Technologies for the Wireless Future: Wireless World Research Forum

TECHNOLOGIES FOR THE WIRELESS FUTURE Wireless World Research Forum (WWRF) Volume 3 Edited by Klaus David ComTec, University of Kassel, Germany

A John Wiley and Sons, Ltd, Publication

TECHNOLOGIES FOR THE WIRELESS FUTURE

TECHNOLOGIES FOR THE WIRELESS FUTURE Wireless World Research Forum (WWRF) Volume 3 Edited by Klaus David ComTec, University of Kassel, Germany

A John Wiley and Sons, Ltd, Publication

Copyright  2008 Wireless World Research Forum (WWRF) This edition first published 2008 by John Wiley & Sons, Ltd. Registered office: John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com. The right of the author to be identified as the author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought.

ISBN

978-0-470-99387-3 (HB)

A catalogue record for this book is available from the British Library Typeset in 10/12 Times by Laserwords Private Limited, Chennai, India Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham, Wiltshire

Contents List of Figures

xiii

List of Tables

xxi

List of Contributors

xxiii

Foreword by Dr Atsushi Murase

xxix

Foreword by Charles Backof

xxxi

Preface Acknowledgements

xxxiii xxxv

1 Introduction 1 Edited by Dr Nigel Jefferies (Vodafone Group R&D, UK) and Prof. Dr Klaus David (ComTec, University of Kassel, Germany) 1.1 A Book of Visions 1 1.2 The Wireless World Research Forum 1 1.3 Current Situation and Trends 3 1.3.1 Next Generation Mobile Networks 5 1.3.2 The International Telecommunications Union (ITU) 5 1.4 Overview of the Following Chapters 6 References 7 2 Vision and Stakeholder Requirements for Future Mobile Systems Edited by Dr Nigel Jefferies (Vodafone Group R&D, UK) 2.1 A Vision of 2017 2.2 Stakeholder Requirements 2.2.1 User Requirements 2.2.2 General Market Requirements 2.2.3 Service Provider Requirements 2.2.4 Backbone Provider Requirements 2.2.5 Access Provider Requirements 2.2.6 Equipment Manufacturer Requirements 2.2.7 Requirements Related to Converging Digital Industries and Application Sectors 2.2.8 Legal and Regulatory Requirements

9 9 10 10 17 21 22 23 24 25 26

vi

Contents

2.3 Acknowledgements References 3 User Requirements, Scenarios and Business Models Edited by Lene Sørensen and Knud Erik Skouby (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark) 3.1 Introduction 3.2 Scenarios 3.2.1 Participatory Design and Creativity 3.2.2 WWRF Reference Scenarios 3.3 User Requirements 3.3.1 Background and Needs 3.3.2 Mobile Probing Kit 3.3.3 The Approach 3.3.4 Short Reflection 3.4 A User-centred Approach to Service Development 3.5 Usability 3.5.1 Low-fi Prototyping 3.5.2 The Actual Test 3.5.3 Short Reflection 3.6 Business Modeling 3.6.1 New Trends in Business Models 3.7 Conclusions and Further Research 3.7.1 User Requirements, Scenarios and Usability 3.7.2 Business Models 3.8 Acknowledgements References

27 28 29

29 30 30 34 43 43 44 44 46 46 47 47 48 50 50 51 53 53 53 54 54

4 Service Infrastructures 57 Edited by Prof. Dr Klaus David (ComTec, University of Kassel, Germany) and Dr Mika Klemettinen (Nokia, Finland) 4.1 Introduction 57 4.2 Semantic Services 58 4.2.1 Objective and Scope 59 4.2.2 From Web Services Towards Semantic Web Services 63 4.2.3 Challenges 72 4.2.4 Business Considerations 76 4.2.5 Summary and Outlook 78 4.3 Service Creation 80 4.3.1 Terms and Definitions 81 4.3.2 Service Domain 82 4.3.3 Technology Domain 96 4.3.4 Business Models for Service Creation 106 4.3.5 Future Research and Development 116 4.3.6 Conclusions 118 4.4 Service Architecture for the Wireless World 118

Contents

vii

4.4.1 Background 119 4.4.2 Key Concepts 123 4.4.3 Enabling Technologies 128 4.4.4 Functional Architecture and Service Delivery Framework 142 4.4.5 Interfaces 144 4.4.6 Groundings and Exploitation 151 4.4.7 Scenarios and Architecture Requirements of Vehicular Communications 156 4.4.8 Conclusions 159 4.5 Acknowledgements 160 References 161 5 The WWI System Architecture for B3G Networks Edited by Andreas Schieder (Ericsson GmbH, Germany), Elias Tragos (National Technical University of Athens, Greece), Andrej Mihailovic (King’s College London, UK), Jukka Salo (Nokia Siemens Networks, Finland) and Jan van der Meer (Ericsson Telecommunicati, The Netherlands) 5.1 Introduction 5.1.1 Key Architectural Principles 5.2 Heterogeneous Radio Resource Management (HRRM) in the WWI System Architecture 5.2.1 Introduction and Motivation 5.2.2 Synthesis of a WWI Architecture for HRRM 5.2.3 HRRM Decision Making 5.2.4 Use Case Examples 5.3 Mobility 5.3.1 Introduction and Motivation 5.3.2 The Functional Architecture 5.3.3 Fuzzy Logic-based Heterogeneous Mobility Management 5.4 Context Provisioning 5.4.1 Introduction and Motivation 5.4.2 The Usage and Ontologies of Context Information 5.4.3 Architecture and Interfaces for Context Awareness 5.4.4 Use Cases 5.5 Network Management in the WWI System Architecture 5.5.1 Analysis of Main Assessment Criteria 5.5.2 Management Planes 5.5.3 Analysis of Management Plane Overlays 5.6 Conclusions 5.7 Acknowledgements References 6 New Air Interface Technologies Edited by Dr Angeliki Alexiou (Bell Labs, Alcatel-Lucent, UK) and Dr Gerhard Bauch (DOCOMO Euro-Labs, Germany) 6.1 Introduction 6.2 Error Control Coding Options for Next-generation Wireless Systems 6.2.1 Introduction

165

165 166 169 169 169 171 172 177 177 177 178 180 180 181 183 184 187 187 191 193 194 194 195 197

197 199 199

viii

Contents

6.2.2 Coding 6.2.3 Decoding 6.2.4 Architecture and HW Requirements 6.2.5 Turbo Principle 6.2.6 Standardization Overview 6.2.7 Conclusions and Challenges 6.3 Multi-dimensional Channel Modeling 6.3.1 Multi-dimensional Modeling Methods 6.3.2 Recent Results 6.3.3 MIMO Radio Channel Models in Wireless Standards 6.3.4 Future Research Topics and Open Issues 6.4 Multiuser MIMO Systems 6.4.1 Introduction 6.4.2 Linear Precoding Techniques 6.4.3 Nonlinear Precoding Techniques 6.4.4 Simulation Results 6.4.5 Conclusion 6.5 Acknowledgements References

201 209 213 218 221 223 224 225 226 230 233 234 234 235 237 238 242 243 243

7 Short-range Wireless Communications 251 Edited by Prof. Rolf Kraemer (IHP, Germany) and Marcos Katz (VTT, Finland) 7.1 Introduction 251 7.2 Integrative and Cooperative Aspects of Short-range Communications: Technologies, Designing Rules and Trends 253 7.2.1 Motivating the Need for Short-range Communications 254 7.2.2 Introduction to Short-range Communications 254 7.2.3 Peer-to-peer Networks and Social Networks 254 7.2.4 Hybrid Mobile Device and Sensor Networks 256 7.2.5 Wireless Grids 257 7.2.6 Cellular Controlled Peer-to-peer Networking 260 7.2.7 Scenarios for Social Networking 264 7.2.8 Short-range Connectivity Measurement Campaign 265 7.2.9 Designing Rules for Future Wireless Short-range Communication Systems 270 7.2.10 Short-range Communication as the Main Driving Force for Cooperative Networking 271 7.2.11 Conclusions 272 7.3 Ultra Wideband Radio over Optical Fibre 272 7.3.1 Introduction 273 7.3.2 Background and Motivation 276 7.3.3 UROOF: User Applications and Basic System Configuration 279 7.3.4 Fundamentals of UROOF Technologies 287 7.3.5 Link Analysis of UROOF Systems 301 7.3.6 Technology Trends to be Explored and Summary 313 7.4 Work in Progress 314

Contents

ix

7.4.1

High Data Rate Wireless Communications in the Unlicensed 60 GHz Band 7.4.2 Ultra Wideband Communication 7.5 Acknowledgements References 8 Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks Edited by Prof. Panagiotis Demestichas, George Dimitrakopoulos and Yiouli Kritikou (University of Piraeus, Greece) 8.1 Introduction 8.2 Overview of Cognitive Wireless Networks 8.2.1 Operation Principles 8.2.2 High-level View of Management Functionality 8.3 Management Mechanisms for Cognitive Wireless Networks 8.3.1 Context Acquisition 8.3.2 Profile Management 8.3.3 Policy-based Management 8.3.4 Configuration of Behaviour of Cognitive Infrastructures 8.3.5 Configuration of Behaviour of Cognitive Terminals 8.4 Supplementary Knowledge Features in Support of Cognition 8.5 Summary 8.6 Acknowledgements References 9 Methods for Spectrum Sharing Edited by Sudhir Dixit (Nokia Siemens Networks) 9.1 Introduction 9.1.1 Drivers for Spectrum Sharing and Spectrum Etiquette 9.1.2 A High-level Functional Model of Spectrum Sharing Management 9.2 Spectrum Sharing Categories Based on Centralized and Distributed Approaches 9.2.1 Centralized Spectrum Sharing between Cooperative Access Networks 9.2.2 Distributed Spectrum Sharing between Cooperative Access Networks 9.2.3 Secondary Networks with Central Controller 9.2.4 Distributed Spectrum Sharing between Noncooperative Access Networks 9.3 Problems and Issues in Flexible Spectrum Use 9.3.1 Higher- (and Cross-) layer Issues 9.4 Conclusion 9.5 Acknowledgements References 10 Ultra Broadband Home Area Network Edited by Djamal-Eddine Meddour (Orange Labs, France Telecom Group) 10.1 Introduction 10.2 Applications Challenges

314 323 342 342

349

349 351 351 351 352 353 354 355 356 360 362 364 364 364 367 367 367 368 369 370 373 376 378 383 383 386 386 386 389 389 390

x

Contents

10.2.1 UBB-HAN as a Convergence Platform 10.2.2 Key Requirements 10.3 Connectivity 10.3.1 Wireless 10.3.2 Wired 10.3.3 Hybrid 10.4 Access Challenges 10.4.1 The Concept of Access Network Continuity 10.4.2 Reference Architecture Model for the Interaction Between the UBB-HAN and the Access Network 10.4.3 The Evolutions Expected in the Access Network 10.4.4 Migration Scenarios 10.5 Architecture 10.5.1 QoS 10.5.2 Topology 10.5.3 Integration with Mobile Network 10.6 Conclusion 10.7 Acknowledgements References 11 Combined View of Future Systems Edited by Mikko A. Uusitalo (Nokia Research Center, Finland) 11.1 Introduction 11.2 Applications and Services 11.2.1 User Interface and Service Adaptation 11.2.2 Personalization 11.2.3 Group Support 11.2.4 Context Awareness 11.2.5 Privacy and Trust 11.2.6 Service Usage and Provisioning 11.2.7 Operational Management 11.2.8 Charging and Billing 11.2.9 Service Layer Mobility Support 11.2.10Peer-to-peer Services Support 11.2.11Negotiation Support 11.3 IP-based Communication Subsystem 11.3.1 Service Support Layer 11.3.2 Provision of Flexible Environments for Applications and Services 11.3.3 Network Control and Management Layer 11.3.4 Business Models and Functions Classification in Space Time Domains 11.3.5 Quality of Service (QoS) 11.3.6 Duplexing 11.3.7 Inter-cell Coordination 11.3.8 Integrated OA&M (Operation, Administration and Management) 11.3.9 Transport and Higher Layers in Internet-based Networks 11.4 Access Network 11.4.1 Flat Full-IP Access Architecture

390 391 392 392 393 394 395 395 395 397 397 398 398 399 402 405 405 405 407 407 409 409 410 410 410 411 411 412 412 414 414 415 415 415 416 416 416 417 418 418 419 419 422 422

Contents

11.4.2 Wired Technologies 11.4.3 Wireless Ubiquitous Coverage 11.4.4 Sensor Networks 11.5 Development of Reconfigurability and Cognitive Wireless Networks 11.5.1 Introduction 11.5.2 Advent of Cognitive Networks 11.5.3 Management Architecture and Operation 11.5.4 Management of Network Segments/Elements 11.5.5 Terminal Management 11.5.6 Reconfigurable Equipment (HW Aspects) 11.6 Other End-to-end Aspects 11.6.1 Self-organization 11.7 Summary and Conclusion 11.8 Acknowledgements References

xi

423 423 432 435 435 436 438 441 445 445 447 447 449 452 452

Appendix: Glossary

455

Index

465

List of Figures Figure 1.1 The structure of WWRF

3

Figure 2.1 Development of the WWRF system concept

10

Figure 2.2 The needs, value and capability planes for the wireless world

11

Figure 3.1 Example of a scenario landscape, including image elements as well as links between them. Strings were used to show links between problems and solutions

33

Figure 3.2 Overview of the scenario approach in developing the WWRF reference scenarios. Notice the feedback loop between the user scenarios and the visions of WGs and SIGs

36

Figure 3.3 The probing kit used in the MAGNET Beyond project for identifying user requirements for nomadic users. The probe was called IDEA MAGNET. (Reproduced with permission from MAGNET)

45

Figure 3.4 Slides from the Visionpool, used to motivate the users to be creative while wearing the mobile probing kit

45

Figure 3.5 Low-fi paper prototype developed in the MAGNET Beyond project [20]. (Reproduced with permission from MAGNET)

48

Figure 3.6 The back of the prototype with the string holding everything together. (Reproduced with permission from MAGNET)

48

Figure 3.7 The business model seen as a reflection of the technical system in the economic system

51

Figure 3.8 Flowchart showing the process from collection of user inputs to an operational system

52

Figure 4.1 Target area of the WG2 architecture work shown in the reference model

61

Figure 4.2 Multi-domain network and service environment

63

Figure 4.3 An example ontology for devices, networks and locations

66

Figure 4.4 Elements involve in Semantic Web service delivery

68

Figure 4.5 Formal mapping of OWL-S to web services (WS) [12]

69

Figure 4.6 GCM comprises three levels of communication

83

Figure 4.7 Example of an activity system

84

Figure 4.8 Archetypical steps of service development

89

Figure 4.9 The JOnAS execution environment [55]

98

Figure 4.10 MUPE application platform

99

xiv

List of Figures

Figure 4.11 SCW State Service Building Blocks Wizard

102

Figure 4.12 SCW Rule Editor

103

Figure 4.13 SCW testing and deployment

104

Figure 4.14 OMA enablers utilizing IMS capabilities (http://www.openmobilealliance.org)

104

Figure 4.15 Relationship between OSE (with OMA enablers), IMS, non-IMS, IMS services and non-OMA enablers (http://www.openmobilealliance.org)

105

Figure 4.16 Customer-centric business model

110

Figure 4.17 Device manufacturer-centric business model

111

Figure 4.18 Service provider-centric business model

111

Figure 4.19 Network operator-centric business model

112

Figure 4.20 Organizational design (adapted from [54])

115

Figure 4.21 Convergence on application and session/control layer based on IMS

122

Figure 4.22 An illustration of role-specific viewpoints of information

128

Figure 4.23 DCS (distributed communication sphere) and its abstraction (communication model) as viewed in SPICE. (Reproduced with permission from SPICE project)

130

Figure 4.24 Service composition with monitoring and controlling

131

Figure 4.25 Service platform open and controlled environment

133

Figure 4.26 Privacy management

134

Figure 4.27 GBA architectural overview

135

Figure 4.28 Combination of GBA and Liberty Alliance

136

Figure 4.29 Modularized logical architecture

137

Figure 4.30 Modularization of the device service architecture

138

Figure 4.31 Platformization of the device service architecture

139

Figure 4.32 Virtualization of the device service architecture

139

Figure 4.33 Full virtualization

140

Figure 4.34 Hypervisor

140

Figure 4.35 VMWare Infrastructure 3

141

Figure 4.36 Overview of the layered architecture

143

Figure 4.37 HAL (hardware abstraction layer) elements

146

Figure 4.38 Ambient network interfaces: ARI, ANI, ASI

148

Figure 4.39 ASI operation

149

Figure 4.40 ASI switchboard

151

Figure 4.41 Reference model for the service architecture of car-to-car and car-to-x

158

Figure 5.1 Layered structure of the WWI reference network architecture

166

List of Figures

xv

Figure 5.2 First synthesis of a WWI system architecture for the provisioning of multi-radio resource management

170

Figure 5.3 Managing different sets in the HRRM

171

Figure 5.4 Range of HRRM use case deployments

173

Figure 5.5 Access selection procedure selecting and invocating the WINNER RAN 174 Figure 5.6 Intersystem handover from a WINNER RAN towards a second (reconfigurable) RAN

175

Figure 5.7 Example of functional architecture for the inter-RAT handover process 179 Figure 5.8 Fuzzy multicriteria vertical handover scheme

179

Figure 5.9 System-level architecture for context awareness

183

Figure 5.10 Use case: Register Context Provider

185

Figure 5.11 Use case: Deregister Context Provider

185

Figure 5.12 Use case: Get a Context Provider for an Entity and Parameter

186

2

Figure 5.13 E R II high-level system architecture for cognitive reconfigurable wireless networks

191

Figure 5.14 AN management node model

192

Figure 6.1 Coding schemes with iterative decoding

201

Figure 6.2 Iterative decoding

202

Figure 6.3 BER of PCCC and SCCC

203

Figure 6.4 Design of codes defined on graphs as an optimization problem

204

Figure 6.5 Performance comparison binary vs. NB-LDPC codes. N D 3008 coded bits and rate R D 1=2

208

Figure 6.6 Performance comparison binary vs. NB-LDPC codes.N D 564 coded bits and rate R D 2=3

209

Figure 6.7 Soft-in/soft-out decoder

212

Figure 6.8 LDPC decoder architecture for layered decoding

214

Figure 6.9 Basic parallel concatenated turbo decoder

216

Figure 6.10 Windowing scheme for one-producer SISO

217

Figure 6.11 SISO architecture with three recursion units

217

Figure 6.12 The equivalent multistream interleaving of the proposed algorithm

220

Figure 6.13 Transmitter turbo structure of the proposed algorithm

220

Figure 6.14 Generic process for developing a standardized radio channel model

224

Figure 6.15 Communication system with M transmit and N receive antennas and an M ð N MIMO radio channel H. Spatial correlation matrices RTX and RRX define the inter-antenna correlation properties

225

Figure 6.16 One measurement setup in a rural environment

227

Figure 6.17 One measurement setup in an outdoor-to-indoor environment

228

xvi

List of Figures

Figure 6.18 One measurement setup in a suburban environment

228

Figure 6.19 Block diagram of multiuser MIMO downlink system

236

Figure 6.20 BER performance comparison of BD, SO THP and SMMSE in configuration

238

Figure 6.21 10 % outage capacity performance of BD and SMMSE, with and without scheduling

239

Figure 6.22 BER performance comparison of BD and SMMSE in configuration {2; 2} ð 4 with and without scheduling

240

Figure 6.23 10 % outage capacity performance of BD, SMMSE and SMMSE THP

240

Figure 6.24 BER performance comparison of SMMSE and SMMSE THP

241

Figure 6.25 BER performance comparison of BD, SMMSE and RBD

241

Figure 7.1 A classification of short-range communications according to the typical supported range

255

Figure 7.2 A general classification of short-range communications

255

Figure 7.3 Screenshot on a NOKIA 6600 of the SMARTEX application developed at Aalborg University

256

Figure 7.4 Details of the wireless sensor board (collaboration of TU Berlin and Aalborg University) (left) and some application scenarios of the parking sensor application (top right, bottom right)

257

Figure 7.5 A mobile device broken up into several capabilities, grouped into user interface, communication interface and built-in resources

258

Figure 7.6 Example of cooperative devices of the same user for a mobile phone and a tablet PC

259

Figure 7.7 Cellular-controlled peer-to-peer network

260

Figure 7.8 Comparison of monolithic and cooperative communication systems with respect to requested services and the related costs

261

Figure 7.9 Possible cooperative application on the mobile device

263

Figure 7.10 Increased market potential for cooperative services

263

Figure 7.11 Wireless grids in different everyday scenarios

264

Figure 7.12 A pervasive wireless social network exploiting the discussed composite network architecture

265

Figure 7.13 Screenshots of the Python S60 application

266

Figure 7.14 Number of wireless devices (Bluetooth-enabled) discovered at Aalborg Airport

268

Figure 7.15 Basic wireless device statistics for Aalborg Airport

268

Figure 7.16 Basic wireless device classification at Aalborg Airport

269

Figure 7.17 Scanning time (mean search to find wireless devices) vs number of devices found

269

List of Figures

xvii

Figure 7.18 MB-OFDM band plan

276

Figure 7.19 A bidirectional RoF BS containing a remote antenna unit (RAU) based on EAT

278

Figure 7.20 UROOF UWB range extension application

280

Figure 7.21 UROOF Case 1.A: in-building UWB extension application

281

Figure 7.22 UROOF Case 1.B: UWB broadcasting

282

Figure 7.23 UROOF Case 2.A: UWB access segmentation

283

Figure 7.24 UROOF Case 2.B: smart antenna MIMO processing

284

Figure 7.25 UROOF Case 3.A: very low-latency applications

284

Figure 7.26 UROOF Case 3.B: localization-based (triangulationCbeamforming) UWB communication

286

Figure 7.27 IUT structure including amplifier in the front end

301

Figure 7.28 Simple point-to-point link

302

Figure 7.29 Intermodulation products and harmonics generated by a three-tone modulation of a nonlinear device

305

Figure 7.30 Total number of third-order intermodulation products as a function of channel number for an N D 122 channel system

305

Figure 7.31 System-level generic electrical-optical converter (VCSEL/E-EAT) model

310

Figure 7.32 System-level generic photoreceiver (APD/PIN/OCMC) model

310

Figure 7.33 Power delay profile of S-V model; here gain coefficients of multipath channel

i ÐÐÐ Þk;`

312

Figure 7.34 Bandwidth in the 60 GHz band in different countries; RTTT D road transport and traffic telematics

316

Figure 7.35 Proposed channel plan for use of the 60 GHz band with wideband channels A1–A4 and narrowband channels B1–B12

317

Figure 7.36 Graphical representation of Usage Model 1 (UM1), ‘Uncompressed HDTV Video Streaming’ (source: IEEE802.15.3c Usage Model Document)

318

Figure 7.37 Graphical representation of Usage Model 5 (UM5) ‘Kiosk File Downloading’ (source: IEEE802.15.3c Usage Model Document)

319

Figure 7.38 Probability of false lock for a UWB communication system

331

Figure 7.39 Probability of false lock for a UWB communication system in the presence of narrowband interference (using a maximum-likelihood estimator). Dots signify an absence of NBI, stars that no interference mitigation is employed, and squares and diamonds that an interference mitigation method (covariance matrix estimation and spectral encoding, respectively) is employed. SIR D -15dB

332

xviii

List of Figures

Figure 7.40 ADC technology trend (source: the series of Proc. of IEEE SolidStated Circuits Conference (ISSCC) and the International Technology Roadmap for Semiconductor (ITRS) [109])

333

Figure 7.41 Edholm’s law of data rate for short-range communications

341

Figure 8.1 Overview of the wireless world in the B3G era

350

Figure 8.2 Cognitive networks operation loop

351

Figure 8.3 Problem description for the operation of cognitive networks

352

Figure 8.4 User–service communication, exploiting a system’s cognition for efficient and effective service delivery

355

Figure 8.5 Representation of policy-based management

356

Figure 8.6 Software download procedure

359

Figure 8.7 The energy gap is growing

360

Figure 8.8 Energy-scalable SDRs achieve low-energy operation through crosslayer QoS and energy management

361

Figure 8.9 Left: power consumption vs. link SINAD tradeoff enabled by scalable front end; Right: average net MAC data rate vs. energy efficiency tradeoff enabled by cross-layer energy management

362

Figure 8.10 Approach for obtaining knowledge useful for decisions at the network segment level

363

Figure 9.1 Spectrum occupancy measurement results

368

Figure 9.2 A functional illustration of the steps involved in spectrum identification and allocation

368

Figure 9.3 Dynamic spectrum management between different radio access technologies

371

Figure 9.4 Dynamic spectrum management between operators inside a single RAT 373 Figure 9.5 Architecture of the secondary wireless access system of IEEE 802.22

377

Figure 9.6 All nodes listen to the other nodes and maintain a database of all channels used, when used, by whom, and how much each is loaded. This information is used by the nodes to select a channel when there is a need to send, or to switch to a new channel when the current one is heavily loaded

380

Figure 9.7 One example of master-slave configuration

381

Figure 9.8 The FCC interference temperature concept (source: Spectrum Policy Task Force)

382

Figure 9.9 Illustration of spectrum sharing RRM architecture

384

Figure 10.1 HAN as a convergence platform

391

Figure 10.2 Reference architecture of a multi-services access network

396

Figure 10.3 Real topology of a UBB-HAN

400

Figure 10.4 Centralized approach

402

List of Figures

xix

Figure 10.5 Distributed approach

403

Figure 11.1 Illustration of the ambient control space and its three external reference points

426

Figure 11.2 The ambient network connectivity abstractions

427

Figure 11.3 Example of cognitive wireless network. Elements may change RAT, frequency, or both, when new conditions are identified

438

Figure 11.4 (a) Overall management architecture; (b) management functionality for individual element/reconfigurable terminal

439

Figure 11.5 Overview of functional blocks

442

Figure 11.6 Dynamic network planning and management – problem description

443

Figure 11.7 Reconfigurable elements approaches: (a) fully reconfigurable equipment; (b) multimode equipment – software-defined signal processing; (c) multimode equipment

446

Figure 11.8 Peer-to-peer networking

448

Figure 11.9 Topology control as a self-organizing paradigm

449

Figure 11.10 Main features of the various approaches to topology control

449

Figure 11.11 Reference I-cake

451

List of Tables Table 4.1 Semantic service provision compared to conventional service provisioning

78

Table 4.2 Comparison of theatre, airline, library and inventory domains [38]

86

Table 4.3 Comparing domain-specific and semantics-driven software architectures 88 Table 4.4 Business issues in the service domain

115

Table 4.5 Business issues in the technology domain

116

Table 4.6 Business issues in the organizational domain

116

Table 4.7 Business issues in the financial domain

116

Table 4.8 Shift in device architecture priorities

120

Table 5.1 Status of management assessment criteria in WWI projects

188

Table 6.1 Synthesis results for the WiMAX 802.16e LDPC decoder [57] turbo codes

215

Table 6.2 Synthesis results for the WiMAX 802.16e duo-binary turbo decoder [61]

218

Table 6.3 Channel model features

229

Table 6.4 Channel model parameters

229

Table 6.5 Comparison of the basic features of two standardized MIMO channel models

233

Table 7.1 Key requirements for future wireless short-range networks

270

Table 7.2 Number of packet exchanges for the WLAN and Bluethooth cases (master and slave separated)

271

Table 7.3 UROOF Case 1.A functionalities

280

Table 7.4 UROOF Case 1.B functionalities

281

Table 7.5 UROOF Case 2.A functionalities

282

Table 7.6 UROOF Case 2.B functionalities

285

Table 7.7 UROOF Case 3.A functionalities

285

Table 7.8 UROOF Case 3.B functionalities

286

Table 7.9 VCSEL parameters

290

Table 7.10 Typical values for power budget calculation

303

Table 7.11 Comparison of 100 % conversion efficiency for VCSEL at UROOF wavelengths

311

xxii

List of Tables

Table 7.12 Comparison of 100 % conversion efficiency for receiver at UROOF wavelengths

311

Table 7.13 Comparition of single-carrier and OFDM for 60 GHz

319

Table 7.14 Estimated power dissipation of a 60 GHz OFDM transceiver in 2008

322

Table 7.15 Power dissipation figures for 60 GHz OFDM transceivers

323

Table 9.1 Spectrum sharing methods

369

List of Contributors Chapter 1 editors Nigel Jefferies (Vodafone Group R&D, UK) Klaus David (ComTec, University of Kassel, Germany)

Lene Sorensen (Aalborg University, Denmark) Mikko Uusitalo (Nokia Research, Finland) Hu Wang (Huawei, China)

Chapter 2 editor Nigel Jefferies (Vodafone Group R&D, UK) Authors/contributors Andrew Aftelak (Motorola, France) Angeliki Alexiou (Alcatel-Lucent, UK) Stefan Arbanowski (Fraunhofer Fokus, Germany) Brigitte Cardinael (Orange Labs, France) Jean-Claude Sapanel (Orange Labs, France) Klaus David (ComTec, University of Kassel, Germany) Panagiotis Demestichas (University of Piraeus, Greece) George Dimitrakopoulos (University of Piraeus, Greece) Sudhir Dixit (Nokia Siemens Networks, USA) Bernard Hunt (Philips, UK) Nigel Jefferies (Vodafone, UK) Wolfgang Kellerer (DOCOMO, EuroLab, Germany) Vinod Kumar (Alcatel-Lucent, France) Werner Mohr (Nokia Siemens Networks, Germany) Angela Sasse (University College London, UK) Knud Erik Skouby (Aalborg University, Denmark)

Chapter 3 editors Knud Erik Skouby (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark) Lene Sørensen (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark) Authors/contributors Anders Henten (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark) Rune Rosewall (Telia Sonera, Sweden) Chapter 4 editors Klaus David (ComTec, University of Kassel, Germany) Mika Klemettinen (Nokia, Finland) Authors/contributors Kazimierz Bałos (University of Science and Technology Krakow, Poland) Alberto Baravaglio (Telecom Italia, Italy) Emmanuel Bertin (France Telecom R&D, France) Joerg Brakensiek (Nokia, Germany) Mohammad M.R. Chowdhury (University Graduate Center, UniK, Kjeller, Norway)

xxiv

Christophe Cordier (France Telecom, France) Noel Crespi (GET-INT, Institut National des Telecommunications, France) Jorge Cuellar (Siemens, Germany) Klaus David (ComTec, University of Kassel, Germany) Wolf-Dietrich Moeller (Siemens, Germany) Olaf Droegehorn (ComTec, University of Kassel, Germany) Babak Farshchian (Telenor, Norway) Matthias Franz (Siemens, Germany) Alex Galis (University College London, UK) Anne Marte Hjemås (Telenor, Norway) Stefan Holtel (Vodafone Group R&D, Germany) Guenther Horn (Siemens, Germany) Kashif Iqbal (DERI, National University of Ireland, Galway, Ireland) Niklas Klein (ComTec, University of Kassel, Germany) SianLun Lau (ComTec, University of Kassel, Germany) Erik Lillevold (University Graduate Center, UniK, Kjeller, Norway) Bernd Mrohs (Fraunhofer FOKUS, Germany) Josef Noll (University Graduate Center, UniK, Kjeller, Norway) Henning Olesen (Technical University of Denmark, Denmark) Massimo Paolucci (DOCOMO, EuroLab, Germany) Andreas Pirali (ComTec, University of Kassel, Germany) Jari Porras (University of Lappeenranta, Finland) Erwin Postmann (Siemens, Germany) Vilho Raisanen (Nokia, Finland) Hariharan Rajasekaran (Siemens, Germany)

List of Contributors

Dumitru Roman (DERI Innsbruck, Austria) Rune Roswall (TeliaSonera, Sweden) Stefano Salsano (University of Rome, Italy) Robert Seidl (Siemens, Germany) Amy Tan (University of Surrey, UK) Sasu Tarkome (Nokia, Finland) Henrik Thuvesson (TeliaSonera, Sweden) Hans Joerg Voegel (BMW, Germany) Matthias Wagner (DOCOMO, EuroLab, Germany) Karim Yaici (University of Surrey, UK) Anna V. Zhdanova (Forschungszentrum Telekommunikation Wien, Austria) Josip Zoric (Telenor, Norway) Chapter 5 editors Andreas Schieder (Ericsson GmbH, Germany) Elias Tragos (National Technical University of Athens, Greece) Andrej Mihailovic (King’s College London, UK) Jukka Salo (Nokia Siemens Networks, Finland) Jan van der Meer (Ericsson Telecommunicati, The Netherlands) Authors/contributors Imran Ashraf (Alcatel Lucent, UK) Giovanni Bartolomeo (University of Rome “Tor Vergata”, Italy) Zachos Boufidis (University of Athens, Greece) Khadija Daoud (Orange Labs, France) Alex Galis (University College London, UK) Raffaele Giaffreda (BT, UK) David Grandblaise (Motorola, France) Oliver Holland (King’s College London, UK) Alexandros Kaloxylos (University of Athens, Greece)

List of Contributors

Herma van Kranenburg (Telematica Instituut, Netherlands) Jijun Luo (Siemens, Germany) Juha Mikola (Nokia, Finland) Markus Muck (Motorola, France) Eleni Patouni (University of Athens, Greece) Mikael Prytz (Ericsson, Sweden) Abed Samhat (Orange Labs, France) Ove Strandberg (Nokia Siemens Networks, Finland) Anthony Tarlano (DOCOMO Eurolabs, Germany) Dorota Witaszek (Fraunhofer Institute FOKUS, Germany) Chapter 6 editors Angeliki Alexiou (Bell Labs, Alcatel-Lucent, UK) Gerhard Bauch (DOCOMO Euro-Labs, Germany) Authors/contributors Angeliki Alexiou (Bell Labs, Alcatel-Lucent, UK) Simon Gale (Wireless Technology Laboratories, Nortel, UK) Martin Haardt (Ilmenau University of Technology, Germany) Tommi J¨ams¨a (Elektrobit, Finland) Andy Jeffries (Wireless Technology Laboratories, Nortel, UK) Pekka Ky¨osti (Elektrobit, Finland) Thierry Lestable (Samsung Electronics Research Institute, UK) Juha Meinil¨a (Elektrobit, Finland) Milan Narandˇzi´c (Ilmenau University of Technology, Germany) Jukka-Pekka Nuutinen (Elektrobit, Finland) Moshe Ran (Holon Institute of Technology, Israel)

xxv

Florian R¨omer (Ilmenau University of Technology, Germany) Christian Schneider (Ilmenau University of Technology, Germany) Veljko Stankovic (Ilmenau University of Technology, Germany) Reiner Thom¨a (Ilmenau University of Technology, Germany) Juha Ylitalo (Elektrobit, Finland) Chapter 7 editors Rolf Kraemer (IHP, Germany) Marcos Katz (VTT, Finland) Authors/contributors Antti Anttonen Maria-Gabrielle di Benedetto Robert W. Brodersen Isabelle Bucaille (Thales group, France) Beatrice Cabon (INPG, France) Emil Dimitrov (Institute of Communications Technology, Leibniz University of Hannover, Germany) Marcus Ehrig (IHP GmbH, Frankfurt (Oder), Germany) Yossef Ben Ezra (Holon Institute of Technology (HIT), Israel) Frank H.P. Fitzek (Aalborg University, Denmark) Gunter Fischer (IHP GmbH, Frankfurt (Oder), Germany) Srdjan Glisic (IHP GmbH, Frankfurt (Oder), Germany) Eckhard Grass (IHP GmbH, Frankfurt (Oder), Germany) Norbert Gungle (TES, Germany) Motti Haridim (Holon Institute of Technology (HIT), Israel) Frank Herzel (IHP GmbH, Frankfurt (Oder), Germany)

xxvi

Thomas Kaiser (Institute of Communications Technology, Leibniz University of Hannover, Germany) Rolf Kraemer (IHP GmbH, Frankfurt (Oder), Germany) Milos Krstic (IHP GmbH, Frankfurt (Oder), Germany) Ronen Korman (Wisair, Israel) Boris I. Lembrikov(Holon Institute of Technology (HIT), Israel) Geert Leus Roberto Llorente (Universidad Politecnica de Valencia, Spain) Aarne Maemmelae Laurence Milstein Andreas Molisch Lorenzo Mucchi Ian O’Donnell Maxim Piz (IHP GmbH, Frankfurt (Oder), Germany) Vincent Poor Moshe Ran (Holon Institute of Technology (HIT), Israel) Juergen Sachs (Institute of Information Technology, Ilmenau University of Technology, Germany) Henrique Salgado (INESC Porto, Portugal) Christoph Scheytt (IHP GmbH, Frankfurt (Oder), Germany) Klaus Schmalz (IHP GmbH, Frankfurt (Oder), Germany) Claudio da Silva Werner Soergel Yaoming Sun (IHP GmbH, Frankfurt (Oder), Germany) Waho Takao Manoj Thakur (University of Essex, UK) Klaus Tittelbach-Helmrich (IHP GmbH, Frankfurt (Oder), Germany)

List of Contributors

Reiner Thomae (Institute of Information Technology, Ilmenau University of Technology, Germany) Alle-Jan van der Veen Wolfgang Winkler (IHP GmbH, Frankfurt (Oder), Germany) Mike Wolf (Institute of Information Technology, Ilmenau University of Technology, Germany) Sven Zeisberg (Institute of Information Technology, Ilmenau University of Technology, Germany) Rudolf Zetik (Institute of Information Technology, Ilmenau University of Technology, Germany) Chapter 8 editors Panagiotis Demestichas (University of Piraeus, Greece) George Dimitrakopoulos (University of Piraeus, Greece) Yiouli Kritikou (University of Piraeus, Greece) Authors/contributors Eugenia Adamopoulou (National Technical University of Athens, Greece) Didier Bourse (Motorola France) Enrico Buracchini (Telecom Italia Labs, Italy) Jan Craninckx (IMEC, Belgium) Antoine Dejonghe (IMEC, Belgium) Kostantinos Demestichas (National Technical University of Athens, Greece) Panagiotis Demestichas (University of Piraeus, Greece) George Dimitrakopoulos (University of Piraeus, Greece) Paolo Goria (Telecom Italia Labs, Italy)

List of Contributors

Yiouli Kritikou (University of Piraeus, Greece) Klaus Moessner (University of Surrey, UK) Markus Muck (Motorola France) Liesbet Van der Perre (IMEC, Belgium) Alessandro Trogolo (Telecom Italia Labs, Italy) Chapter 9 editors Sudhir Dixit (Nokia Siemens Networks, USA) Authors/contributors Sudhir Dixit (Nokia Siemens Networks, USA) Juha O. Juntunen (Nokia) Kimmo Kalliola (Nokia) Jean-Philippe Kermoal (Nokia) Juha Pihlaja (Nokia) Yi Wang (Huawei, China) Chapter 10 editors Djamal-Eddine Meddour (Orange Labs, France Telecom Group, France) Authors/contributors Eftychia Alexandri (Orange Labs, France Telecom Group, France) Martial Bellec (Orange Labs, France Telecom Group, France) Pierre Jaffr´e (Orange Labs, France Telecom Group, France) Abdesselem Kortebi (Orange Labs, France Telecom Group, France) Riadh Kortebi (Orange Labs, France Telecom Group, France) Philippe Niger (Orange Labs, France Telecom Group, France) Chapter 11 editor Mikko A. Uusitalo (Nokia Research Center, Finland)

xxvii

Authors/contributors Andrew Aftelak (Motorola) Angeliki Alexiou (Alcatel-Lucent, UK) Stefan Arbanowski (Fraunhofer Fokus, Germany) Brigitte Cardina¨el (FT, France) Klaus David (ComTec, University of Kassel, Germany) Panagiotis Demestichas (University of Piraeus, Greece) George Dimitrakopoulos (University of Piraeus, Greece) Sudhir Dixit (Nokia Siemens Networks, USA) Pertti H¨oltt¨a (Elisa, Finland) Bernard Hunt (Philips, UK) Nigel Jefferies (Vodafone Group R&D, UK) Olavi Karasti (Elisa, Finland) Wolfgang Kellerer (DOCOMO Euro-Labs, Germany) Mika Klemettinen (Nokia, Finland) Vinod Kumar (Alcatel, France) Werner Mohr (Nokia Siemens Networks, Germany) Jean-Claude Sapanel (France Telecom, France) Jukka T Salo (NSN, Finland) Angela Sasse (University College London, UK) Knud Erik Skouby (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark) Lene Sorensen (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark) Mikko A. Uusitalo (Nokia Research Center, Finland) Hu Wang (Huawei, China)

Foreword by Dr Atsushi Murase Dr Atsushi MURASE NTT, DOCOMO Managing Director of Research Laboratories

Welcome to this newest volume of technology advances that will pave the road for the wireless future ahead. As a reader, you will find in this book the latest research results from the Wireless World Research Forum. These results, as they are structured and explained in the following chapters, describe the great variety of technologies, components and building blocks needed for the emergence of a new generation of mobile communication systems. In the different working groups within the Wireless World Research Forum, many aspects of such systems have been continuously elaborated on and deeply discussed in technical detail, so that the outcomes, laid down in this book, represent a rich set of innovations beyond the state of the art. These innovations relate to user requirements, service scenarios, service support platforms, cooperative networking scenarios and new air interface technologies to name just a few. However, by setting the stage so that the user’s needs are always kept in the centre, the book leads on towards new insights of market trends, including ways to achieve greater user satisfaction. This will allow us a better understanding of the emergence of new business environments. This is of great importance since competing solutions are continually pushing onto the market from different directions and at unprecedented speed. Naturally the progress in demand and development of new mobile services has resulted in an increased demand for radio-frequency spectrum. The expected allocation of frequencies at the upcoming World Radio Communication Conference 2007 will change the regulatory framework within which the international use of radio-frequency spectrum is managed to the mutual benefit of all partners involved. This will also determine to what extent the technology advances described here will lead to success in the marketplace for next-generation mobile systems. Technology advances by themselves will not always have the strongest influence on market success. Equally important is the adoption of new ideas through the appropriate global standardization bodies. At DOCOMO we watch and contribute to that development, as many other companies around the world do, to make certain that portability and interoperability of new and innovative solutions can be achieved on a global scale. I expect this book to foster a common understanding on global trends and directions for new mobile communication systems.

xxx

Foreword by Dr Atsushi Murase

It is a great achievement of this book to collect and summarize so many of the ongoing activities and to describe them, despite their complexity, in an easy-to-read manner. I wish you an enjoyable reading experience.

Foreword by Charles Backof Charles BACKOF President, IEEE Vehicular Technology Society

On behalf of the IEEE Vehicular Technology Society (VTS), I congratulate the Wireless World Research Forum (WWRF) for publishing Technologies for the Wireless Future Volume 3, and thank them for assembling such a comprehensive view of emerging mobile communications. This book contains the latest thinking on the requirements and technologies to meet the challenges of worldwide wireless systems beyond 2010. The contributions of hundreds of researchers from both industry and academia have been distilled and organized into a well-structured framework. The book takes a top-down approach, with user requirements and market size as the justification for research issues, spectrum requirements, and new technologies to meet system needs. The book will be a useful resource for those involved in 3G Evolution, 4G, and short-range communication systems. Those involved in technology development, system implementation, spectrum regulation, and economics will find relevant and timely information. Because of its relevance to technologists, this book will be available on the IEEE bookshop website, in addition to other sources. The initial chapters cover the vision, requirements and user scenarios in the wireless future. The following chapters deal with service infrastructure, cooperative and ad hoc networking, new air interface technologies, short-range wireless communications, and reconfigurable cognitive networks. The concluding chapters deal with the issues of flexible spectrum, home area networks, and the system concept to unify all these themes. IEEE VTS is the original mobile wireless communications professional organization, with over 68 years of critical contributions to this important field. VTS has worked closely with WWRF in recent years to accelerate the development and implementation of future wireless systems. As just an example, in September 2006 the VTS Conference and WWRF meeting were co-located in Montreal, to encourage closer cooperation. Going forward, the WWRF and IEEE VTS will continue mutual collaborations in order to make future communications visions a reality.

Preface Welcome to the third volume of the book entitled Technologies for the Wireless Future, which is produced by WWRF. The idea is to take the most important outputs from the working groups and special-interest groups that compose the Forum and bring them together in a series of one-volume surveys. The latest of these will give the reader a good overview of the WWRF approach to analyzing the future of wireless and mobile communications, as well as an insight into the trends themselves and the key technologies that will be deployed. The WWRF’s objective to formulate strategic visions and future research directions in the wireless field and to generate, identify and promote technical trends for mobile and wireless system technologies is reflected in this unique and timely book. This fully updated book is a comprehensive single point of reference and features a wealth of new material. The introduction provides an overview of the Wireless World Research Forum, the current situation and trends, as well as an overview of the following chapters of the book, which include: ž the stakeholders and their requirements ž future application scenarios ž a semantic service description, how to create services and the key requirements, as well as concepts of a future service architecture ž the WWI (Wireless World Initiative) reference architecture ž physical layer aspects for future air interfaces ž short-range communication ž reconfigurable platforms ž flexible spectrum use ž future ultra-broadband home area networks ž the WWRF system concept. In April 2008 the total connections to GSM mobile communications networks have passed the 3 billion mark globally. This happened just 17 years after the first GSM network launch in 1991. Nevertheless, we are just at the beginning of the next phase of exciting applications, new systems and further growths. Many of the facilitators for this new phase are further elaborated in this book. The contents will have a wide-ranging appeal to engineers, researchers, managers and students with an interest in the future of wireless. We are sure that you will find this helpful for your study or work – please enjoy your reading. For any questions or suggestions, please contact: [email protected].

Acknowledgements It is a great pleasure for me to acknowledge and give thanks for the contributions of the large number of dedicated and inspiring experts from all over the world. Based on your contributions this book became a reality. I would like to thank the steering board and the vision committee of WWRF for creating the stimulating environment for shaping the vision for the technologies for the wireless future, and the team from Wiley (Mark Hammond, Sarah Hinton and Katharine Unwin) for the good cooperation for this book project. Special thanks for very constructive and productive cooperation in editing the various chapters go out to: Dr Nigel Jefferies – Vodafone, UK – Chair of WWRF Associate Professor Dr Lene Sørensen and Prof. Dr Knud Erik Skouby – Chair of Working Group 1 – both Center for Information and Communication Technologies/IMM, Danish Technical University, Denmark Dr Mika Klemettinen – Nokia, Finland – Vice Chair of Working Group 2 Dr Angeliki Alexiou – Alcatel – Lucent, UK – Chair of Working Group 4 Dr Andreas Schieder – Ericsson, Germany – Vice Chair of Working Group 3 Prof. Rolf Kraemer – IHP, Germany – Chair of Working Group 5 and Dr Marcos Katz – VTT, Finland – Vice Chair of Working Group 5 Prof. Panagiotis Demestichas – Chair of Working Group 6 – George Dimitrakopoulos and Yiouli Kritikou – all Univ. of Piraeus, Greece Dr Sudhir Dixit – Nokia Siemens Networks, USA – Chair of Special Interest Group 3 Dr Djamal-Eddine Meddour – France Telecom, France – Vice Chair of Special Interest Group 4 Dr Mikko Uusitalo – Nokia, Finland

Prof. Dr Klaus David Chair for Communication Technology (ComTec) University of Kassel, Germany

1 Introduction Edited by Dr Nigel Jefferies (Vodafone Group R&D, UK) and Prof. Dr Klaus David (ComTec, University of Kassel, Germany)

1.1 A Book of Visions This is the third in a series of books, entitled Technologies for the Wireless Future and produced by the Wireless World Research Forum (WWRF). The idea is to take the most important outputs from the Working Groups (WGs) and Special-Interest Groups (SIGs) that comprise the Forum, and bring them together in a series of one-volume surveys. The latest of these will give the reader a good overview of the WWRF approach to analysing the future of wireless and mobile communications, as well as an insight into the trends themselves and the key technologies that will be deployed. Previous volumes were published by Wiley in 2005 [1] and 2006 [2], and this volume should be seen as being mostly complementary to them in terms of the specific research issues raised. The specific technology areas that are featured in the new volume have, in most cases, emerged since the last volume was published. One area in which the present volume builds on the previous ones is in the development of the WWRF system concept, which is described in Chapter 11. This concept is a significant development beyond that expounded in the first volume, and is based on the requirements of the various entities, or stakeholders, that have an interest in the future of mobile communications. It builds on the user-centred approach that was pioneered in the previous volumes. Each of the volumes comprises the contributions and views of hundreds of researchers from industry and academic institutions, brought together through discussion and the synthesis of white papers from individual contributions at WWRF meetings. The editors would like to acknowledge these contributions.

1.2 The Wireless World Research Forum The Wireless World Research Forum was established in 2001 to bring together industry and academic researchers to create a clear view of the future of mobile communications and of Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

2

Technologies for the Wireless Future – Volume 3

the research issues that need to be addressed [3]. The Forum has developed over the years, but has not lost sight of its original mission. New working groups and special interest groups have been formed to tackle new problems and opportunities. Links have been established with new industry, research and standards fora to make sure that our messages are heard by those who will determine the future technology of our industry. The goal of WWRF is to develop a common global vision for the future of wireless, to drive research and standardization. This is to be used to influence major decision makers’ views of the wireless world to ensure that the right technology is in place to achieve our vision. WWRF also brings together partners to form powerful R&D collaborations. For instance, it played a major role in the establishment of WWI (the Wireless World Initiative), a connected set of projects under the EU’s 6th Framework Programme, which comprised over 100 organizations with a budget of over 100 million euros [4]. Countless smaller projects have also been set up through relationships established at WWRF or through technical discussions that have identified important research issues. It should never be forgotten that we are advancing the frontiers of wireless technology to better serve our customers, and WWRF has led the way in promoting user-centred design for mobile and wireless telecommunication systems [1]. The objectives of the forum have been established as follows: ž To develop and maintain a consistent vision of the wireless world. ž To generate, identify and promote research areas and technical and society trends for mobile and wireless systems towards the wireless world. ž To identify and assess the potential of new technologies and trends for the wireless world. ž To contribute to the definition of international and national research programmes. ž To simplify future standardization processes by harmonization and dissemination of views. ž To inform a wider audience about research activities that are focussed on the wireless world. An example of the harmonization of views is the so-called ‘Cross-Forum’ group that WWRF has established with similar regional bodies, including mITF (Mobile IT Forum) (Japan), NGMC (Next-Generation Mobile Communications) (Korea) and the Future Forum (China). The Cross-Forum meets regularly to exchange information and views on the development of new mobile technologies in the different regions. WWRF has also set up liaison agreements with other bodies to ensure a regular flow of information between them. Examples include: the UMTS Forum, IEEE Communication Society, the SDR Forum, the Autonomic Communications Forum (ACF) and the eMobility European Technology Platform. The WWRF is open to all those who can sign up to our objectives. Different levels of membership are available. Our founder and sponsor members (currently comprising: Alcatel-Lucent, China Mobile, Ericsson, France Telecom, Huawei, Intel, LG Electronics, NEC, Nokia, Nokia Siemens Networks, Nortel, Research in Motion, Samsung and Vodafone) have seats on the Steering Board. Full members make up the remainder. In all, there are over 140 members from five continents, consisting of academic institutions, network operators, manufacturers, regulators and research organizations. All members are represented in the General Assembly, see Figure 1.1 for the structure of WWRF. Executives are elected by the General Assembly, including the Chairman, Treasurer and the Vice-Chairs from three regions: EMEA (Europe, Middle East and Africa), the Americas and Asia. Together with the chairs of the Working Groups and the Special Interest Groups, and the founding and sponsor

Introduction

3

Chair General Assembly Secretariat Steering Board

Vision Committee

WG6: Cognitive Wireless Networks & Systems

WG5: Short-range Radio Communication Systems

WG4: New Radio Interfaces, Relaybased Systems & Smart Antennas

WG3: Communication Architectures

WG2: Service Architecture

WG1: Human Perspective and Future Service Concepts

SIG1: Spectrum Topics

SIG2: Security and Trust

SIG3: Self-Organization

SIG4: Convergence in Home and Enterprise Networks

Figure 1.1 The structure of WWRF

members, they form the Steering Board, which runs the Forum in line with the Articles of Association and the decisions of the General Assembly. The Working Groups and Special Interest Groups are at the heart of WWRF’s activities. Each is responsible for developing white papers and briefings exploring the research agenda in its own area of operation, and for developing a compelling agenda of invited talks and contributions at its meetings. The groups meet during the WWRF plenaries and according to need between them. Collaborative working tools such as Wikis are also used to develop documents.

1.3 Current Situation and Trends At the beginning of 2008 the following situation and short-term trends can be observed: In Western Europe a saturation point has been reached in terms of the number of mobile phone users, with more than 100 % user penetration. However, a majority of voice calls are still generated by fixed-line phones, meaning that there is still significant potential for growth of mobile airtime, if not in terms of number of customers. In addition to ever more competitive tariffs, an important driver might also be the even-better voice quality announced for 2008 with the wideband AMR voice codec for UMTS. Currently operators and manufacturers also put major efforts into further development of data network performance. Specifically, the HSDPA (High Speed Downlink Packet Access) version currently being deployed at 7.2 Mbps and even higher bit rates will be offered on the downlink, and with HSUPA (High Speed Uplink Packet Access) 1.4 Mbps and more will

4

Technologies for the Wireless Future – Volume 3

be offered on the uplink. Further developments are already planned under the name of LTE (Long Term Evolution), including R&D projects and trials with the aim of achieving significant advances in the up- as well as the downlink during the next few years, as specified by NGMN (Next Generation Mobile Networks) [5]. In terms of growth of number of subscribers, a large increase can be observed in emerging markets such as Africa, China and India. Particularly in Africa, mobile communication, which in many cases means having phone communication for the first time, is a key element in improving the economic situation of countries. Worldwide, numbers of phones are showing strong growth, and there will be 3.5 billion mobile customers by the end of 2008. In 2007 alone, 1.14 billion mobile terminals were sold worldwide, clearly showing that the mobile market is still one of the largest ‘high tech’ markets. The appearance of the iPhone from Apple has initiated a lot of discussion on the design, man–machine interface (the touch screen used with fingers and no stylus), future applications and further innovations (such as using sensors for automatic orientation of the screen) of mobile phones. In addition, there are new business models to be considered (where for instance the handheld manufacturer takes a share of the airtime revenue). Another area with lots of innovation and competition is the operating system (OS) for mobile terminals, and smart phones in particular. The following OSs are amongst those currently competing: ž ž ž ž

The market leader, Symbian Microsoft Windows Mobile Research in Motion Various versions of Linux, such as the Android announced by Google or those from the LiMo Foundation (Linux Mobile Foundation) ž MAC OS X ž Palm OS. When it comes to new applications, the following trends can be seen: ž For navigation in cars, dedicated devices with built-in GPS and maps and a small screen of a few inches in width are becoming more and more successful in the market place. A new trend in Europe is to have a terminal with built-in GPS, so that navigation will also be possible using the mobile terminal. The acquisition of Navteq by Nokia for several billion euros provides evidence for the ever-increasing importance of navigation and related services such as points of interests. ž Mobile television, already a success story in some Asian countries like Korea and Japan, is coming to the European market, facilitated by the standard DVB-H. ž Download of music is becoming more and more important and is already offered by operators as well as handheld manufacturers. ž Push email and PIM functionalities, as pioneered by Blackberry/RIM, are highly valued by many business users. ž Social networks such as MySpace and Facebook are becoming important on mobiles as well.

Introduction

5

1.3.1 Next Generation Mobile Networks Next Generation Mobile Networks (NGMN) is an initiative by a number of mobile network operators to make recommendations for the creation of networks suitable for the competitive delivery of mobile broadband services and cost-efficient eventual replacement of existing networks [5]. It now has 16 network operator members, with a further 26 manufacturer sponsors and two advisor universities. NGMN is looking for a platform for innovation by moving towards one integrated network for the seamless introduction of mobile broadband services. This network will coexist with other networks while it facilitates smooth migration from them. The target architecture defined by these recommendations is an optimized Packet Switched (PS) network architecture, which will provide a smooth migration of existing 2G and 3G networks towards an IP network with improved cost competitiveness and broadband performance. This is a further step in the evolution of current industry efforts in HSDPA, HSUPA and EVDO arenas enabling a personalized broadband access experience and consolidating the diversity of networks operated by mobile network operators. NGMN addresses other issues affecting the mobile industry, such as management of Intellectual Property Rights (IPR), interworking of different technologies and operational aspects of running successful services. Recommendations will be submitted to appropriate standards bodies. In general NGMN operates on a shorter-term, more focussed agenda than WWRF, but there is an ongoing liaison between the two bodies to ensure that information and ideas of significance can be exchanged. 1.3.2 The International Telecommunications Union (ITU) Since the publication of the last volume, there have been significant developments within the ITU. The World Radio Conference (WRC), held in Geneva in October and November 2007, established the allocation of spectrum for mobile services in the period [6]. In particular, additional spectrum was made available for IMT Advanced or 4G. Over the next 15 years the following spectrum will become available in various countries at varying times: ž ž ž ž ž

450–470 MHz band 698–806 MHz band in Region 2 and nine countries of Region 3 790–862 MHz band in Regions 1 2.3–2.4 GHz band 3.4–3.5 or 3.5–3.6 GHz band (no global allocation, but accepted by many countries).

The regions can be roughly defined as: Region 1 – Europe (including all of Russia), Middle East and Africa Region 2 – North and South America Region 3 – Asia and Pacific.

6

Technologies for the Wireless Future – Volume 3

In addition, the IMT Advanced concept developed by ITU-R WP8F (Radio Sector Working Party 8F) has progressed, thanks in part to contributions from WWRF, to the stage of inviting candidate technologies to be considered.

1.4 Overview of the Following Chapters Having introduced this book and the WWRF, we have presented some important aspects of the current situation and short-term trends. Here we give a short overview and discussion of the remaining chapters of the book. In Chapter 2 the stakeholders, such as users, operators and content providers and their requirements are discussed. This provides the starting point for further discussions about requirements and concepts from more specific viewpoints, as presented in the following chapters. Future trends and the vision of the WWRF are also presented and discussed in this chapter. Mobile networks have already initiated a very successful revolution in communications. Nevertheless, many – including the WWRF community – are convinced that truly mobile communication is only just starting. This assertion can be put into perspective by looking at some future application scenarios, including a home scenario in California, one in rural China, business activities in Frankfurt and public services in Kenya, as detailed in Chapter 3. The methodology used to derive such scenarios and requirements and innovative business models is also given. To facilitate innovative services and applications to the end users beyond just voice and connectivity, an appropriate service infrastructure is critical. Key aspects of such an infrastructure presented in Chapter 4 are: a machine-understandable semantic service description, how to create services, and the key requirements as well as concepts of a future service architecture. The layered structure of the WWI (Wireless World Initiative) reference architecture is described in Chapter 5. This includes the reference points such as the Ambient Network Interface (ANI), the Cognitive Service Provisioning Interface, the Decision Making and Reconfiguration Management Interface, the Self-Configuration and Self-Management Interface, and the Application Interface. Three functional architectures, Heterogeneous Radio Resource Management (HRRM), Mobility Management and Context Awareness, are also discussed. In Chapter 6 the focus is on physical layer aspects for future air interfaces. A number of aspects are detailed in more depth here. For error correcting codes, LDPC codes that approach the Shannon limit are investigated. Advanced multi-dimensional radio channel models are needed to meet the requirements of future wireless systems. The requirements include new frequency bands and propagation scenarios, large bandwidth, advanced multi-antenna structures and new network topologies. The state of the art of multi-dimensional radio channel modelling and recent developments in channel model standardization activities are reviewed. Finally, a review and comparison of linear and nonlinear multi-user MIMO precoding techniques is presented. Another ‘wireless approach’, complementary to the cellular wireless approach and gaining more and more importance is short-range communication, is presented in Chapter 7. Aspects elaborated here are: Ultra Wideband Radio over Optical Fibre, Integrative and Cooperative Aspects, High Data Rate Wireless Communications in the Unlicensed 60 GHz Band, and UWB: Applications and Limitations.

Introduction

7

Today a wide variety of air interfaces are in use, including for example: GSM/GPRS/EDGE, UMTS/HSDPA/HSUPA, different standards of WLAN, Bluetooth and others. In the future, this heterogeneity may well increase further as NFC (near field communication) or LTE air interfaces come along. One interesting approach to managing this heterogeneity is to use ‘cognitive networking’. In general, cognitive systems are able to retain knowledge from previous interactions with the environment and to behave according to this knowledge, together with other goals and policies. This enables the system to adapt to external stimuli and optimize its performance. In Chapter 8, two subjects are presented in more depth: ž The basic characteristics of reconfigurable platforms, with an explanation of how they can support cognitive networks. ž Policy-based management and autonomic radio resource mechanisms that use the cognitive networking paradigm. One key enabler for mobile communication has been and will always be radio spectrum. Today, spectrum is either used as licensed spectrum, typically allocated to a specific operator, or used in unlicensed bands, as is the case for W-LAN or Bluetooth. In the future, different approaches to spectrum sharing may be used, and a number of such approaches are considered in Chapter 9. The private home is an important application domain for networking. A deeper analysis of future ultra-broadband Home Area Networks, including requirements and application scenarios for bandwidth greater than 1 Gbps, is presented in Chapter 10. Finally, in Chapter 11, a combined view of the system concept for the future is summarized. The goal is to specify an environment that can realize our vision of the future wireless world where users are able to access, anytime and anywhere, services that best fit their preferences and environment. Context-aware applications will provide relevant information about users to enable this. The requirements and system concept are discussed. The latter includes all aspects, including applications, services, the IP-based communication subsystem, access networks, reconfigurability and cognitive wireless networks.

References [1] R. Tafazolli (ed.) “Technologies for the Wireless Future: Wireless World Research Forum (WWRF)”, John Wiley & Sons, Ltd, 2005. [2] R. Tafazolli (ed.), “Technologies for the Wireless Future Volume 2: Wireless World Research Forum (WWRF)”, John Wiley & Sons, Ltd, 2006. [3] [4] [5] [6]

http://www.wireless-world-research.org/. http://www.wireless-world-initiative.org/. http://www.ngmn.org/. http://www.itu.int/ITU-R/index.asp?categoryDconferences&linkDwrc-07&langDen.

2 Vision and Stakeholder Requirements for Future Mobile Systems Edited by Dr Nigel Jefferies (Vodafone Group R&D, UK)

In its work, WWRF has dedicated itself to the derivation of two families of requirements for future mobile systems: stakeholder requirements and system requirements, see Figure 2.1. The idea is to provide a framework and rationale for the technology to be developed within the coming years. In this chapter we consider a motivating vision of the development of mobile communications by the year 2017. Following this, we will develop a set of stakeholder requirements that will be used in the light of the technology developments discussed in the intervening chapters to create a system concept as described in Chapter 11.

2.1 A Vision of 2017 In [1] we identified the following as a key visionary statement from WWRF, which would motivate our efforts to provide better mobile communication for more people: 7 trillion wireless devices serving 7 billion people by 2017

This essentially means that the entire world population will be served by wireless communicating devices. The simplest forms of wireless devices will have to be affordable and easily operable. A thousand devices per user seems ambitious as a target, but is achievable, though it presents us with huge challenges. As the vision becomes reality, the variety of devices serving people will become huge. Many new cars already contain tens of communicating devices. Communication between machines will grow faster than communication between humans. Sensors and RFID tags will be added to more and more goods, and they will communicate wirelessly. Part of their role will be to provide context sensitivity. Sensors are to be embedded in, for example, vehicles, transport systems, weather systems and building infrastructure Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

10

Technologies for the Wireless Future – Volume 3

1 – Elaborating the Vision

Usage trends

+

Technology perspectives

Current situation Identifiable trends Projection up to 2017

2 – From the Vision to WWRF System Requirements

Vision Stakeholder requirements Reference model

System requirements and concepts

Figure 2.1

Development of the WWRF system concept

(furniture and lights for context sensitivity, doors and windows for security). With a tag and a sensor in a package of food, one could read with a handset the origin and history of the food. This would improve safety of food. Sensors and tags will begin to inhabit every object. There will be an emergence of smart sensors with local intelligence. Most of the devices will be part of the mobile Internet, connected, independent of wires and ubiquitous. The IP networking layer will be the common integrating layer to deliver services to the end users, while sensors and tags could use more simple communication protocols.

2.2 Stakeholder Requirements Each stakeholder or participant in the system has a distinct set of requirements that must be captured and considered in any system design. 2.2.1 User Requirements We start our consideration of user requirements by identifying a set of needs and values that drive human activity. These will then provide a framework for the adoption of, and interaction

Vision and Stakeholder Requirements for Future Mobile Systems

11

Capability Plane

Context Awareness

Personalization

Basic functionalities needed to meet the user needs

Ubiquitous Information Access

Ubiquitous Communication

Value Plane

Accessibility

Values for interaction with technology

Reliability

Trust

Utility

Control

Privacy

Needs Plane Core human needs relevant to the wireless system

Human Capability Augmentation Self-Actualisation Belonging Safety & Security Subsistence

Figure 2.2 The needs, value and capability planes for the wireless world

with, future mobile services and applications. The top-level requirement structures that can be used and adapted to the situations necessary are divided into three planes: ž The Needs Plane, which includes core human needs relevant to the wireless system. ž The Value Plane, which expresses the values for interaction with technology. ž The Capability Plane, which identifies the basic functionalities needed to meet the user needs. The needs, values and functionalities that can be affiliated with each plane are presented in Figure 2.2. These planes are a core part of the user requirements for the wireless world. The contents of each plane will change over time with trends in society, with changes in lifestyle and with new technological developments. The requirements are a mixture of core human needs, the values to be met using technology, and basic functions that the user experiences. 2.2.1.1 The Needs Plane The following core human needs, focusing on values that apply specifically to human interaction with technology, are relevant to wireless systems: Subsistence Some user populations, for instance in the developing world, will require mobile technology to address basic subsistence needs: finding work, realizing additional income opportunities, purchasing basic goods at lowest possible prices, managing perishable goods to avoid waste, organizing childcare, creating better housing, etc. Such services need to be affordable by the intended users. It is worth pointing out that in many developing countries, future mobile services may not have to compete with existing technologies and infrastructures (such as the Internet and fixed phone systems), and thus both rate and scale of adoption of a

12

Technologies for the Wireless Future – Volume 3

suitable wireless technology may be high (which could put strain on the system). Additionally, services in these environments need to be robust and accessible by populations with low literacy. Safety and Security Technology must be safe to use, and be perceived by potential customers to be safe. For any future application or service, a risk analysis should be carried out to identify potential risks to users’ safety and security. Devices may be valuable, and so may the data stored on them. Unintended usage for undesirable purposes (for instance, SMS bullying and so-called ‘happy slapping’ among children) needs to be identified, and countermeasures designed in. The wealth of data created through ambient services, for instance, offers many possibilities for abuse (for instance, for tracking and sophisticated stalking). In addition, there will be demand for an increasingly sophisticated range of applications and services that enhance personal safety – in the home, at school or while travelling. This can be achieved by monitoring of persons or areas, providing information about risks, and information about how to avoid them, but such services must be reliable and effective. Remote monitoring of vulnerable people (such as children and the elderly) is likely to increase. Furthermore, there are likely to be demands for services that allow local communities to take an active role in their organizing their own security. At a corporate level, demand for tracking valuable, perishable and potentially dangerous goods, and remote management, will increase. Those responsible for public safety and security will seek to make increasing use of technology in general and mobile technology in particular, for tasks such as transport of dangerous goods, managing crowds, access control, securing access and tracking suspicious individuals. Belonging Most individuals belong to a multitude of groups and social networks, such as families, local communities, and social and professional peer groups. Technology, applications and services must therefore support the interactions and activities defining these groups. However, different communities can have highly specific needs and these must be accommodated. One example is the way in which a professional business operates compared to a recreational, hobby environment. The technology structure must accordingly be able to support communication within groups and communities, as well as across geographical, language and social barriers. Initiating and maintaining human relationships will continue to be a significant driver of demand for communication services. One-to-one communication will continue to predominate, and there is a need to make such communication easier for user groups who struggle with current mobile technologies, such as disabled users and the elderly. For many such groups there will be a demand for low-cost services. Growth is likely to occur for applications that support group communication. With increasing mobility, particularly among younger people, the need to maintain relationships over large distances and for long periods of time is increasing. Multimedia communication is increasingly used for personal communication. There is a need for the user to control which communities they are part of. Some kind of negotiation with current members is needed before they can be seen as part of the community. In concert with other members, users should be able to form, control and manage the rules related to the communities and groups they are part of. The group policy determines the rules. Many communities cut across the work and home domains, calling for mobile applications and services able to support the user in their multiple roles.

Vision and Stakeholder Requirements for Future Mobile Systems

13

Ubiquitous support is needed through a variety of media such as text, speech, graphics, still and moving images, 3D, haptics and olfactory sensing. Ubiquitous communication, ubiquitous access and presence awareness are key prerequisites to support initiation and maintenance of relationships and communities. Trust is essential for keeping and evolving relationships and communications in communities and networks. Safety and security therefore also needs to be in accordance with this. Self Actualization Human behaviour is essentially goal-driven, and users will assess mobile services and applications by how useful they are in achieving their goals. These achievements may be professional ones, or part of personal growth and development. Services that support self actualization need to provide fast and easy access to relevant information and expertise. People increasingly accumulate personalized data on mobile devices that are not only valuable, but may be essential for them to reach their goals. Consider the impact on a busy person of losing access to their diary. Availability and integrity of personal and professional information is vital to building trust with users who will rely on future services and technologies. Systems should support the transition from supply-driven to user-centric demand-driven value creation. Consumer electronics (CE), information technology (IT), telecommunication and media industries are converging, and more and more services will be IP-based. Increasingly, mobile services tend to be user-centric, demand based and context aware. This means that the end user gets a more central and active role in mobile services value networks, for instance by not only consuming but also producing content. This user-centricity should be one of the main concepts in the architecture design. The production, elaboration and delivery of user content needs to be supported. Multi-device and multi-channel content self-production should be supported, with special regard to storage, delivery, control mechanism and possibilities for moderation of the content. Increased user content production can only be supported by making system functions simple to acquire, set up, use, manage and retire. Services have to be at the forefront in terms of helping the user to discover their benefits. The usage experience should be seamless and transparent for the user. With new devices the user profile, services and applications should be automatically available. The user should also be able to easily and securely transfer their data from one storage provider to another one without having to put too much thought into it. The management of interorganizational cooperation between many different companies and stakeholders is required. For the provisioning of mobile services, cooperation between different actors in a complex value network is necessary. A value network is a generalization of the traditional value chain reflecting the more complex relationships that exist between commercial entities. The architecture should support this interorganizational cooperation. Examples would be enabling monitoring of the performance of the value network (for instance, sales information from an entity that sells the service could be distributed to and used by other relevant network entities), or supporting revenue-sharing mechanisms that allow for fair distribution of revenues. Technological standards and interoperability play an important role in the orchestration of cooperation on a technological level. Human Capability Augmentation Applications and services can augment human capabilities in two ways: by helping to overcome missing capabilities or by enhancing existing capabilities. Examples of the former are applications and services that increase mobility and independence of people with physical

14

Technologies for the Wireless Future – Volume 3

disabilities or enrich the communication possibilities of those whose vision or hearing is impaired. An example of capability augmentation for all people is an application or service that overcomes the fallibility of human memory by providing reliable records (‘memories for life’). However, when contemplating such enhancements, we must consider the potential downsides of enhancement; forgetting has a key function both for individual performance and well-being and for human relationships (see ‘Belonging’, above). 2.2.1.2 The Value Plane This plane contains values for interaction with the wireless technology and these are naturally linked to the functions of a device user interface (UI). The UI basically defines the user experience and user preferences and how these are supported. The following items are some of the most important user needs when it comes to the value plane: Accessibility Accessibility is essential to many users. For example, if the keys of a mobile device are too small, or the text on the display is too small to be read, the user will not be able to use the device. Accessibility requirements depend on the user and their environment. Older people, for instance, may not need to have the same functions available on devices as younger people, whereas they may need the display to be bigger. Furthermore, some functions may depend on the availability of, for example, reliable power and battery supplies; some displays can drain the batteries so that the device is low on power when important functions (such as a phone call) need to be carried out. Accessibility is also a matter of how well the user can operate the device and set up the wireless network for information and communication purposes. For example, there is a need to simplify the setting up of WLAN profiles for ordinary users. Trust and Privacy Privacy is the claim of individuals, groups and institutions to determine for themselves when, how and to what extent information about them is communicated to others. Trust can be described as the firm belief in the competence of an entity to act dependably, securely and reliably within a specified context. Generally, even though users want to be in control of trust issues, they don’t want to maintain and configure their trust all the time. Users require that control over sharing information is balanced with automation, and is always in the perceived control of the user. The system should enable easy identification of users. Easy identification becomes important when consumers are accessing a multitude of systems and services. Identity management can be a valuable asset and control point. Those who can manage consumer identity effectively and are trusted to maintain user privacy can sell this service to others. For simplicity (from the user’s point of view), there should be only a few necessarily trusted entities. Users want to have control of information created about themselves; they also want to know who has information about them, and what information. The level of privacy should be adjustable under the control of the user according to the ongoing task. If the required level of privacy and trust is threatened, the user should be informed.

Vision and Stakeholder Requirements for Future Mobile Systems

15

Due to mobility it will be important to harmonize related national and international legislation. Standardization is needed to homogenize the systems of security, and their accesses. Single sign-on is needed to avoid unnecessary complexity. Control Lack of control makes many users anxious or stressed when it comes to using wireless systems and the information which is linked to the individuals or groups. The fact that a network, or server, is processing information and provides many users with a lack of control and therefore there is a need to establish a special trust relationship. Users must be able to ‘switch off’ when they feel like it. The feeling that users are always accessible to others can in many situations be a living nightmare. Therefore, it is necessary that, again, the user is in control so that it is possible to become out of reach in a period of time, or simply out of reach to some selected persons. Communities must be able to screen participants and restrict access. This is particularly relevant when communities are experiencing intruders or persons who misuse their access to, for example, a particular web site. Reliability The main concern for the user in terms of reliability is the robustness of the interaction with the device. The reliability may be in relation to the accuracy of the text entry mechanisms (e.g. handwriting recognition) and speech input (e.g. voice controlled dialling) in noisy conditions. It may simply be a matter of ensuring that the device is stable and robust enough to handle different uses and user situations. Also, the way different devices communicate, for example through Bluetooth connections, needs to be reliable. Furthermore, the stability and performance of the operating system and underlying software is of high importance. From a user perspective, the reliability should always be high and without discussion. However, there may be a need to define different reliability levels for different users and usage situations. Utility Utility is for the user a matter of evaluating the ‘whole package’ of using the technology. This includes the devices, the network connections, the services available, the reliability and robustness, the security and trust, the level of control, etc. This has to be taken into consideration along with the fact that users do not want to know about the technology or network but only to use the services and devices. Personalization can be expected to be a main parameter for adding utility to a technology in future. However, this is a challenge for the service providers to find the right balance between user needs and non-needs. 2.2.1.3 The Capability Plane In this section the capabilities of the system required for it to meet the basic human needs described earlier are listed.

16

Technologies for the Wireless Future – Volume 3

Context Awareness Context is any information that can be used to characterize the situation of an entity. An entity is a person, place or object that is considered relevant to the interaction between a user and an application. Devices with context-aware recognition and awareness are seen to have great potential in next-generation wireless systems. Firstly, context awareness may help people in managing complexity. Secondly, context recognition may serve as a foundation on which altogether new types of mobile applications and services can be built. Thirdly, it is a key part of facilitating autonomic communications. It is expected that users in the future will use many different entities or devices and that they will expect that their personal data can be obtained independently of entity. This means that the exchange of context information should be supported and that there have to be mechanisms to link information to context ontologies. It should be easy for the consumer to discover services suitable for the context (place, situation). Many consumers are not even conscious of existing services. Easy discovery is especially essential in services that are to be used in the mobile context. Personalization (Broad and Narrow) Personalization is a process that changes the functionality, interface, information content or distinctiveness of a system according to the explicit or derived wishes of a user. Personalization can be system- or user-initiated. When a system initiates a change in accordance with the perceived preference of a user, we usually think of the user preferences as being stored in a user profile. In current thinking the user profile is initially set up by the user, but the system has some means of learning more about the user’s preferences and changing the profile automatically. The motivations for using personalization fall into two areas, those that are primarily to facilitate work and those that are primarily to accommodate social requirements. The former motivational category contains three subcategories: enabling access to information content (e.g. personalizing news), accommodating work goals (e.g. customizing the tool bar in MS Word) and accommodating individual differences (e.g. selecting large font size to accommodate visual impairment); the latter contains two subcategories: eliciting an emotional response (e.g. selecting a nature-themed wallpaper for one’s PC to counter-balance stress) and expressing identity (e.g. selecting a ringtone that reflects one’s personality). Users should be able to access and modify their personal information, profile and preferences regardless of the place, time and devices they are using. Users may use many types of device. Ideally, they should be able to access and change their personal data, preferences and profiles with any device so that the information is updated consistently everywhere. Personalization technologies are required to be highly accurate in order to meet user needs. The performance accuracy will depend on the application or service supported by the personalization function. The customer should be able to choose services that meet their need and the kind of features these will include. Services should be able to be constructed from collections of service components and thus possible to use according to customer expectations and decisions. In mass customization the customers define the content of the service by themselves, enabling services.

Vision and Stakeholder Requirements for Future Mobile Systems

17

Ubiquitous Information Access and Communication Ubiquity refers to the ability to offer the experience of services with an acceptable level of service anywhere, anytime and with any access device. Services may include communication services, access to or management of contents, and access to or management of applications. The services should be able to be delivered over different platforms. The content and the delivery of services have to work separately. Content and communications may be included in different databases with different specifications. Depending on the used device or method of using the service the system has to support the usage and deliver the service with right definitions. Also, different methods for producing the service may be required. Devices for using the services are currently very complicated; many of their features are not used in an optimal way. Future systems must support convergence with currently disjointed networks and services: fixed wireless, mobile media, home technology. As a result of convergence, technologies that were applied in one area of application should now be available in another area. This should allow new competition and create competing business models, so both peer-to-peer and centralized paradigms could be supported, for instance. 2.2.2 General Market Requirements Over the last few years there has been an explosion in new information and communication technologies. A converged, intelligent IP-based network that integrates data, voice and video provides the foundation for an endless set of applications designed to make people more productive and businesses more competitive, all by increasing efficiency, saving time and reducing cost. The convergence processes extend and open many new opportunities for new communication services. These will play an important role in the global economy, creating new markets. Convergence is and will be an important catalyst for developing demand for new services and developing the capability of industry to satisfy these demands. In particular, future mobile systems will put new demands on the network to fulfil the user expectations of the future communication society. These demands may lead to new and profitable business opportunities on service delivery and content provisioning. 2.2.2.1 Blurring Business Roles Radical changes evident during recent years in the field of communication are first of all related to the offer of new services, available as the result of interpenetration of many convergent communication processes. The convergence of traditional telecommunication systems, Internet-based systems and the emergence of new applications all require new business models. The borders between traditional roles and administrative domains (network provider, content provider, service provider retailer) are blurring. Even a user may become a service provider, content provider or retailer. Additionally the roles may change in the same active context, implying a very flexible business model. Individual users may also appear in different roles: as employees, as private persons or in other specific roles.

18

Technologies for the Wireless Future – Volume 3

The access networks of the future will comprise several types of wired and wireless technologies. Therefore, the future will be based on seamless fixed-mobile convergence in the sense that a user may seamlessly move between different access technologies while maintaining a communication session. Effectively combining new technologies will provide full coverage for voice, data, multimedia or any combination of these services and at the same time operators are assured that the network selects the most cost-effective method of delivering these services. The trend with B3G systems is that the customers are increasingly interested in a wide service portfolio rather than the use of a specific access technology. Customers will connect to an access network based on a combination of personal preferences. This means that, in the new B3G environment, the end user will have an interest in customising their services with regard to parameters such as quality of service (QoS), security, price and network operator selection. Rapid service creation and delivery will be among the most vital success factors in that market. The need for personalization will create new business models based on value creation, which will change the existing mobile service value chains. The traditional role of mobile operators will change. Open B3G systems will invite new players to enter the market, shifting the focus from competition on geographical coverage and price to competition on services. B3G systems that are open for a wide variety of access types will strongly affect the business potential for operators. Competition will be heavily increased on nearly every network segment from the radio interface, access supplier and service and content provisioning. The success factors for the telecoms business will be similar to those that are important to the users. In B3G systems, it is evident that the network operators will be dependent on external service providers and content providers to be able to support the customers with the services they demand. At the same time the service and content providers and virtual network operators will be dependent on the infrastructure owners to be able to deliver the services to the users. 2.2.2.2 Augmented Environments as Part of the Ubiquitous Communication System Augmented environments superimpose computer enhancements on the real world. Such augmented environments are well suited for collaboration of multiple users. To improve the quality and consistency of the augmentation the superimposition of real objects by computer-generated objects and vice versa has to be implemented. Auditory augmentation of the visually dominated everyday environment is a new and very promising approach to creating user-friendly information systems that are accessible to everybody. 2.2.2.3 Sustainable Development Sustainable development is a global objective and means adopting business strategies and activities that meet the needs of the enterprise and its stakeholders today while protecting, sustaining and enhancing the human and natural resources that will be needed in the future [2]. Economic development must meet the needs of business and the community. This includes shareholders, customers, employees, suppliers and those communities who are affected by the organization’s activities.

Vision and Stakeholder Requirements for Future Mobile Systems

19

It is important to ensure that the sustainable development objectives that are established complement the enterprise’s existing competitive strategies. In other words, sustainable development should provide an additional dimension to new business strategy. It should also be emphasized that sustainable development cannot be achieved by a single enterprise in isolation. Sustainable development is a pervasive philosophy to which every participant in the global economy must subscribe. Sustainable development should become a central objective of all sectors and policies in the B3G era. In particular, B3G systems must contribute to a better quality of life and help improve and simplify everyday life for people and companies. New technologies must be used to promote sustainable growth of the societies. 2.2.2.4 Supporting Value-based Pricing Mechanisms Previously accepted pricing structures for mobile data services, based on charging per megabyte, are insufficient for the introduction of wireless broadband services. Charging per megabyte would make some services too inexpensive to generate sufficient revenue, whereas others would be too expensive to create any demand. Therefore, pricing needs to be value based; for example, measured from demand or related to context of use. Pricing of services needs to be taken at a service-specific level to generate satisfactory profits in 3G. The basis upon which each service is charged is likely to vary over services. The system should support the gathering of relevant pricing information and the pricing process. The pricing mechanism of new services will be different from that of traditional telecom services. What people are willing to pay for will be service specific. Important concepts like pricing and revenue sharing therefore need to follow a more value-based approach. Convergence, use of unlicensed frequencies and P2P networks enable new business models. A successful business model needs to accommodate companies with suitable core competencies. The challenge in the B3G environment is to develop charging schemes that adequately support diverse business models while being sufficiently simple for user acceptability and cost-effective implementation. A diversity of business models needs a diversity of charging models. A universal model cannot satisfy all the various requirements. Charging methods in B3G systems have to be much more flexible and diverse and focus on QoS requirements, user profiles and security. In B3G systems, different types of bundling rates have to be provided. Operators will have to change the perspective from managing networks to managing services and experience. New service provisioning will define a complex value chain, in which many business entities contribute to the end service and so share the resulting charges. The new pricing model of B3G services will involve a large number of entities and complex economic incentives. 2.2.2.5 Supporting Revenue-sharing Mechanisms In the network of actors the services are bundled from components created by different actors. Some actors collect the revenues and share them according to some ‘rules’ with other actors. Sharing of investments, costs and revenues is complex and should be such that each actor is properly compensated. The system should support the gathering of information and performance monitoring on which revenue sharing may be based.

20

Technologies for the Wireless Future – Volume 3

The revenue figures for B3G systems will be more unpredictable, requiring businesses to make rapid changes in strategy and service innovation. New business roles and revenue models have to be considered as new opportunities arise and competition increases. In the B3G model, several entities will be involved. The revenue for operators will depend on the role, the service offerings, the charging and the market share per service. Revenue potentials will be shared between the different players in the value chain, such as service providers, content owners and brokers. It is important to develop the right business models between the players, considering their roles and features, as well as implementing tariff procedures and principles for revenue-sharing mechanisms among these players. 2.2.2.6 Supporting Cost-effective Service Provisioning To provision viable and sustainable services, the architecture should support cost-effective provisioning. More and more new services and applications will emerge, and parties participating in the provision of operations will include network operators, access network providers, business operators, software developers and content providers, as well as users themselves. Charging modes will be important to all parties in the B3G value chain, including users, with each wanting to maximize its profit. 2.2.2.7 Generally Markets and Users Require Low-cost Services There will be a clear differentiation between providing low-cost, mass-market services and highly-valued, upmarket services. Users will expect to be able to do more with their communications services, for less money, and are showing an interest in services beyond voice. They will be attracted by services that offer them access to a wide range of communications information and entertainment services in a user-friendly, cost-effective way. Customers will expect to get the service that they want almost anywhere at any time, at a good price and at the required quality. That is why the focus in the B3G era will be I-centric, with different underlying technologies supporting different users’ needs. Rapid service creation and delivery will be vital success factors in that market. This recognizes that services are not just about connectivity but also about content, context, security and price. Some users want to have ubiquitous global wireless data coverage and interconnection. This should be automatic and go hand in hand with data services that work without the user needing to read the manual or call for assistance. Users will continue to be willing to pay a premium for such a service because time is valuable to them and they want to stay connected no matter where they are around the world. 2.2.2.8 The Number of Mobile Users in the Developing Markets is Increasing Rapidly and this Requires New Business Models The number of subscribers for mobile communications has increased much faster than predicted. The majority of traffic is changing from speech-oriented communications to multimedia communications. Services are becoming more and more important for the

Vision and Stakeholder Requirements for Future Mobile Systems

21

future success of mobile operators. Services will increasingly be delivered over the access technology most appropriate to user requirements and circumstances. As technology has evolved and service has changed, operators have had to adopt new, business models. Old business models will no longer work for industry players. Therefore business models will have to evolve and change with changes in technology. The interrelationships between the different companies that work together to provide services to users have also undergone changes, and the new technology itself has changed too. With previous generations, the model was a rather simplistic one. The mobile operator owned the network infrastructure and provided voice services and simple data services to customers. The mobile operator could often control the interactions with other players such as suppliers and collaborators. In the B3G business model, more parties will be involved in the process of providing B3G services and applications to users. The new business model should define the relationships between all acting roles within the system.

2.2.3 Service Provider Requirements The definition of a service provider encompasses anyone who provides a commercial service directly to users making use of a mobile communications network. The following requirements have been identified: 1. Interface specifications should be kept simple, and the number of options should be kept to the minimum in order to reduce the cost of the equipment and to improve interoperability between different network elements. 2. Service providers should have the ability to control Operational Expenditure (OpEx). 3. Service providers should have the capability to monitor the end user experience of their subscribers, even in a roaming situation. 4. Service providers should have the capability to access information at different levels of the system (e.g. core network, access network) in order to ensure QoS levels that are compliant with the established service level agreements (SLAs). 5. Service providers should have the capability to enact specific business agreements with any networks (in case of roaming conditions) for ensuring appropriate resource reservation to the roaming end user in the visited network. 6. Service providers should have the capability to support media sharing, integrating automation of regulatory mechanisms. Portability is a great enhancer of this phenomenon, as it can guarantee even more freedom and agility in sharing and delivering content, perhaps bypassing fixed and wireless network operator infrastructures with local sharing, either through increasingly cheap storage on portable devices or wirelessly over personal networking communication channels. 7. Service providers should have the capability to support free access to applications and services enabled by marketing and advertising. These strategies provide the core engine of some of the most successful enterprises enabled by innovative Internet technologies and business models. Business-to-business revenue streams (e.g. advertisers that buy space or search-related visibility on a portal or search engine) allow the offering of free solutions to the user (e.g. searching on the Internet, on the desktop etc., as in the case of Google and others).

22

Technologies for the Wireless Future – Volume 3

8. Service providers should have the capability to support flexible multi-brand management. Services can be delivered in several ways. When several organizations are working together, several brands may be used simultaneously and there is a risk that customers or end users do not know who is responsible for the provisioning and how they should proceed when they have questions or complaints. The system should support multi-branding and provider recognition. 9. From the service provider’s perspective, the underlying communication infrastructure should be agnostic to the services and applications being offered. They should follow the principle of technology neutrality and the concept of bearer services. 10. Service providers require innovative, converged infrastructures to improve delivery of current services and provide a scalable framework for new, bandwidth-intensive services. Solutions that provide greater network intelligence, integration and flexibility will position them to seize new market opportunities. 11. Service providers need an infrastructure capable of evolving without disruption to provide a wide range of new services. Fast, open service creation, secure services with QoS and flexible billing capabilities are new requirements that can be identified related to the service provider’s role in the B3G market 12. Service providers need also to increase revenue and profit while reducing cost of service delivery. This is possible by offering services that are I-centric. New B3G architectures will satisfy the requirements for rapid migration to an I-centric environment, enabling service providers to deliver the full service flexibility expected by users. 2.2.4 Backbone Provider Requirements The following main requirements are seen for those who provide the backbone network. The main reason to invest in the backbone domain will be the reduction of operating expenses brought by some of the new technologies. 1. Capital Expenditure (CapEx) reduction: ž There should be one network for all types of service. ž Core-infrastructure sharing should be facilitated among the different technology generations from the initial deployment phase, with optimizations to follow. ž There should be the capability to transport all types of communication and content exchange. ž Backwards capability is required. ž There should be minimal dependence on hardware upgrades to provide new features, capabilities and revisions; rather it should be the software that does this as much as possible with remote upload from the network management centre. 2. Operational Expenditure (OpEx) reduction: ž There should be better and cheaper capabilities for policy-driven network management. ž Increased optimal routing is required, without incremental cost. ž There should be fast and automatic reconfiguration of network resources without any loss of data. 3. Networks that meet market and customer requirements: ž Providers will need the capability to act at different levels of the network to ensure QoS levels that are compliant with the established SLAs. ž There should be a diverse set of service classes and means of charging (value-based).

Vision and Stakeholder Requirements for Future Mobile Systems

23

ž Providers will require enhanced robustness for the backbone networks, and provably quick recovery from service disruptions. The protection, restoration and resilience should be built from the beginning. ž Operators will have end-to-end control of services and networks from the initial deployment phase. 4. Technical requirements: ž A multi-service backbone network must be capable of handling different types of traffic with widely differing characteristics, for example: video and voice streaming, Internet traffic and real-time traffic for P2P communication. Multi-service backbone networks need advanced QoS and high network resilience. ž To deliver high quality of experience (QoE), network elements at the network edge and backbone must interoperate to maximize throughput and performance and minimize the impact of congestion. ž Backbone networks usually offer transport services over significant distances, therefore high-capacity optical transmission systems with low cost per bit and per kilometre are required. 2.2.5 Access Provider Requirements The access provider’s goal is the competitive delivery of ubiquitous broadband service. So they need the capability to support full global coverage, anywhere, anytime, and to be able to support fully broadband access. 1. Capital Expenditure (CapEx) reduction: ž The provider should be able to reuse existing antenna systems and sites across all phases of evolution and at the highest spectrum efficiency. ž There should be coexistence of various technologies with no requirement for periodic hardware upgrades. ž Core-infrastructure sharing among different technology generations will be required from the initial deployment phase, with optimizations to follow. ž There should be an optimized solution for backhaul transmission. ž Cost-optimized indoor nodes should be designed. 2. Access providers should have the ability to control their operational expenses: ž Network configuration/optimization should be possible with the minimum of effort. ž To minimize the OpEx, the end user terminal, if leased or under the control of the access provider, should be remotely diagnosable and capable of receiving bug fixes, initialization and system-level upgrades. 3. Future systems should provide backward compatibility with existing networks and mobile terminals should be maintained (including support of the current 2G/3G cellular access systems and noncellular access systems such as PSTN, xDSL and WLAN). 4. Open and standardized interface are required to ensure: ž A multi-vendor environment. ž Rapid network deployment. ž Interoperability and interworking. ž Lower cost. 5. Interface specifications should be kept simple and the number of options should be kept to the minimum in order to reduce the cost of the equipment and to improve interoperability

24

6. 7.

8.

9.

10.

11.

12.

13. 14. 15. 16. 17. 18.

Technologies for the Wireless Future – Volume 3

between different network elements. Interoperability of the heterogeneous systems is needed. Access providers should have the capability to monitor the end user experience of their subscribers, even in roaming situations. Access providers should have the capability to act at different levels of the network (including core and access networks) to ensure QoS levels that are compliant with the established SLAs. Access providers should have the ability to enact specific business agreements with all visited networks (in case of roaming conditions) for ensuring appropriate resource reservation to the roaming end user in the visited network. Access providers should have the capability to support emerging architectures and services like multi-hop ad hoc networks and P2P paradigm with respect to users and application requirements (routing, content aware, security, QoS awareness). Access providers should have the capability to support using unlicensed frequencies. Unlicensed frequencies, like using WiFi to provide broadband Internet access in hotspots, make wireless access provisioning possible for new entrants such as corporations or groups of users. Access providers should have the capability to support P2P models for communication and sharing of data. P2P environments utilize resources contributed by the participating nodes. Control and coordination are at the edge of the network. P2P systems will enable new types of applications and business models. Access providers should have the capability to enable proximity communication (touch paradigm). Proximity technologies (RFID, proximity transactions) can enable use of mobile devices for industries that are currently not utilizing mobile terminals. This may be an opportunity for players controlling the mobile area to provide value for players like retail merchants. Technologies and business models should allow anyone at anytime and anywhere to take the role of the access provider. This should not take place at the cost of reliability, however. The relation between the PAN (personal area network) and PN (public network) management should be defined. Future systems should allow for market-driven revolution and evolution of access provision. Excessive complexity of systems and interoperability of access networks should be avoided. Improved terminal certification schemes need to be introduced to facilitate early terminal availability with high quality. The access provider should have the ability to support multiple radio interfaces. There should be intelligent selection and seamless handover from one RAT to the other.

2.2.6 Equipment Manufacturer Requirements 1. Equipment manufacturers should have the ability to provide interoperable equipment in all parts of future systems. 2. The system should support a diversity of terminals. Adaptation to terminal characteristics may become part of the value chain. 3. There should always be open interfaces between the different parts of the system.

Vision and Stakeholder Requirements for Future Mobile Systems

25

4. Architecture should allow flexibility for appropriate amount of capacity. 5. The equipment should be modular and designed in such a way as to be incrementally scalable, upgradeable and backward compatible with its previous release (or releases). 6. Equipment manufacturers should have the ability to develop proprietary interfaces and service platforms to meet their customer needs. 7. The system should support easy interoperability of equipment. 8. Future systems should be easy to implement. Therefore there should be only as much complexity as needed to fulfil the requirements. 9. The equipment should be energy and power friendly and provide solutions supporting intelligent energy and power management frameworks. 10. Equipment manufacturers should have the ability to bring innovations to the market. Therefore it is important to establish a new environment which will support the process for enhancing quality and enabling innovation, as well as an environment that promotes collaboration and supplier integration. Establishing a new working environment will help maximize manufacturer competitiveness, efficiency, productivity and profitability. 11. Equipment manufacturers should share critical product knowledge at every stage of the production lifecycle. This requires investments in the latest collaboration technology to expedite product development and time to revenues. Equipment manufacturers also need to deploy infrastructure and processes that support the capture and reuse of designs and intellectual capital, thereby creating an environment for increased innovation. 12. Any equipment manufactured must comply with all safety and regulatory requirements, and in case the equipment is to be placed in a central office it should comply with the Network Equipment Building System (NEBS) requirements if the operator so requires. 2.2.7 Requirements Related to Converging Digital Industries and Application Sectors In the future, the telecommunications industry will join with the information technology, consumer electronics, broadcasting and media and entertainment industries to form a common digital industry. Content will be in digital form and usable across the various media and industries. The same will apply to services and applications. The different industries currently have different modes of operation and the merger will change these. One example of converging industries is the emergence of IEEE 802.XX standards from the communication needs of information technology. 1. The following should be supported in the world of converging digital industries: ž Open and spam-free Internet. ž IP-based architecture in all domains. ž Interoperability, commercial and technical. ž Open standardized interfaces. ž Common platforms for software and services. ž Deregulation. ž Coexistence and free competition of IPR protection and open source. ž Security, privacy and trust. 2. Uniform user-centric experience: ž In the near future, B3G systems will support people in their daily life in a flexible manner. Taking into account user habits, trends, competencies and levels of acceptance chances, the systems should provide users with richer, higher quality, ubiquitous

26

Technologies for the Wireless Future – Volume 3

and affordable services. This requires a user-centric converged system providing applications from the telecommunications, information technology, consumer electronics, broadcasting, media and entertainment domains. ž User acceptance is essential to the success of this converged world. From the user point of view, they do not care about the heterogeneity of the system and do not want to know the details of the technology issues. What they need is to access their services with rich and consistent user experience, no matter whether the services are provided via an increasingly heterogeneous communication infrastructure, or via fixed communication links, or in an ad hoc manner. Therefore, it is required that the future converged system will provide a uniform user-centric experience even if the service is conveyed across multiple industry domains, multiple networks and multiple access technologies. 3. Seamless integration with wired world. The gradual migration of today’s wireless communications towards B3G reflects the most recent trend in the communications landscape. Future wireless systems are expected to be recognized as those which can achieve high data-rate transmissions and provide enhanced capabilities, cost efficiency and highly sophisticated services, comparable to those offered by wired networks, for a variety of applications, such as interactive multimedia, VoIP, network games and videoconference. Moreover, future systems are expected to be based on IP technology, yielding into a common, agile and seamless all-IP architecture design, supporting scalability and mobility. 4. Technologies of the wireless world will be used by several industries in different application areas. There are also many requirements from this direction. Such application areas include automotive and health care. 2.2.8 Legal and Regulatory Requirements In every country there are some laws and regulations with which network operators, service providers and other involved business partners of mobile and wireless communication systems must comply. These laws and regulations vary from country to country; the following are some legal and regulatory requirements which are practiced in many countries. There are SAR requirements on radio systems, including radio base station, mobile terminal and other radio devices. 1. Radio power of antennae must be strictly controlled according to respective regulations. This requirement has a direct effect on a SDR system, which supports dynamic and/or online change of radio. 2. Future mobile communication systems should support lawful interception, which is already a requirement nowadays in many countries. 3. Network operators and service providers are usually required to authenticate end users before providing connection or service. In some countries service providers are also required to log communication actions of all users and to store the logs for a minimum time. 4. Protection of user privacy is not only a user requirement; it is also a legislative requirement according to data protection laws, which makes it compulsory for network operator and service providers to take good care of user privacy. 5. Consumer protection laws may have effects on future mobile communication systems, especially on accounting and billing.

Vision and Stakeholder Requirements for Future Mobile Systems

27

6. Some basic functions to deal with emergency situations are required to be provided in many countries, e.g. emergency calls should be free of charge and always available even when a user roams to a visited network; position information of mobile terminals should be available when it is required by a public security body. 7. Future mobile communications networks should not provide or support any service or application that violates copyright of various types of media. 8. The system should be based on real knowledge (not assumptions) of legal issues and avoid dependence on legal areas in which processes are unclear. The more certain the outcome of a legal process, the better for all entities involved. This will also decrease transaction costs as it will not be necessary to make separate agreements to fill gaps in laws or fix the problems in outdated statutes. 9. The system should enable balanced legal rights and allocate them fairly. A business model can benefit if intellectual properties are allocated fairly. That is, a business model that is not based on one party hoarding all the rights, but acknowledges the reasonable expectations of other partners, will have fewer conflicts, promote cooperation and win-win situations, and be more likely to succeed. Unbalanced legal rights, on the other hand, may form a severe threat to businesses. Over-strong privacy protection may make otherwise interesting business models unavailable, while weak privacy protection in many countries outside the EU diminishes trust and slows down business. Likewise, unbalanced consumer protection, unfair allocation of intellectual property rights and so on may either forbid certain business models or decrease people’s trust in businesses. 10. The system should adjust to the legal uncertainties related to user communities. For a business model that is based on a community, for instance using an interest group as a content producer, a community of hobbyists as a distributor, or a social group as a customer, the unclear legal status of those communities is a threat. The business model may turn out to be infeasible if it is not possible to make agreements with a legally nonexistent entity. 11. The user must be able to accept separately the use by each service of their personal information, and the system must provide continually the possibility of using a simple means, free of charge, of temporarily refusing the processing of personal information for each connection to the network or for each transmission of a communication.

2.3 Acknowledgements The contents of this chapter represent the views of the WWRF Working Groups (WG), Special Interest Groups (SIG) and the Vision Committee, but not necessarily the views of all member organizations. It has been put together by the editorial team of the WWRF Vision Committee, who have collected input from the WWRF and external sources, with the most essential input coming from the WWRF Working and Special Interest Groups. The members of the editorial team were: Chair of Vision Committee and editorial team in the first phase: Mikko Uusitalo, Nokia Research Chair of editorial team in the second phase: Jean-Claude Sapanel, Orange Labs Other contributors, in alphabetical order: Andrew Aftelak, Motorola Angeliki Alexiou, Alcatel-Lucent

28

Technologies for the Wireless Future – Volume 3

Stefan Arbanowski, Fraunhofer Fokus Brigitte Cardina¨el, Orange Labs Klaus David, University of Kassel Panagiotis Demestichas, University of Piraeus George Dimitrakopoulos, University of Piraeus Sudhir Dixit, Nokia Siemens Networks Bernard Hunt, Philips Nigel Jefferies, Vodafone Wolfgang Kellerer, DOCOMO Vinod Kumar, Alcatel-Lucent Werner Mohr, Nokia Siemens Networks Angela Sasse, University College London Knud Erik Skouby, Aalborg University Lene Sorensen, Aalborg University Hu Wang, Huawei

References [1] R. Tafazolli (ed.), “Technologies for the Wireless Future Volume 2: Wireless World Research Forum (WWRF)”, John Wiley & Sons, Ltd, 2006. [2] United Nations World Commission on Environment and Development, “Our Common Future”, Oxford University Press, Oxford, 1987.

3 User Requirements, Scenarios and Business Models Edited by Lene Sørensen and Knud Erik Skouby (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark)

3.1 Introduction Dealing with the concept of user centricity in relation to developments in future wireless technologies is a serious challenge. Each new technology needs an individually designed approach to capture the ever changing user needs, specific user groups, specifics of the new technology and the new opportunities it provides. Developments in wireless technologies change the way users interact, communicate and handle everyday situations, and following these changes users need and wish for further changes. This means that there is a need to update current understandings of user centricity following the development of technology. In this relation, an important aspect of perspective related to wireless technologies is the concept of nomadicity (nomadic computing and communications) [1], [2]. In a technical sense, this describes the system support needed to provide a rich set of computing and communication capabilities and services to nomadic users, in a transparent, integrated and convenient form, when they move from one place and context to another. This technical element challenges traditional thinking about usability and user experience and demands that wireless technologies support capabilities that enable independence of location, motion, computing platform, communication device, bandwidth, processing power, etc. [3]. Within user-centred design much focus is on developing new methods and approaches to deal with the changing technologies and user needs, in order to derive necessary confidence in wireless developments. Many user-centred design methods have been developed around wired technologies. In general, these methods try to capture user needs either in laboratory facilities or in situ – where the user experience problems and needs. Wireless technologies increase the need for revision of existing methods, for developing new methods and for taking into consideration the nomadicity concept in the identification of user requirements and usability testing of products, devices and services. Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

30

Technologies for the Wireless Future – Volume 3

This chapter presents selected examples of work in developing new methods and approaches for dealing with the user-centric concept in a wireless world. Examples are found in IST-EU projects such as MAGNET and MAGNET Beyond (My personal Adaptive Global NET and Beyond) [4], [5]. This chapter focuses on providing insight into practical handling of the development of scenarios, user requirements, usability and business modeling, which are all areas closely linked to our application of the user-centric concept. This implies, for example, focus on the user as a person who interacts with devices, applications and services. The concept of user experience is essential to the situation where the user chooses to use and buy a technology, service or application. The focus on users as persons is illustrated by examples of methods where real users have been included in the development of scenarios, identification of user requirements and usability evaluations.

3.2 Scenarios Traditionally, scenarios are used in many commercial and R&D projects to express visions and user requirements. Increasingly, however, they are also used to bridge the social-oriented user requirements and the technical-oriented system capabilities and specifications. Additionally, scenarios are used to visualize to an audience the overall achievements of the technology with respect to the benefits and gains from the user perspective. An extensive review of different scenario approaches was conducted in [6]. The following subsections present two different approaches to developing scenarios. The first example includes scenarios developed in the MAGNET project [4] in order to analyse user needs and visions for the specific project technologies. The second example illustrates the process of developing WWRF Reference Scenarios in order to express a vision for the whole area of interest for the WWRF. This involves both user-centric narratives and descriptions of the expected capabilities of the future wireless architecture linked to the year 2020. The two examples illustrate a difference in both purpose and approach. 3.2.1 Participatory Design and Creativity Enabled by the technological development and driven by the market, designing new technologies for mass markets today demands a strong emphasis on user centricity and commercial potentials. One approach to securing user centricity in the design and development of wireless technologies is the application of participatory design [7], [8]. The overall purpose of participatory design is to ensure that the final applications and services will be adaptive to the users and not the other way around. Involvement of users is central to this methodology. There are different approaches to a participatory design process (see [3]), although there is a tendency that the approaches are converging to a form where the fundamental goal for the design process is to increase efficiency as well as ‘democracy’, i.e. securing the influence of ‘all’ groups involved, including the users. One element essential to applying a participatory design process is the stimulation of communication and learning of multidisciplinary teams where designers, developers and users are present. In order to facilitate and stimulate communication and learning, M¨uller [9] has created an approach to developing a shared language amongst the persons involved. The shared language can be built on external cognitive aids such as pictures, different elements for prototyping and cultural probes – all elements have the function to support users in telling a story and being aware of technology options.

User Requirements, Scenarios and Business Models

31

Looking at a design process, the participatory design principles only support the involvement of users and an exchange of information amongst the developers and users. When it comes to expressing views and needs for the users, scenarios are often used to express narrative stories that show how the users interact with technology in different situations and contexts. However, the involvement of users in a participatory design process means that the users have to be able to envision their own needs in a context, situation and time in the future. Often, or rather usually, this is very difficult for the users and it may therefore be necessary to find approaches to stimulate and motivate a creative thinking of the users. It is therefore useful to combine creativity theory with participatory design. There exist numerous creativity techniques that can be used in a participatory design process. Brainstorming [10] in particular has shown to be useful in stimulating the generation of multiple ideas and facilitating their expression. Using brainstorming often results in increased enthusiasm as well as spontaneity, creativity and a lot of different ideas [11]. Another technique in line with creating a shared language is the picture stimulation technique [12]. This technique encourages participants to think at a problem from different angles and to break away from traditional thinking.

3.2.1.1 Background The above-mentioned combined approach was used in the MAGNET project [4] with the overall purpose of developing user scenarios expressing high-level user requirements. The overall purpose of the MAGNET project was to enable commercially viable personal networks that are affordable, user friendly and beneficial for all kinds of users in all aspects of their daily life. The personal network concept allows users to interact with a multitude of entities wherever they are. Personal networks are related to personal communication networks consisting of a large number of devices, which can be interconnected independently of time and place; furthermore they enable collaborative environments within a distributed network setting to support private and professional activities [3]. Identifying user requirements for a technology that does not yet exist and which, for example in the case of a network technology, the users cannot see and never will see as a technology is a challenge. The solution is mainly related to first making the users aware of the potential benefits to their life of this technology and then to creating visionary requirements that link in detail the expressed potential user benefits to the functional capabilities of the network architecture. It was decided to develop user scenarios consisting of a narrative user story and mix these with high-level functional requirements. Since the user requirements had to be linked to a future, not-yet-existing technology, it was decided to use a futuristic approach to the scenario development; a seven-step template formed the basis for the scenarios addressing historic and present trends as well as expectations on drivers for future developments (see [3] for more details). As a central part of the scenario development, user expectations had to be expressed. It was therefore decided to organize two different workshops: a creative user-focused workshop and an expert workshop in which technical experts and developers from the project worked with the user ideas. The results of both workshops would end up in scenarios with both a user narrative and high-level capabilities of the PN architecture. The above-mentioned approach was chosen as there was a certain focus on a particular group of users in this project – namely users who demanded higher efficiency in treatment

32

Technologies for the Wireless Future – Volume 3

and monitoring in the handling of a chronic disease, as well as enhanced communication and efficiency when consulting medical treatment personnel. A group of 10 patients and health care personnel was invited to join such a workshop. It was seen as important to include all levels of users in the specific environment in order to express all aspects of user requirements, including those of the health care personnel. 3.2.1.2 Creative Workshop The overall purpose of the creative workshop was to identify elements of user requirements from the selected group of users and to be able to develop relevant user scenarios. The elements constituting the workshop were a mix of participatory design and creativity methods and techniques. The creative workshop was built up around the following elements [3]: ž Establishment of a shared language in which the users identified all types of device and technology of importance to them in their daily life and in particular in the treatment of their disease. This exercise built on simple brainstorming in the group of participants. ž A predefined conceptual landscape scenario introducing the participants to the task of being creative and thinking about the future. This was done as a fantasy journey where all participants were asked to close their eyes, listen to the soft music in the background and at the same time listen to and make their own cognitive picture of a story told to them. The story was about a person with a chronic disease and how they, in a very positive spirit, could be helped in the future (though the result and effect of the help was left as an open question). The guided meditation was used to stimulate the participants’ creativity and to help connect now with the future. ž A conceptual text-based scenario landscape: a conceptual, physical paper landscape showing different situations and pictures of how users thought about the future. The pictures and situations were focused on the situations of the present users, and some on technologies stimulating their thinking about future possibilities in their treatment and health management. The landscape was made up by the participants themselves, while the elements (pictures etc.) were premade by the organizers. ž Additional external cognitive aids such as image elements: pictures, words or short sentences. These were produced before the workshop on the basis of an understanding of the users’ situations, environmental trends identified (which might have an influence on the users and their thinking), humans and activities, and relevant high-level capabilities of the technology as envisioned in the project. These image elements were introduced stepwise under the construction of the scenario landscape. The introduction of the image elements had the overall purpose of directing the participants to think more broadly about issues relevant to the MAGNET project. Image elements showed high-level system requirements, economic, ethical and functional issues and questions. ž Modeling wax, used to stimulate the participants to think about mock-ups linked to their ideas and to show their thoughts visibly. Figure 3.1 shows some of the image elements made in relation to one of the scenario landscapes. After the workshop, the scenario landscapes were analysed, and predefined elements of relevance to the MAGNET project (such as personalization perspectives and security aspects) were combined and used as the basis for writing up user scenarios.

User Requirements, Scenarios and Business Models

33

Figure 3.1 Example of a scenario landscape, including image elements as well as links between them. Strings were used to show links between problems and solutions

3.2.1.3 Expert Workshop Shortly after the creative user workshop, representatives from all identified technical areas in the MAGNET project gathered to work together. The overall purpose of the expert workshop was to discuss different technological solutions to support the visions expressed in the user scenarios. The participants in the expert workshop had a rather specific task compared with the users. However, the experts needed to think as a team and were required to be creative (within the frames of the scenarios) at the same time. Elements of the expert workshop were: ž An introduction to the user group, context and relevant issues. This was done to create a fundamental understanding among the technical experts of who the users were and what solutions they would be looking for. ž Teamwork, where two groups worked in parallel to understand different user scenarios and to brainstorm on different ideas, solutions and relevant capabilities to support the user scenarios. The teams were given different scenarios in order to make the workshop more effective. First they were asked to read the scenario. Then they were given questions based on subsets of previously defined, high-level system requirements. The teams were furthermore encouraged to add sub-scenarios to the user scenarios. This gave the experts a chance to express their technology visions from their perspective of the user side. The expert workshop ended with a number of tables and network figures relating to the user scenarios. The results from this workshop were then merged with the user scenarios to construct the final scenarios. 3.2.1.4 Short Reflection The overall approach to the development of the scenarios worked well. Participants in the creative workshop were positive and rather enthusiastic about the idea that somebody

34

Technologies for the Wireless Future – Volume 3

would make their life a bit easier with respect to managing their disease. The results of this workshop were a lot of ideas and visions on how to do that. However, it was clear that the participants had a hard time imagining technologies and possibilities not yet existing. Further, they had difficulties in expressing their needs and requirements for a technology they had not had any chance of testing or trying as a hands-on experience. Following this, the experts in the expert workshop complained a bit about the visionary level in the user scenarios – it was not seen as really visionary. This had some influence on the expert workshop, where some experts initially showed reluctance to participate, thinking that the task was not interesting or challenging enough to result in any data or knowledge relevant and usable in their work. During the workshop, however, the attitude generally changed to engaged and positive. The data from the workshops were novel and useful inputs for the construction of use cases. Later in the MAGNET project, for project strategic reasons, it was decided not to focus so much on the specific application analysed in the workshops, but to look more broadly at health care (and then discard the chronic disease perspective). The results from the workshops have consequently not been used directly in the further work of the project and therefore it is not possible to say anything precise about the acceptance among the technical developers. Details on the scenario writing, workshops and results can be found in [3], where a deeper discussion of the approach can also be found. 3.2.2 WWRF Reference Scenarios Scenarios are not only useful in design and development processes, they can also be used to display specific visions in a more elaborate and detailed way. This has been the case within the WWRF, where an exercise on developing reference scenarios was started in 2006, with a consolidated version expected in 2008. 3.2.2.1 Motivation and Purpose The idea for the scenario development came as a reflection on the WWRF vision for the global wireless future to drive research and standardization for the year 2017. The vision for 2017 is: ‘7 trillion wireless devices serving 7 billion people in 2017’ [13]. Central to this vision is that technologies, services and applications will be based to a higher degree than today on user requirements and commercial market interests. It is therefore important to discuss the wireless future in a common framework covering technology developments and possibilities, user requirements and commercial market interests. One way of identifying and initiating discussion on the technology developments, user requirements and commercial market interests is to display the vision in a more profound and visible way through scenarios. The scenarios can serve to express the vision in all three areas but can also facilitate development of new ideas and possibilities, serving as a platform for identifying new research areas. In spring 2006 it was agreed within the WWRF to develop a common set of user-centric reference scenarios to visually present the views and research areas of WWRF. The purpose of the exercise has been to develop scenarios which can: ž Relate the WWRF vision to the requirements, limitations and challenges seen by the WWRF environment.

User Requirements, Scenarios and Business Models

35

ž Represent a dynamic and user-centred view of the wireless future for 2017 (this has since been changed to 2020). ž Represent challenges in future technologies, services and applications, and research areas of the WWRF. ž Serve as a common reference frame, as well as a reference for defining future WWRF activities. The development of the reference scenarios has been based on input from all WWRF Working Groups (WGs) and Special Interest Groups (SIGs) and has been coordinated and driven by the Working Group 1 on Human Perspective and Future Service Concepts. 3.2.2.2 The Approach Since the WWRF reflects many different research areas, it was decided to broaden the user-centric scenario concept to include short descriptions of expectations of capabilities, or even specifications of selected parameters of the envisioned system architecture in the future. This means that the scenarios should ideally represent all the WWRF in the relevant research areas and at the same time point to new areas of interests. Construction of user-centric scenarios demands a certain focus on real users and their requirements. To be interesting the situations and actions expressed in the scenarios have to be linked to real life situations and as ‘stylized facts’ be a representation of the environment relevant to the WWRF research areas. Therefore, principles from user-centric design [14] have been applied. For this particular exercise, it was decided to use a more futuristic template for the scenario construction. The template used was built on [15] as well as on elements and trends in [16]. In summary, the scenarios are constructed using the following approach: ž Identification of driving forces and fundamental drivers. ž Identification of relevant driving forces to take into account fundamental uncertainties. ž Selection of different scenario settings: locations, persons, focus and interaction with the wireless systems, devices and applications. ž Identification of ‘amazing’ or ‘wau’ effects for each scenario. ž Writing of the scenarios. ž Discussion of the impacts of the scenarios. Driving forces are elements which move the plot of the scenario and vary between the scenarios. Fundamental drivers are elements which have a reasonably high probability of coming true in all scenarios [15]. The identified points cover overall technology trends, social and user perspectives, environmental and economic perspectives in the Wireless World. Fundamentally, this approach means that the user-centric scenario writing will drive the process to identify system capabilities and specifications. However, an adaptation process between the two sides has to be introduced in order to ensure that technology visions (which may be different from users’ expectations) can also be displayed. The overall process is shown in Figure 3.2. The motivation for the coordinated scenario work is that it provides a structure for the exemplification of user needs related to the different technical aspects and serves as a guideline for future relevant research projects.

36

Technologies for the Wireless Future – Volume 3

Input from several information sources • • •

Survey of literature and state of the art Industry input Research projects

WWRF Vision

User scenario 1

User scenario 2

User scenario 3

•

•

•

High-level story line of a private sphere scenario

High level story line of a work sphere scenario

High level story line of a public sphere scenario

Visions of WGs and SIGs

Interface and integration of WG/SIG inputs

Detailed scenario 1

Detailed scenario 2

Detailed scenario 3

Figure 3.2 Overview of the scenario approach in developing the WWRF reference scenarios. Notice the feedback loop between the user scenarios and the visions of WGs and SIGs

3.2.2.3 The Actual Process In practice, the scenario construction process started in spring 2006. Members of the WWRF were asked to give feedback on a very loose set of scenario headlines identified by the WG1. Over the summer all WGs and SIGs were again asked to provide written feedback on a scenario draft in the form of a questionnaire relating to different capabilities and specifications of the anticipated future wireless systems. They were again related to immediate expectations of the requirements of the scenario sphere outlined in the draft. The drafts and questionnaires were then followed up by phone interviews/discussions with the WG/SIG leaders to focus on the parameters, specifications and trends that represented the forum the best. Comments and feedback were reflected in the first version of the reference scenarios (see for example [17]). Simultaneously with this interactive process, a desk study was conducted focusing on other scenario-making exercises carried out for the construction of wireless scenarios (see for example [16]). Reports and surveys relating to specific requirements in different countries were also used (for example [18]), and user-centred requirement studies were included [5]. The first version of the reference scenarios was presented at the WWRF meeting in Heidelberg, 2006. This caused many reflections and suggestions for changes over the months up

User Requirements, Scenarios and Business Models

37

to and around Christmas. During spring 2007 changes were made in a number of ways to develop the second user scenario version. Examples of changes were: ž ž ž ž

Alteration of the scenarios to reflect 2020 instead of 2017. Specification of the driving forces and their importance in the different scenarios. Removal of most references to direct technologies in the user scenarios. Various modifications and rewritings of the different scenarios in order to comply with comments.

In June 2007 it was decided to create a scenario group consisting of representatives from the WGs and SIGs. This was done in order to secure targeted feedback and contributions to the final scenarios and to be able to carry out discussions within a smaller group of people. This will lead to a stable version of the scenarios in the autumn of 2007. An evaluation process has also been planned. It is of high importance for the WWRF that the scenarios which represent the forum actually have value, validity and interest to the surrounding society. The evaluation process will take place during 2008 and the outcome will be the final scenarios. 3.2.2.4 The User Scenarios Even if the final version of the scenarios are not ready yet, a framework exists for the content and structure of the scenarios, as well as a set of user scenarios. The overall structure of the scenarios is based on: ž A combination of driving forces (elements which can be assessed at different levels in the various scenarios) as well as a number of fundamental drivers (which are equally represented in all scenarios). ž A narrative (with selected pictures) which describes a user/group of users and how they interact with wireless technologies in daily life situations. ž A set of specifications linked to wireless components of wireless systems such as bit rate, frequency bands, spectrum, security, network interconnections, mobility, terminals and devices, etc. The intention is to represent visionary elements in the user activities and uses of technology, but also in the representation of wireless components and specifications. Setting user centricity at the centre of the scenarios, it was at an early stage decided to represent three thematic areas: the residential, the professional business and the public/health-related. This was to capture the comprehensive user sphere as stylized facts and to address different uses of, for example, seamless networking, sensor networks, security, mobility, etc. For each of the thematic areas, two sub-scenarios have been constructed, focusing on different geographical areas; it was decided to ‘cover the world’ with the scenarios, highlighting the differences in technology adaptation this will show. The sub-scenarios cover (in terms of context and location): ž Home environments (activities related to what users are doing in their home and when carrying out activities related to being off work, such as shopping) in the USA and rural China.

38

Technologies for the Wireless Future – Volume 3

ž Nomadic professional working environments in Germany and Argentina. ž Public/health environments in Australia and Kenya. The scenarios do not include a particular stakeholder analysis. The intention has been to balance the representation of gender and different ages in the scenarios in order to represent different user groups. A relatively long list of driving forces and fundamental drivers has been identified for the scenarios. The full list can be seen in [17]. Examples of driving forces are: ž ž ž ž ž ž ž

developments will generally be more user driven user mobility will increase the service and application market will grow user security, integrity and privacy will become more important the market concentration in the wireless industry will change the fight for market dominance in the wireless industry will intensify short terminal usage time and complexity management will become increasingly important problems.

All six scenarios represented in this paper build on the assumption that users in 2020, in general, are happy with and open to new technologies. The reference scenarios aim at addressing values and capabilities central to the WWRF. Some examples are: ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž ž

high mobility of users usability issues devices and device capabilities ethic perspectives billing and payment issues seamless communication security personalization privacy and trust context awareness and management peer-to-peer communication authentication self actualization rich communication ambient technologies ubiquitous access pervasive services high bandwidth and high system loading capacity operator issues nomadic work global connectivity presence awareness control sensor networks thing-to-thing communication.

User Requirements, Scenarios and Business Models

39

3.2.2.5 The Scenario Settings and Examples The current version 2.0 of the user reference scenarios show the diversity of different users around the world, their requirements and needs for wireless technologies and how they interact with different devices and applications. Below is a short characterization of the scenario settings and two examples of the narrative parts. It has been realized that machine-to-machine communication cannot be visibly shown in the already formulated user scenarios. Since it is assumed that a high percentage of devices in the future will be machines, tags, sensors etc. the user will never see or know about, there is a need to develop a specific machine-to-machine scenario. This part of the scenario development is still ongoing. Home Environments – California In many cultures around the world, managing daily life at home is not necessarily only done at home. For many knowledge and business workers, daily life is often mixed with working life, without clean-cut boarders between them. This scenario focuses on some of the basic human requirements involved in managing residential matters. It looks at the need for monitoring and dealing with children to make sure that they are secure on their way home or in the home. It also deals with all the daily life issues that need to be taken care of, including grocery shopping, picking up children, cooking, communicating with friends/family and entertainment. The scenario is related to a suburb in California. It could be such a suburb as La Jolla, north of San Diego, where there is a large percentage of people who commute every day through, to and from the metropolitan area of San Diego. Because of the high density of people living and commuting, public mass transportation can be essential. In many American cities like San Diego, metros, trolleys and busses take care of public mass transportation. In San Diego proposals have been made to build monorails on top of this. The monorail idea is here taken a step forward and combined with an existing idea concerning combining individual wagons/cars with a sort of trolley or monorail, which can then be transported over specific transportation nets. The San Diego area (and perhaps most of California) is characterized by a high number of community groups which work to unify the voice of the community in different concerns and issues. The need for communication within a community is also a focus point here. Technologically, this scenario focuses on the handling of mesh-networks, establishment of secure and trusted network connections in public and private spheres, sensor networks in the private home, and e-commerce in push-pull relations. Business perspectives look into the construction of business models for individuals in the RUT (Rapid Urban Transport; an extended trolley/monorail), connecting to a hosting network and using services on this. There are also business perspectives related to ads and shopping, to setting up games and to establishing large networks for community communication. A short example of narrative in the homecoming scenario (California) follows: What a day! It had been one of the busiest days in a very long time. Juliet took a deep breath and pushed the button to lock her car to the RUT: an individual car which could be linked to a group of other cars and transported together via a specific monorail type of line running above all other traffic. Juliet was a 38-year-old woman working as a photographer in a big broadcasting company. News about the disaster in Mexico had filled the headlines all day and it was difficult to leave the company but she had to go home to take care of the children as her husband was away on a business trip. Fortunately, she could follow the development of the situation while she was linked

40

Technologies for the Wireless Future – Volume 3

to the RUT, looking at the built-in screen. Using her personal mobile device, she could connect to the network available in the RUT and transfer her personal preferences in terms of what she would like to watch on television or listen to on the radio, or whether she would like to use the built-in display to retrieve personal information such as e-mail, voice mail and maps showing the location of her children. Furthermore, she was able to continue working on her personal laptop in order to finish off some work while traveling. Juliet liked to use some of the advanced services on the RUT network. One of the specific services was that she could get information about other commuters on the RUT. Sometimes she discovered friends who were commuting at the same time; since the commuting time was sometimes considered ‘wasted time’ it was a good chance to have a ‘friends’ talk before she arrives at home to all the responsibilities there. On the other hand, she could also select not to be visible to others if she did not want to be disturbed on the ride. This feature was turned on today. She was too preoccupied with watching the news on the disaster and spent the trip following the television news. There was a fee for using the RUT technological network and the special services. However, Juliet thought that the price was fine, particularly as she had to spend almost two hours commuting every day.

Home Environments – China Technology adaptation in rural areas of Asia is rising exponentially and is expected to continue over the coming years [3]. In China today, more than half of the population lives in rural areas. China has a penetration rate of mobile phones of 33 %, and this is expected to rise over the years and therefore to have a significant effect on rural areas. This scenario looks at rural China and expected and likely achievements in this area. Focus is on a young adult who goes shopping. Different services are being put forward for the user, but for a relatively low cost or no cost. It is expected that most services offered in rural China will be attainable for everyone. The scenario explores the broad usage of mobile smart phones and RFID tags combined with search technology. Security plays a small role but is not limiting peoples’ usage of the mobile smart phones. Networks and hotspots are expected to be present at most locations; however, there may be some problems with stability and interruptions of communication. Business Activities – Frankfurt This scenario focuses on a high-density professional business environment: Frankfurt. It can be assumed that for this scenario, Svend (the main character) represents a large group of users requesting the same sort of capabilities and resources. Frankfurt is already a metropolis in terms of international business, with around 320 000 companies situated in the city. The region is considered to be one of the most productive in Europe and is an international financial centre. It is expected that the need for office space in Frankfurt will continue for years to come and therefore, indirectly, technology support for a large group of users will have to be developed. Svend, a media journalist, is representative of a large number of knowledge and business workers around the world. The knowledge worker/journalist is characterized by being highly mobile and with high demands on bandwidth, seamless connectivity and security. The scenario assumes that Svend’s businesses can be carried out seamlessly with the use of a cellular set-up in Frankfurt – this being one of the wireless technologies in the area. This illustrates issues and challenges particularly in respect of spectrum capacity and cognitive radio usage.

User Requirements, Scenarios and Business Models

41

As a professional business person, Svend does not consider payment for the advanced services to be an issue. However, there are business perspectives to explore when it comes to the sharing and engagement of operators when capacity needs are not met with the work office’s normal network capacity. Business Activities – Argentina Argentina has an old tradition of exporting goods to (amongst others) Europe: many Argentineans originates from European countries and have in that way paved the way for performing business today. The export industry centres around food, meat and manufacturing products [5]. This scenario takes place around the large city of Mendoza in the middle of the wine district of Argentina. Argentina is a worldwide number five in wine production [5]. Wine serves as a source of income in different ways: as an agricultural product to be sold in Argentina and exported to other countries; and as a tourist attraction, with possibilities to stay on wineries for some time. This scenario takes place on a large winery with exports to Europe and a growing tourism. Looking at a winery in Argentina in 2020 gives several challenges technologically: a large winery with tourism can be assumed to be open to adaptation and implementation of wireless technologies. On the other hand, will the location of such a winery give limitations to the types of technology which can be implemented; to the number of operators which can be assumed to operate in that location; and perhaps to the quality of service, electricity availability and the robustness of network connections? The export perspective for such a winery will, again technologically, demand a certain electronic handling of communication, billing and payment to make a smooth business with European partners. This will again demand a certain level of security in spite of, perhaps, problems with trusting network connections and shortages in electricity. Today, radio technology is important in Argentina. This is expected to still be the case in 2020, at least for older people, and as a back-up if Internet connections are unstable. Technologically, this scenario focuses on radio technology, RFID tagging, information systems for export business, simple wireless communication over WiMAX or similar technologies, security and problems with display and battery time. An example from the narrative part of the scenario in Argentina: After welcoming the tourist group, Pedro Salazar headed for his car. It was a busy day and he and his men would be checking a large amount of the wine stocks and had to supervise bottling of the first barrel of the organic wine. Organic wine was a new product from the Salazar winery. However, it was now a clear demand from the Dutch customers they exported to: organic wines had been upcoming for years and a large percentage of their customers wouldn’t buy anything but. Pedro was not so sure about the quality but he knew that the wine would be documented in the right way to satisfy the demanding customers. Pedro stopped the car around 2 km from the hacienda. A group of his men were working in the field checking the wine stocks. Pedro looked at his mobile smart phone. There he could find the details on the wine stocks they were checking right now. One of his men had been making adjustments to the fertilizing of some of the new grape stocks all morning, and had tried to update the tags on the stocks next to the plants. However, his mobile device failed (the display couldn’t manage the heat or the batteries failed in humid weather – problems they unfortunately often encountered) and therefore Pedro now had to make the changes to the stock tags and at the same time update his own mobile smart phone. The information was then updated on Pedro’s mobile smart phone so that he could later place it on the ERP system on his laptop so that it would be

42

Technologies for the Wireless Future – Volume 3

visible for the export partners in Europe as well. This was all part of the increased documentation procedure for wine production, combined with the increasing environmentalism in the world.

Public Services – Kenya Worldwide education is key to economic and social development. In many low-income countries the enrolment rate of children is a major concern for the governments and is a factor which politically they try to turn in the right direction using different steering parameters. In these years, several initiatives have been created in order to produce and offer cheap laptops so that all students in low-income countries will have the ability (probably with the help of the local government or school) to pursue such devices and thus ease the way to better learning and understanding of the technology and Internet. Kenya is a low-income country where a relatively high percentage (around 85 %) of children go to school [6] and therefore political focus is on the education perspective. This scenario focuses on e-learning perspectives in Kenya. In the scenario, the idea of offering relatively cheap laptops is used as a factor to enhance higher education and to offer distance learning. The scenario focuses mainly on the features of Internet applications and browsers in 2020: context awareness, privatization and trust issues are here relatively important. Cheap computers can today work as a basis for creating ad hoc networks/mesh networks, and this perspective is underlined in the scenario as well, since it is expected to be a feature of technologies in the future in more rural areas of low-income countries. Public Services – Australia Health care and management of health-related concerns is a worldwide issue which is expected to increase in importance over the coming 10–20 years: the number of deaths caused by chronic diseases is estimated to reach 41 million persons worldwide by 2015 – a significant rise compared with numbers from 2005. This naturally will be a large problem economically (it will create a loss in terms of income and expenditure in terms of treatment) and some challenges in treatment of people across borders (since it can be expected that mobility of people naturally will increase – chronic disease or not). One basic requirement for the future (and in the present) will be the ability to receive the right level of basic treatment for all users everywhere. This scenario focuses on an older couple that travels and at the same time needs supervision of the functioning of a pacemaker. There is an increased need for monitoring of diseases and treatments among older persons compared with younger adults (who often manage their own diseases and treatments as far as they can). The couple in the scenario is from India, traveling to Australia. The couple must be assumed to represent the upper layers of society and therefore also the upper level of health care treatment. Sydney, Australia is already today a well-developed city with a high level of care-taking and hospital equipment. In the scenario, location-based services for which payment is required are specifically mentioned. In the contact between the hospital and the patient, payment for the treatment is also mentioned, however no details are given on how that could take place. It will most likely be that the hospital service needs some kind of business model, either as a direct contact with the patient or an indirect payment between the hospitals (so that the patient has a pre-specified rate for all types of treatment in terms of the pacemaker with his home hospital, and this rate is unaffected by the place at which he is treated).

User Requirements, Scenarios and Business Models

43

Technologically, the scenario will include special consideration of and requirements for the handling of security perspectives, machine-to-machine communication, BANs, interference between airwaves, medical equipment and the pacemaker, as well as location-based services. In all the scenarios, a number of pictures have been included for illustrative purposes. In the final version, it is the intention to have specifically drawn illustrations associated with the scenarios. The full description of the user scenarios version 2.0, as well as assumptions for these, can be found in [19]. 3.2.2.6 Short Reflection The process for the reference scenario development has been relatively long. With respect to the process developed for this particular purpose, the challenges and problems have mainly been in three areas: 1) to create the link between the user scenarios and the identification of the capabilities of the future wireless systems; 2) to create user scenarios that are sufficiently visionary to challenge the technology and sufficiently realistic to create recognizable social settings; 3) cross-disciplinary exchanges take time. However, development of cross-disciplinary discussions is one of the fundamental purposes of the WWRF and these exchanges have therefore served more than one purpose.

3.3 User Requirements Using a participatory design approach to design and develop technologies demands a specific adaptation for the users involved in the process. In relation to wireless technologies the concept of nomadicity is essential. The following section presents one approach to identifying user requirements for nomadic workers. The approach has been developed as part of the MAGNET Beyond project [5]. 3.3.1 Background and Needs The MAGNET Beyond project is an extension of the MAGNET project aiming at implementing the solutions developed during the first phase. MAGNET was a purely R&D project while the MAGNET Beyond project has its focus on development and implementation with the aim of demonstrating achievements through, amongst others, pilot services. Within the MAGNET Beyond project, it was decided to focus on two particular application areas: health care (MAGNET.care) and nomadic, professional business (Nomadic@work). The health care area focuses on how to manage health care in a more efficient and intelligent way, while the professional, nomadic business area focuses on supporting workers with a high demand on the capabilities of wireless systems in their daily life. For each of the two cases, different ways of identifying user requirements have been developed (see [20]); below the focus is on the Nomadic@work case. The essence of nomadic workers is that their normal situation is being on the move. Furthermore, they demand (or wish for) the full technology support available in their offices when traveling. Take for example a journalist from an advanced industrialized area who travels around the world to make reports on disasters. In such cases, it is likely that the wireless technologies he is used to at home will not be available to the same extent abroad and a lot of different short cuts or solutions will have to be made in order to file a reasonable report (audio

44

Technologies for the Wireless Future – Volume 3

and video) to be shown on the home area’s television. His requirements are, by the nature of his work, rather high, and therefore he represents a demanding tech user. When it comes to identifying user requirements from mobile, professional users, an obvious solution would be to follow persons for a period of time and register their requirements. However, as with the example from the MAGNET project presented above, the process here would be to explore requirements relating to the future developments of the personal networks and the opportunities they create. This would require following groups of mobile, professional users and interrupting them constantly with questions leading to the development of technologies addressing the problems and needs arising during situations encountered during the day. Within the MAGNET Beyond project this approach was not considered as an option. Therefore a new approach was developed in order to deal with the nomadicity of the users and still get insight into requirements related to future achievements in technology. 3.3.2 Mobile Probing Kit The overall approach to dealing with the nomadicity issue in MAGNET Beyond was to design a so-called probing kit, with inspiration from [21], [22]. The idea was to let a mobile probing kit follow a group of nomadic knowledge users in order to identify their needs, ideas and requirements on problems encountered in their everyday working situations. The inspiration for the probing kit came from the designer Jeff Hawkins, who used a carved-up piece of wood (the size of a shirt pocket) in the very early stage of a design process [23], [24]. The carved piece of wood was used as an actual device that he brought along in his activities. Furthermore, he simulated entering information on the mock-up and in that way envisioned how he would use this ‘device’ if it really existed. However, it is one thing to gather this kind of information oneself, and another to collect the user requirements of a group of users. For that particular purpose, a probing kit was developed. The probing kit was a 5 cm by 7 cm notebook with a nice metal cover and an integrated pen. The small size of the notebook and the nice design were chosen intentionally to make it appealing for users to ‘wear’ the notebook for a period of time. Under the metal cover was placed a brief explanation page describing what the participants should do. Furthermore, small bumper stickers were placed on selected pages in the notebook. The bumper stickers should motivate the participants to think about possible problems and solutions. The sheets in the notebook were used to note ideas, problems and issues of interest when the participant had time or was in a situation with a problem or a specific wish. Figure 3.3 shows a picture of the mock-up. 3.3.3 The Approach The implementation of the probing kit took place over a longer period of time and was kicked off with a workshop [24]. The overall approach can be summarized in the following steps: ž A workshop started the implementation. At the workshop a group of nomadic workers were first introduced to the concept of a personal network and to the task they had volunteered for. ž At the workshop, the participants were introduced to a stimulation creativity tool named Visionpool [25]. Visionpool consists of a large number of small slides (5 cm by 5 cm)

User Requirements, Scenarios and Business Models

45

Figure 3.3 The probing kit used in the MAGNET Beyond project for identifying user requirements for nomadic users. The probe was called IDEA MAGNET. (Reproduced with permission from MAGNET)

Figure 3.4 Slides from the Visionpool, used to motivate the users to be creative while wearing the mobile probing kit

showing diverse colours, ambiguous elements and figures and forms, and parts of welldefined, well-known things. The participants were asked to select five of these slides, which they then used to develop a short story or an idea related to a technology, a problem or an idea on how new technology would support them in their daily life. The overall idea of the introduction of the Visionpool was to provide motivation for the participants to be creative when later walking around with the probing kit themselves. Figure 3.4 shows some of the slides from the Visionpool part of the workshop.

46

Technologies for the Wireless Future – Volume 3

ž After kick-off and introduction at the workshop, the participants were asked to carry the probing kit around for a total of three weeks and to note every idea, problem or wish they encountered when working or carrying out activities related to work. In order to make sure that the notebook was not forgotten, the participants received a number of text messages (with their agreement) that should remind them of their task. A text message could be: ‘what information do you need right now?’ (more can be found in [24]). ž After three weeks alone with the probing kit, the participants returned to another workshop. At this workshop, the noted ideas were presented in plenum, causing a new discussion (and new ideas). Finally, an evaluation of the whole process was made. 3.3.4 Short Reflection A total of 175 ideas were made over the three weeks, of which half were relevant to the MAGNET Beyond project. That was considered satisfactory in relation to the number of participants (11) and the fact that many of the participants actually lost the motivation for wearing the probing kit and making notes over the whole period. Typically, the participants were rather motivated at the beginning of the three weeks, then forgot (or almost forgot) about it, and then in the end found some motivation again. Some of the participants liked the text messages they received, while others became annoyed and a bit frustrated because of being forgetful or not wanting to make notes. Conclusions on the approach were that it is useful but that the time period that the participants have to walk around with the probing kit should be significantly decreased. More discussions on the approach can be found in [24].

3.4 A User-centred Approach to Service Development As discussed in Chapter 4, the development of technology and especially the associated convergence of technologies enables a change in the roles of users in relation to the creation of new services. The changing technologies have led both to a new network structure and to availability of new tools empowering the user. The networks were originally owned and operated by the operators: telegraph messages were typed and decoded directly by employees of the telegraph company. With the emergence of voice and later data communication, users could send and receive messages themselves, but the equipment comprising the network was owned and operated by the telephone company, limiting messages to fixed formats and in principle even limiting the content. This was drastically changed during the last two decades of the last century by the liberalization associated with the new technology: equipment owned directly by the user is increasingly used to format and transmit messages. Spectacular examples of users taking the new opportunity to create their own services are related to the use of SMS, general use of the Internet and more specific use of P2P networks (e.g. Skype, Flickr and YouTube). Common among these heterogeneous examples is that they enable end users to participate in an active role using, generating and even distributing content and additional services either in a peer-to-peer manner or through portals or other hosted services in the network. The trend towards making the users of a service participate in its creation and provision is sometimes called ‘McDonaldization’, after the fast food chain that invites its customers to serve themselves and clear their table after they have eaten [26]. The expression also hints that the trend is towards a lower level of sophistication/quality than professional services. Whether this is a feasible and acceptable long-term trend is difficult or impossible to answer in general. A basic criterion for acceptability is that the content creation mechanisms are easy

User Requirements, Scenarios and Business Models

47

and intuitive ways to create services for all environments broadly used. For the specific service creation usability perceived by the users – as discussed below for commercial services – is the ultimate test both for commercial and noncommercial services. More information on service creation can be found in Chapter 4.

3.5 Usability Usability and user experience terms are highly important to designing commercially interesting products. When developing applications or services it is important to make user-friendly and user-acceptable solutions and this has its direct and immediate expression in the interface between technologies and users. As part of this, the testing and evaluation of early interface prototyping is necessary. This section reports on a case in which an interface design has been developed and tested in situ in a low-fi mockup. The test and the approach to testing the first prototype design were developed within the MAGNET Beyond project [5]. 3.5.1 Low-fi Prototyping Dealing with mobile users in the MAGNET Beyond project, it was a challenge to design the usability testing of the low-fi prototype design with real users. It has been the underlying principle within the project to make developments from a participatory design view, and therefore a special approach was developed to test an early prototype interface for mobile users. The overall aim was to develop a conceptual user interface that could be used by nonsavvy technology users. Therefore a first analysis was made to identify the characteristics and needs of a nonsavvy technology user. Important user qualities for such a person could be [20]: ž ž ž ž

He/she feels that he/she is in control of every step/action. He/she can predict the outcome of the action. He/she is able to overview the system (aerial view). He/she gets a conceptual understanding of the system and the actions (transparency). Introducing him/her to MAGNET Beyond, the sequential goals will be to:

ž Make him/her trust the PN and let him/her be in control of it. ž Let him/her establish and use PN-fed concerning documents with trusted well-known persons (probably in a face-to-face situation and with privacy management). ž Let him/her start trusting and using selected, well-known services, offered by providers he/she knows and trusts. Because the nomadicity concept is so relevant, it was decided to conduct the first evaluations in a situation with the users placed in an environment in which they would be present anyway, meaning they are potential real users [20]. In order to carry out relevant tests, a low-fi paper prototype was developed. This can be seen in Figure 3.5. The paper prototype included a Nokia 770 as an underlying device (any other device could have been used, but the application and pilot services developed in the project were initially targeted to this platform/device). The device would at all times be closed. On top of the screen, a number of paper ‘screens’ were placed. In order to navigate around on the prototype, the test person would have to flip over the pages and in that way imagine that the selected buttons on the paper screen had a consequence and brought them further in the navigation of the device.

48

Technologies for the Wireless Future – Volume 3

Figure 3.5 Low-fi paper prototype developed in the MAGNET Beyond project [20]. (Reproduced with permission from MAGNET)

On the side of the paper screens were placed a number of small stickers with headlines listing activities which could be carried out using the paper prototype. This was done in order to make it a bit easy to navigate through the 52-page-long paper prototype. Everything was tied to the Nokia device with a string (see Figure 3.6). The paper prototype was used in order to achieve an understanding of users’ experiences of the envisioned services and generic MAGNET Beyond concepts and to evaluate whether the concepts envisioned were as good as possible before an actual implementation. 3.5.2 The Actual Test In a field type of evaluation, the physical context and uncontrollable events occurring as part of the evaluation test can promote a better understanding of the use context for the envisioned concepts. This demands an understanding of both cognitive constraints, such as visual, light and sound, and spatial conditions in the use (test) situation. The social aspects are embedded in the physical context and also have to be addressed in the analysis.

Figure 3.6 The back of the prototype with the string holding everything together. (Reproduced with permission from MAGNET)

User Requirements, Scenarios and Business Models

49

The set-up of tasks which the users have to perform during the test was explicitly chosen beforehand in accordance with a cognitive walkthrough [27]. Other tasks which emerge during the evaluation as everyday users’ activities are handled, for instance, in SPES (Situated and Participative Enactment of Scenarios) [28]. In order to carry out the situated environment set-up, evaluations were carried out at a conference held in March 2007. The theme of the conference was the TETRA system (Terrestrial Trunked Radio, a European-developed set of standards for an emergency communication system) and its economic, regulatory and technological aspects of implementation. Present at the conference were representatives from a large number of companies with interests in TETRA systems, as well as academic representatives. The conference was assumed to provide a good basis for a situated environment set-up since a large number of different knowledge persons would participate and would want to carry out specific activities during the day. The evaluations (and the screens on the paper prototype) were set up to follow the activities of the day at the conference (registering, attending sessions, lunch, meeting people and exchanging information, transport when leaving the conference). The test for the dialogical approach was carried out in the following steps: ž The evaluation was planned. Specifically, it was planned what to observe, which actions to perform with the test person and what the evaluator should say. It was necessary to use nonemotional words when asking test persons to elaborate on their ‘think aloud’ comments so as not to influence the answers given. ž One evaluator was devoted to one test person. The evaluator had to introduce the test person to the test, take notes and document what was taking place in respect to the conceptual design, use and surrounding environment (expected as well as unexpected actions). The test person’s attitude or emotional outbursts were also noted. ž A number of ordinary conference participants were (by agreement) followed by an evaluator and asked to carry out different specific tasks such as registering for the conference, looking up information on speakers, etc. – doing everything using the low-fi device with the test screens. A list of activities and questions was predefined and presented stepwise to the test person (details on questions and activities in the given example can be found in [20]). ž It was an underlying assumption that the test person was a regular participant at the conference. This also implied that initial questions and the introduction to the test evaluations would have to take place before a conference session. When the session started, the evaluator followed the test person and sat beside them during the session. During a regular conference session the test person did not have a specific evaluation task to carry out but was encouraged to look through the paper screens in order to be a bit more confident with the set-up. ž During the session, the evaluator again made notes regarding the test person and how they were interacting with the paper prototype. ž After the conference session, the test person was asked to carry out some more activities relating to the conference, and again problems, questions and observations weredocumented. This was the end of the test, and a new test person could be followed. The length of the evaluations was around 2–2.5 hours in total. Other types of evaluation test were carried out during the conference. Details on these can be found in [20].

50

Technologies for the Wireless Future – Volume 3

3.5.3 Short Reflection With respect to the test, it was concluded that the paper prototype needed a significant redesign: many problems, questions and issues were raised as the test persons went along with the tests. Since the aim was to develop a user interface design for nonsavvy technology users, too many concepts and links in the prototype were difficult. In terms of the approach, there was a general satisfaction with the way things tested. Most users were pleased that an actual Nokia was used, giving them the feeling of a real device (what it was). However, it was also clear that it was difficult for the users to get an overview of the many paper screens and to understand that not all the buttons in the graphical interface could be ‘active’ and lead to a new paper screen. Details on the whole process, concepts tested and results can be found in [20].

3.6 Business Modeling The trend towards user participation in the creation of content is having an important impact on value creation and business models. It challenges the way business models are constructed and used. Business models have in a sense existed as long as there have been business interactions between people and businesses. One of the first business models in a modern sense was the ‘shopkeeper’, but businesses have always organized their activities in accordance with ‘business models’, whether shaped by the tradition of the specific trade or out of theoretically-backed considerations. The meaning of the concept is, nevertheless, not quite clear. In academic literature the first appearance of the term ‘business models’ was in 1957 [29]. It was, however, not widely used until the emergence of the Internet, but now a general understanding has emerged that ‘The term business model describes a broad range of informal and formal models that are used by enterprises to represent various aspects of business, such as operational processes, organizational structures’. Thus, a business model typically describes a company’s business with respect to the following building blocks: production capacity, supply, customers and finances; i.e. the business model takes the technological characteristics and potentials as input and transforms them via customers and markets into economic values. From a theoretically economic point of view, it is maybe not too clear what this concept adds to the understanding of the economic process, but the strength of business models is that they consider how new technologies are taken into market – a process traditionally neglected by economic theory – and they even offer a structure to analyse the consequences of the development of different technological ideas. When business models were introduced the interest of market analysts and researchers centred on categorizations of different kinds of business model in order to understand the new possibilities related to the Internet in the form of, for example, e-commerce and brokerage. Later, the main emphasis was on dividing up and reconfiguring the different business processes and market players that are parts of value complexes or networks in order to understand the development in supply chains. A business model in the current general understanding relates to many different issues, such as the value proposition to different market segments, the players involved in the processes of production, the revenue model and the distribution of the revenue among the different companies involved, the flows of information among the business model entities, etc. This generally provides a framework for understanding how the technical system is reflected in the economic system transmitted via the market, as illustrated in Figure 3.7.

User Requirements, Scenarios and Business Models

The economic system

Flow of money

Flow of goods and services

The technical system

51

Suppliers

Figure 3.7

The business model seen as a reflection of the technical system in the economic system

This also illustrates, however, that generally business modeling is a supply-side exercise dealing with the relationships between the players on the supply side in order to determine how they can service the needs on the demand side. There is obviously a need to introduce user needs/requirements in the modeling exercise. A way of doing this is illustrated in Figure 3.8. Figure 3.8 illustrates how the demand side is introduced in the MAGNET methodology. An overall ‘distillation’ and synthesis process is introduced ‘between’ the technical and economic systems. In practice this is done following the procedure described above. User workshops are organized in order to gather user inputs to develop and qualify the specific services, which are then assumed to cover the needs and demands in the different situations, which are generalized as ‘themes’. The outcome is user scenarios that are evaluated technically in expert workshops, resulting in user requirements described in technical terms, and these are subsequently used to define the system requirements. Throughout the process, technical potentials and constraints and business aspects provide input to shape, check and complement the user requirements as illustrated in Figure 3.8. The technical aspects outline the sphere of possible services, whereas the business aspects characterize the economically viable services. The final results of MAGNET are prototypes and pilot services. Some user requirements cannot be directly transformed to system requirements, but must be fed directly into the operational system (illustrated by the dotted line in Figure 3.7). 3.6.1 New Trends in Business Models Figure 3.7 illustrates that a change in, for example, the technical supply system will lead to a change in business models. Following the general liberalization of the telecom markets there have been major changes in the supply system. First the dominating role of the incumbent was challenged, but the changes are ongoing, for example affecting the traditional role of mobile

52

Technologies for the Wireless Future – Volume 3

Themes

User workshops

Expert workshop

Business aspects

Technical aspects

User scenerios

User requirements

System requirements

System prototype

Operational system

Figure 3.8 Flowchart showing the process from collection of user inputs to an operational system

operators. Open systems for a wide variety of access types will invite new players to enter the market. Focus will shift from competition on geographical coverage and price to competition on services. Competition will be heavily increased on nearly every network segment from the radio interface, access supplier and service and content provisioning. A second significant example is the role of equipment manufacturing. The current mobile device has different types of access, and a wide variety of applications and services as well as a hyped and interesting design, all in one small, user-friendly, complete terminal with a long-life battery. Device manufacturers are moving away from smart phone technology into mobile devices with lots of new functionality. This is associated with a development where R&D has virtually disappeared as a telecom operator activity, generally shifting to equipment producers as Motorola, Nokia, Samsung, etc., leading to an increasingly prominent role for these actors in the development of telecom and more broadly ICT. This is exemplified by their dominating roles in new converged standards such as 3G and mobile TV. The first

User Requirements, Scenarios and Business Models

53

modification of the traditional mobile business model was equipment producers meeting user requirements for advanced/sophisticated handsets outside the operator arrangement. The next step was operators using advanced handsets as competitive arguments in their communication packages. Recently Nokia has taken this development a step further by not only offering connections to preinstalled services, but also offering directly a communication service, GPS and selling enhancements. This is in direct competition with operator offerings, but no doubt prompted by a user demand for specifically GPS services.

3.7 Conclusions and Further Research 3.7.1 User Requirements, Scenarios and Usability Many user-centred design methods have been developed around wired technologies, but new technologies and changes in behaviour, e.g. the introduction of the nomadicity concept as a reflection of an important development, increase the need for revision of existing methods and for developing new methods. This also relates to the identification of user requirements and usability testing of products, devices and services. The discussion in this chapter is mainly related to the process of identifying and implementing revised and new methods. A number of issues needing further reflection and research have been identified: ž Construction of scenarios that are seen as relevant and challenging in both user and expert workshops. The problems are mainly related to: ž Creating the link between the user scenarios and the identification of the capabilities of the future wireless systems. ž Creating user scenarios that are sufficiently visionary to challenge the technology and sufficiently realistic to create recognizable social settings. ž The user-centred design activity is cross disciplinary by nature and there is a need for a (better) common language. ž There are limits to the time period in which a given group of users can effectively be engaged in participatory design activities; this leads to a need for careful planning of sequences and topics. ž Usability tests demand a careful balance of realism and complexity. ž Identifying user requirements for mobile, wireless technologies or services is a challenge. It calls for changing the currently existing methods or even designing new methods of involving mobile users in the user requirement identification. New developments in mobile, wireless technologies call for an evaluation of the participatory design process in which user requirements and usability tests are carried out traditionally. This chapter illustrates some of the trends and new ways of thinking aiming at improved user input for better future technologies and services. 3.7.2 Business Models Traditionally, close cooperation between the device manufacturer and the mobile operator have created a well-established value chain in the provision of mobile services to users. New services and devices were introduced by the network operator. This cooperation is under stress. Device manufacturers today spend a large amount of their profits on research and

54

Technologies for the Wireless Future – Volume 3

development of new technologies and services, and market them directly to users. The future shape of business models will depend on the interrelated questions: ž Which services will users demand? ž Which technology systems will be developed? ž Which combination of organization and technology will be most profitable? As illustrated above, there is no given/fixed answer to these questions. They may, however, be addressed, and insight be gained in an ongoing cross-disciplinary process. The basic research question related to business modeling is: ž Which actors will drive the development and how will relations be between the actors?

3.8 Acknowledgements The following contributed to this chapter (directly as well as indirectly): The MAGNET and MAGNET Beyond projects – in particular, input for 3.6 from Anders Henten and Rune Roswall is acknowledged; co-authors on projects and publications presenting the examples given in this chapter; anonymous persons in workshops, interviews and tests.

References [1] L. Kleinrock, “Nomadic computing, information network and data communication”, Presented at the international conference on Information Network and Data Communication, June, pp. 223– 233, Trondheim, Norway, 1996a. [2] L. Kleinrock, “Nomadicity: anytime, anywhere in a disconnected world”, Mobile Networks and Applications, 1(4), pp. 351– 357, 1996b. [3] N. Schultz, L. Sørensen and D. Saugstrup, “Participatory design and creativity in development of information and communication technologies”, in S.B. Heilesen and S.S. Jensen, “Designing for Networked Communications Strategies and Development”, pp. 75–96, Idea Group Publishing, Covent Garden, 2007. [4] MAGNET, http://www.ist-magnet.org, 2004. [5] MAGNET Beyond, http://ist-magnet.org, 2006. [6] A. Sasse (ed.), “User requirements and expectations”, in R. Tafazolli (ed.), “Technologies for the Wireless Future Volume 2: Wireless World Research Forum (WWRF)”, pp. 15–58, John Wiley & Sons, Ltd, 2006. [7] P. Ehn, “Work-oriented design of computer artifacts”, Stockholm: Arbeitslivcentrum. [8] S. Bødker, J. Greenbaum, and M. Kyng, “Setting the stage for design as action”, in J. Greenbaum and M. Kyng (eds), “Design at work: Cooperative design of computer systems”, pp. 139–154, Lawrence Erlbaum Associates, Hillsdale, NJ, 1991. [9] M.J. M¨uller, “PICTIVE – an exploration in participatory design”, Presented at the Conference on Human Factors in Computing Systems. “Proceedings of the SIGCHI conference on Human Factors in Computing Systems”, ACM Press, 1991. [10] A. Osborn, “Applied imagination”. New York: Schreibners, 1953. [11] K. Geoff, “Everyday creativity”, Little Ox Books, Stillwater, 1998. [12] R.V.V. Vidal, “Creativity and problem solving” (No. 2002– 3), Kgs. Lyngby: Technical University of Denmark, 1993. [13] R. Tafazolli (ed.), “Technologies for the Wireless Future Volume 2: Wireless World Research Forum (WWRF)”, John Wiley & Sons, Ltd, 2006. [14] K. Chrisler (ed.), “A user-centred approach to the wireless world”, in R. Tafazolli (ed.), “Technologies for the Wireless Future: Wireless World Research Forum (WWRF)”, John Wiley & Sons, Ltd, 2005. [15] P. Schwartz, “The art of the long view”, Currency Doubleday, New York, 1991.

User Requirements, Scenarios and Business Models

55

[16] B. Karlson, A. Bria, J. Lind, P. L¨onnqvist and C. Norlin, “Wireless foresight scenarios of the mobile world in 2015”, John Wiley & Sons, Ltd., 2003. [17] L. Sørensen and K.E. Skouby “Summary of the WWRF Reference Scenarios version 1”, 2006 (the note can be obtained by contacting the authors). [18] MOCCA project, http://MOCCA.projectweb.org. [19] L. Sørensen and K.E. Skouby, “WWRF Reference Scenarios, version 2.0”, 2007 (the note can be obtained by contacting the authors). [20] J.S. Pedersen et al. “Usability of PN services (low-fi prototyping)”, MAGNET Beyond Deliverable D1.4.1, 2007. [21] B. Garver, T. Dunne and E. Pacenti, “Cultural probes”, Interactions, Jan. & Feb. 1999. [22] S. Hulkko, T. Mattelm¨aki, K. Virtanen and T. Keinonen, “Mobile probes”, in “Proceedings of the Third Nordic Conference on Human-Computer Interaction”, Tampere, Finland, 23–27 October 2004, NordiCHI, 82 pp. 43–51, ACM Press, New York, 2004. [23] E. Bergman, “Information Appliances and Beyond”, Chapter 4, Morgan Kaufmann Publishers, 2000. [24] J.E. Larsen, M. Proschowsky, D. Saugstrup, N. Schultz, L. Sørensen et al., “Mobile probing kit: a method for obtaining novel ideas and requirements for mobile personal applications and services”, Submitted for International Journal of Human-computer Studies, 2007. [25] J.K. Sørensen, “The development of a visual design tool: VisionPool”, in “Proceedings of Nordic Design Research Conference, May 29–31”, Copenhagen, Denmark, http://www.tii.se/reform/inthemakting/ files/p13.pdf, 2005. [26] ITU, digital.life, ITU, Geneva, p. 71. [27] C. Wharton, J. Rieman, C. Lewis, and P. Polson, “The cognitive walkthrough method: a practitioner’s guide”, in J. Nielsen and R. Mack (eds) “Usability Inspection Methods”, pp. 105– 140, John Wiley & Sons, Ltd, 1994. [28] G. Iacucci, K. Kuutti and M. Ranta, “On the move with a Magic Thing: role playing in the concept design of mobile services and devices”, in DIS’00, Brooklyn, New York, 2000. [29] R. Bellman, C. Clark et al. “On the construction of a multi-stage, multi-person business game”, Operations Research, pp. 469– 503, 1957.

4 Service Infrastructures Edited by Prof. Dr. Klaus David (ComTec, University of Kassel, Germany) and Dr. Mika Klemettinen (Nokia, Finland)

4.1 Introduction Service infrastructures, advanced applications and service architectures are among the most important areas of wireless communications due to the recent developments in the markets. This is obviously related to the fact that several mobile markets, such as the markets in Western Europe, are reaching saturation in terms of number of customers for voice, the ‘killer service’ or ‘driver’ for mobile systems such as GSM so far. To facilitate further growth, the industry is currently making quite an effort to provide ever higher bit rates, as can be seen from the evolution plan of HSDPA, with 3.6 Mbps already being deployed and 7.2 Mbps just about to be deployed around 2008, and with the ongoing planning and R&D work for even higher bit rates. In addition, as described in this book, several other radio technologies like improved WLAN technologies, high-speed short-range technologies, etc. with even higher bit rates are being researched and developed to reach the market. This means that we will have a situation where a variety of wireless network technologies will allow for broadband connectivity in most places where customers happen to be. This will allow for similar services to those offered in the wired networks, e.g. web browsing, e-mail and VoIP, to name a few. For these types of service a fierce price battle can already be observed today and it can be expected that this will continue in the future. One key question is whether there are other services – not necessarily available and sensible in the fixed networks, not necessarily directly related to bit rate – that would excite customers and provide real value to them? Even given the fact that this area has hardly been exploited in the marketplace up to 2007, according to the WWRF vision these services will become a reality; see for example the related scenarios in Chapter 3. Other questions emerging from the previous one include: What kind of service infrastructure and enabling technologies would such services require? To what extent can one utilize available solutions familiar from the Internet and PC world, and what are the special Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

58

Technologies for the Wireless Future – Volume 3

opportunities and challenges related to wireless communications, where services, devices and people are on the move? How can one enable easy service creation in this setting that will lead to positive user experiences? The work in this chapter tries to tackle the challenges listed above from different angles and is based on three recent Working Group 2 white papers: ž ‘Semantic Services’ edited by Josef Noll from the University Graduate Center of Kjeller (UNIK), Norway and Matthias Wagner from NTT DOCOMO Euro-labs, Germany: ž In the heterogeneous wireless communications and services environment, machine understandable descriptions of both user context, preferences and service capabilities are the key for an automated service adaptation. Semantic technologies support the machine readability of content, and become part of the service-oriented architecture (SOA). This adaptability provides means for improved and personalized user experience especially in situations where the usage context limits the user if compared to the more traditional PC experience. ž ‘Service Creation’ edited by Olaf Dr¨ogehorn from ComTec, University of Kassel, Germany and Henning Olesen from the Center for Information and Communication Technologies (CICT), Technical University of Denmark (DTU), Denmark: ž As the technologies are converging due to trends like FMC (fixed-mobile convergence), a more unified approach for creating new services is desired both for the fixed PC-driven world and for the mobile world. Furthermore, the use of mobile devices enables end users to participate in a passive or even active role to use, generate and even distribute content and additional services either in a peer-to-peer manner or through portals or other hosted services in the network. This trend can be observed with services like Flickr and YouTube, and the upcoming blogging systems. ž ‘Service Architecture’ edited by Christophe Cordier from France Telecom, France and Josip Zoric from Telenor, Norway: ž Today’s wireless and mobile services are typically monolithic and centralized in nature, and service access is tied to a single device or delivery channel. There is a growing need for supporting heterogeneous service access and sharing service usage experience. Envisaged new sources of revenue for Internet Service Providers include tailored, personalized and dynamically composed services. The key requirements for service platforms include achieving faster time to market, cost efficiency and compelling user experience. In addition, requirements of the car-to-x application scenario are given. The issues of semantic services, service creation and service architecture are discussed in Sections 4.2, 4.3 and 4.4, respectively.

4.2 Semantic Services The next generation (beyond 3G) of mobile services and applications must offer an increasing level of value and differentiation; they will have to adapt to the user preferences, be aware of the user context, and adapt to the device and the radio environment. These services must also be developed easily, deployed quickly and, if necessary, altered efficiently. Machine understandable descriptions of user context, preferences and service capabilities are the key for an automated service adaptation. Semantic technologies support the machine readability of content, and became part of the service-oriented architecture (SOA). This

Service Infrastructures

59

section focuses on the challenges of service delivery in a ubiquitous service environment, supporting the preferences and context of the user and her communication devices. Historically, a service-centric architecture was introduced to let services communicate with each other. WWRF’s user- or I-centric approach is based on the transition of access delivery to service delivery [1]. Current rule-based algorithms become too complex when handling user context and preferences, thus demanding new mechanisms allowing dynamic adaptability of services. The service-centric world was introduced based on Service Level Agreements (SLAs) between trusted partners. In a more dynamic service provisioning world, as envisaged in a Semantic Web services environment, privacy and security become key issues [2]. Our approach is to take advantage of developments in both worlds, using the security and privacy mechanisms of the I-centric world and combining them with the semantic representation of data as known from the Semantic Web (services) world [3]. The key challenge in a user-centric approach is the handling of user preferences, context, devices and connectivity. Experiences from implementations in international projects have shown that managing and updating preferences is a tedious work and that users often disagree with the services selected by a rule-based decision engine. While the home is a rather controlled environment, with trusted and known constellations of devices, service delivery in the mobile/wireless world is more complex. Louis V Gerstner, Jr of IBM said: Picture a day when a billion people will interact with a million eBusinesses via a trillion interconnected, intelligent devices. Pervasive systems do not just mean computers everywhere; it means computers, networks, applications, and services everywhere.

The report from the UK Technology Strategy Board [4] pointed out that the high added value comes from: ž ž ž ž ž ž ž ž ž

Always on – availability of the right content at the right place and time. User-centric solutions – simple and practical person-oriented solutions. Invisibility – numerous, casually accessible, often invisible computing devices. Intelligence – removing the cognitive load through devices with embedded sensing and processing capabilities. Increasing productivity – market value propositions: saving time, saving money. Life-enhancing – penetration of technology into mainstream mass market applications. Innovation – using technology in ways that empower people to work, live and play in radically new ways. Omnipresent – embedded into everyday devices and objects all around. Ubiquity – everyone and everything connected to an increasingly ubiquitous network structure.

To build these types of personalized services is a challenge to the system design as well as user description. This section will address some of the challenges. 4.2.1 Objective and Scope 4.2.1.1 Objectives The objective of the WWRF WG2 is to create a technology-independent blueprint of next-generation service platform architecture. The architecture is used as a basis for a

60

Technologies for the Wireless Future – Volume 3

discussion with the different players; this includes understanding why market participants have come up with new architectures as well as a clear understanding of the underlying network architectures. Semantic service provision is seen as a glue to connect the different platform parts in an economic way, and thus link service environment, user aspects and user context together to form a dynamic service world. Further steps in WWRF WG2 are to bring the WG2 results into a system architecture, identifying (high-level) components, their distribution and their interworking (interfaces and protocols) based on intensive discussions among experts from operators, manufacturers, content providers, IT/software industry and academia. The task of this section is to elaborate where semantic definitions can support the foreseen WG2 architecture for personalized and trusted services. 4.2.1.2 Scope A next-generation mobile service network has to establish its own value proposition between the stakeholders of next-generation mobile service provisioning. It has to place its unique selling points into the overall service-centric environment dealing with a number of different access systems. Mobile and wireless networks have the potential to act in a central role within this service environment and are therefore required to be capable of acting as a service control environment. It is not the intention to copy all successful Internet services but to support them in an efficient and trustworthy way. The service platform architecture is interfacing with the services and applications at the upper layer of a communication system. The service platform architecture covers service support components (such as generic service elements), their relationship and their internal and external interfaces. It usually targets the upper system layers, as depicted in Figure 4.1, however some functions or parts of them reside in lower system layers, e.g. location or mobility. When considering ubiquitous communications environments such as sensor networks, the layering is diminishing anyway. Service architecture for future mobile systems must be able to satisfy requirements from a broad range of services and different kinds of service usage. This can be reflected in the consideration and analysis of various future life scenarios. With this approach a scenario-centred design process can be applied. Although a number of scenarios are available, their impact on service architectures is rarely obvious. The reason might be that service architectures have not been in the focus of the scenario creator at all. Alternatively, the scenarios might be tailored to specific service architectures, i.e. those activities do not comply with the scenario-centred design process. In cases where scenarios were in fact considered for the design process, they often describe very specific aspects and the resulting service architectures are very much tailored to them. To gain the desired demands from future life scenarios on service architectures, it is required to evaluate a selection of scenarios to derive commonalities and their specific features. A first analysis results in a number of aspects, which can be taken as initial input for the service architecture design: future services will be context-sensitive, adaptive and personalized. They will be available in different networks, with different bandwidth/QoS and for different devices or multimodal UIs, respectively. Service architectures need to support sophisticated charging and billing, security and privacy, identity management, DRM and trust. The complete service lifecycle is to be reflected, from service creation and composition

Service Infrastructures

61

semantic preference description? ... User Model & Ambient Awareness

Content & Communications

Appl. Scenarios Adaptation

Personalization

Natural Interaction

Conflict Resolution

Service Deployment

Environment Monitoring

Service Creation

Service Discovery

Service Control

Service Bundling

The semantic glue

Application Support Layer Service Execution Layer Service Support Layer Network Control & Management Layer IP Transport Layer Networks Terminals

Figure 4.1 Target area of the WG2 architecture work shown in the reference model

to service discovery and delivery. Semantically described services can provide a cost-effective service environment, supporting in particular service testing and service update. In addition, semantic descriptions ease user preference descriptions and support community tagging. 4.2.1.3 Approach Based on the current developments in semantic service delivery, WWRF WG2 proposes the following approach to address the above-stated objectives and scope of future service architectures: ž Use semantics to describe the service environment, including: 1. User preferences, profiles and roles. 2. User context like location, communication capabilities and connectivity. 3. Interfaces to the service world. ž Establish a semantic service architecture, supporting automated service incorporation and configuration of complex services. ž Specify the transmission zone between the remote service world (Internet services) and the wireless (phone or proximity) service environment of the user. ž Identify unique selling points as value added services (D high-level features) provided by mobile service platforms. ž Verify the value added services with service scenarios. ž Identify a suitable architecture supporting those features and addressing the above-stated scope.

62

Technologies for the Wireless Future – Volume 3

ž Design an architecture blueprint in terms of high-level components and interfaces. ž Identify platforms and software engineering methods for its realization. WWRF WG2 sees a semantic user preference description as a potential way to establish a dynamic service offer to the end user, adjusted to the needs and the context of the user. 4.2.1.4 Semantic Service Scenarios In this section, we will concentrate on just one scenario, describing several aspects of semantic service provision. The scenario is taken from the EU FP6 project Adaptive Services Grid (ASG) and provides a service world for a mobile traveler [5]. Similar scenarios are known from other projects describing the dynamics of a future service world. The mobile traveler scenario belongs to the group of location based services (LBSs). The scenario is targeted at mobile users with limited web access as well as at all (other) users with respect to the composition of a number of atomic services which together fulfil the demands of the end service user. The typical chain of events is that an end service user is in a situation where she needs a specific service and uses, for example, a cell phone to access such a service. The service itself is composed and presented by the ASG platform. To give examples, these services address situations in which the user needs location-specific information on (this list is not exhaustive): ž ž ž ž ž ž ž ž ž

hotels restaurants restroom facilities fuelling stations service stations garages/repair shops pharmacies dentists medics/hospitals.

This information can be enhanced through services, for example a route guidance to the destination. An important aspect of the scenario is that it covers the complete service chain from service discovery to billing. In fact, billing is a crucial aspect since it has been shown over the last years that end service users are used to getting services for free from the Internet and, hence, are not (or hardly) willing to pay for such services. A common attempt to deal with this is to try to get money from advertising. Telenor Research, for example, has delivered a number of interesting aspects on this and further aspects regarding the willingness to pay. Another situation where an end service user is willing to pay is in an emergency: emergency services (not only sirens and flashing lights but also finding a pharmacy or even an available room) are of special interest since they form one of very few classes of service where the end service customer is willing to pay for the use. So, profitable business will be possible on the market of services with this service class. In an abstraction of a location based service (LBS), it does not matter whether a local attraction is booked or whether a taxi or hotel is needed in an unknown environment, or even if a traffic accident occurred and immediate help is needed. At this point, it shall be pointed out that mobile users – who either travel on their own or by car – are a very important target

Service Infrastructures

63

group for this kind of service. The car is the most appealing environment for mobile users, as all the services addressed here can easily be ported to public transportation or other means of private transportation. This is why the following part of this subsection focuses on the use of ASG mechanisms in the vehicle. The Mobile User in the Scenario There are some aspects that have to be considered when offering mobile services in a vehicle: ž The vehicle human machine interface (HMI) should be used to minimize distraction and to make use of the optimal resources. ž A travel destination (coming from a navigation system) might be used for input instead of the current location (provided, for example, by a GPS receiver or the localization service of a GSM provider). ž Further data may be available to optimize the service (e.g. ETA (estimated time of arrival)). ž Vehicle hardware could be used to enhance the service (e.g. feeding the target address into the navigation system). The hardware/software setup needed to install such a LBS service in a vehicle is shown, for illustration, schematically in Figure 4.2. This scenario addresses the dynamic service interface, the mobile interface and the user-related service requirements. Similar requirements for service delivery are expected to be found in other mobile/wireless environments. 4.2.2 From Web Services Towards Semantic Web Services 4.2.2.1 Evolution to Future Service Provisioning Personalization is one of the most important characteristics of service provisioning in future telecommunication systems. Personalization means tailoring of services and applications to

My Company App 1

App 2

App 3

Alternative access Work Home Friend Train Car Hotel

Interface Support Comp. WEB

SAP

Telecom Operator Location Personal Security Profile Telecom access

Roaming

Figure 4.2 Multi-domain network and service environment

64

Technologies for the Wireless Future – Volume 3

the very specific needs of a user in a ubiquitous, comfortable service environment, together with a single bill service independent of the partners involved in the end-to-end value chain. Personalization also means allowing fully manual service configuration, carrier access without any automatic vertical handover procedures and individual invoicing of the individual service partners, in case this is required by any user for any reason. Market participants face the following trends: ž Content and services are distributed across domains including fixed line and mobile networks as well as open (e.g. World Wide Web) and closed service environments. Business relations might be distributed. ž Internet services are mainly used by fixed line access, whereas more and more initiatives are evolving to provide Internet services through wireless rather than mobile access. ž Technology evolution separates call control and transport. The directions of spread out of the service control are regarded as a chance and a threat for mobile operators. ž Transport charges are dropping tremendously due to excessive IP transport capacities in back-bone networks and increased competition in access networks (flat rates). ž IT technologies have a strong impact on telecommunications protocols, system design and interfaces. Convergence is no longer restricted to the protocol layer, but addresses the whole value chain. ž Semantics have entered the business world, e.g. Enterprise Service Architecture (ESA) by SAP and Microsoft, and will influence the service provision platform of mobile operators. ž The growing importance of peer-to-peer services integrates the single user into the group of service and application providers. ž The increasing importance of ubiquitous computing technologies (sensor networks, RFID, NFC) leads to pervasive interaction with service systems and provides new possibilities for services based on rich information. ž Trust in context-aware services still suffers from privacy concerns of the users. In addition, legal aspects restrict providers in offering rich context-aware services. These aspects lead to the following conclusions: ž Users will maintain several business relations in different service domains. ž Access procedures including authentication and authorization will become more and more complex and will happen more often. ž IP-based services will dominate. ž Service control is not by definition in the domain of the telecommunication operator. ž Telecommunication network operators can hardly survive by providing bit pipes only. Value added services are a prerequisite for future business of telecommunication operators. Mobile network operators have to identify their unique selling points and force them into the market. ž A trusted position in the market is important for acceptance by users. Figure 4.2 illustrates our vision of a multi-domain network and service environment. Mobile services platforms of telecommunication providers and mobile operators in parallel to third party service environments provide services through application servers to the customers. Additional functionalities, for example identity management or billing, are depicted as separate functions that may also be provided by third parties.

Service Infrastructures

65

4.2.2.2 Semantic Web The Semantic Web is an evolving extension of the World Wide Web in which web content can be expressed not only in natural language, but also in a form that can be understood, interpreted and used by software agents, thus permitting them to find, share and integrate information more easily [6]. The web is no longer just a medium for human-readable information, but a universal medium for data, information and knowledge exchange. Ontologies are introduced to support the understanding of terms used in the web. Ontologies In order to enable meaningful communication between different service components, a common understanding of used terms and definitions needs to be achieved. But as there is no global common understanding and use of terms, this understanding needs to be established between communicating parties on the fly. In order to enable this mechanism, semantic descriptions, using taxonomies and ontologies, need to be used. Ontologies are for knowledge sharing and reuse, while languages such as XML are perfect to express information in a structured and efficient way. To be able to discuss with one another, communicating parties need to share a common terminology and meaning of terms used. Otherwise, profitable communication is infeasible because of lack of shared understanding. With software systems, this is especially true – two applications cannot interact with each other without common understanding of terms used in the communication. Until now, this common understanding has been achieved awkwardly by hard-coding this information into applications. This is where ontologies come into the picture. Ontologies describe the concepts and their relationships – with different levels of formality – in a domain of discourse. An ontology is more than just a taxonomy (classification of terms) since it can include richer relationships between defined terms. For some applications, a taxonomy can be enough, but without rich relationships between terms it is not possible to express domain-specific knowledge except by defining new terms. Ontologies have been an active research area for a long time. The hardest issue in developing ontologies is the actual conceptualization of the domain. Additionally, to be shared, the ontologies need a representation language. Languages like XML that define the structure of a document, but lack a semantic model, are not enough for describing ontologies – intuitively an XML document may be clear, but computers lack the intuition. In recent years, various ontology languages based on web technologies have been introduced, the two main ones being OWL and WSML. DAMLCOIL, which is based on RDF Schema [7], is another such language. It provides a basic infrastructure that allows machines to make simple inferences. Recently, the DAMLCOIL language was adopted by W3C, which is developing a Web Ontology Language (OWL) [8] based on DAMLCOIL [9]. Like DAMLCOIL, OWL is based on RDF Schema, but both of these languages provide additional vocabulary – for example, relations between classes, cardinality, equality, richer typing of properties, characteristics of properties and enumerated classes – along with a formal semantic to facilitate greater machine readability. The OWL language has a quite strong industry support, and therefore it is expected to become a dominant ontology language for the Semantic Web. Figure 4.3 provides example ontologies for devices, networks and locations. Ontologies are the key components in semantic service provision. Key issues in ontology creation are the complexity of the ontologies and how to enable interworking between

66

Technologies for the Wireless Future – Volume 3

ObjectProperty

Class Longitude

range

Class Latitude

range

hasLongitude

domain

Class

ObjectProperty

domain

Location

Class

ObjectProperty range

domain

hasLocation

Phone

hasLatitude

Location Ontology ObjectProperty

availableAt

range

Class AccessInterface

domain

Network

range

hasAccessInterface domain

domain ObjectProperty

Class subClassOf

ObjectProperty

Class Location

Class

subClassOf subClassOf

Network

range

Class

to subClassOf

subClassOf

Device subClassOf

Class WirelessNetwork

disjointWith

Class

Class

WirelineNetwork

Wireline

subClassOf

subClassOf

Class

Class

Class

Computer

Positioning

Phone

subClassOf subClassOf subClassOf

Class

subClassOf subClassOf

Class

Class

Class

Class

Class

Wireless

Desktop

PDA

GPS

Galileo

GSM

Class GSM

Class

subClassOf

Class subClassOf

UMTS

Mobile subClassOf

Class

Network Ontology

UMTS Device Ontology

Figure 4.3 An example ontology for devices, networks and locations

different ontologies. Creating one main ontology covering user goals, service platform capabilities and service capabilities would make interworking easier, but will result in a complex and hard to handle ontology. When distributing the ontologies into service ontologies, user goal ontologies and platform ontologies, one needs to address interworking, called ‘mediation of ontologies’. Mobile Ontology: Ontology Adoption for Mobile Services Mobile environments and the web converge are forming a shared distributed communication sphere (DCS). This causes the appearance of new settings to be supported, for example when the user utilizes mobile and fixed devices to interact with systems. Interaction and connectivity of mobile applications with the Internet services will become a substantial service offer in the mobile world. To ensure interoperation of mobile and web services, applications and tools (running on heterogeneous various service platforms in such a sphere), developers need to have a shared specification of objects belonging to the sphere and their roles. Certain ontologies have already been developed for the mobile communications domain with the employment of Semantic Web formalisms [10]. However, widespread and global adoption of such ontologies remains a challenge. A Mobile Ontology is being developed as a comprehensive ‘higher-level’ ontology for mobile communication domain. The ontology is a machine readable schema intended for sharing knowledge and exchanging information both across people and across services/applications, and it covers domains related to mobile communications, specifically:

Service Infrastructures

67

addressing persons, terminals, services and networks [11]. Currently, definition and implementation of this Mobile Ontology is managed as a collaborative effort amongst participants of the EU IST SPICE Integrated Project.1 The added values of a Mobile Ontology are: ž Providing an easy and formal way to reference objects from the mobile communications domain (in particular, to serve as an exchange format between mobile service enablers). ž Providing an opportunity to implement enhanced, ontology-based reasoning. ž Providing a formal representation of the domain to be used in research and development projects, and for educational purposes. The Mobile Ontology, as developed in the SPICE project, covers basic communication terms, characterization of the user, terminals and networks, as well as service specifications. The ontologies were developed using RDF and OWL [7], [8]. It reuses existing ontologies such as UAProf from the Open Mobile Alliance (OMA), FOAF from the foaf-project and vCards from W3C. Semantic Web Services To develop and maintain services for the future that are attractive, easy to use and cheap enough, developers have realized that new methodologies, techniques and tools are necessary. Based on these facts, concepts and technologies like service-oriented architectures (SOA), web services (WS), Semantic Web (SW) and Semantic Web services (SWS) have gradually grown up to show their viability, especially if they are used in combination. Ongoing work in standardization bodies (e.g. W3C and OASIS) and research forums (e.g. EU IST FP6/FP7) support these developments. The ASG (Adaptive Service Grid) and the SPICE projects are typical examples of EU projects that have as goals utilization of concepts and technologies from SOA, WS, SW and SWS, and demonstration of their maturity by developing services within selected business domains. All major software vendors have already adopted web service (WS) technology, i.e. WSDL (Web Services Description Language), SOAP (Simple Object Access Protocol) and UDDI (Universal Description, Discovery and Integration), as a cornerstone in their future tools for service creation. They move to SWS to meet the requirements of future enterprise systems, to create and execute in near real-time and cope with context awareness, personalization and mobility. SWS adds extra semantics to the service descriptions, enabling WS to work together in a more flexible, intelligent and automatic way. There are two major approaches to extending WS to SWS, as indicated in Figure 4.4. Most approaches, such as OWL-S, WSMO and SWSF, create a SWS description language and then map it to the web service. WSDL-S and SAWSDL, on the other hand, directly annotate the WSDL file and define semantics as part of the file. The terminology used in SWS is recalled here: XML provides a syntax for structured documents, but imposes no semantic constraints on the meaning of these documents. XML Schema is a language for restricting the structure of XML documents and it also extends XML with data types, which might be seen as a first step to adding semantics. RDF is a data model for objects (‘resources’) and relations between them, and provides a simple semantics for this data model. These data models can be represented in XML syntax. RDF Schema is a vocabulary for describing properties and classes of RDF resources, with a semantics for generalization 1 SPICE:

http://www.ist-spice.org/.

68

Technologies for the Wireless Future – Volume 3

WS platform Service Registry

Published in

Searches

Service request

Semantic Web Service Service Discovery

Sent To

Input For Message Exchange

Creates Service Request

Message Exchange

Service Invocation

Client

Described In

Web Service

Service Description

Understands

Semantic Description

Understands

Expressed In Terms OF

Requestor Ontology

Provider Ontology

Web Service Ontology

Understands

Expressed In Terms Of

Figure 4.4 Elements involve in Semantic Web service delivery

hierarchies of such properties and classes [8]. This section will describe the basic relationships; a detailed description of the Semantic Web services approaches (OWL-S, WSMO, WSDL-S, SWSF and SAWSDL) is found in the W3C submissions and related documents [9], [12], [13], [14], [15], [16], [17], [18]. Web Ontology Language for Services (OWL-S) OWL-S, former DAML-S, is an OWL-based web service ontology [12] with first specifications named DAML-S 0.5 back in 2001. It provides a core set of markup language constructs for describing properties and capabilities of web services in a computer-interpretable form. It is the first well-researched web service ontology, and as such the reference for all further developments. OWL-S provides a semantic view of the service through a profile. The link to WS or other types of service is called grounding. Figure 4.5 provides the formal mapping between the OWL-S-described ontology and the WSDL-described web service. It subdivides between the operation/process of the service and the input/output from the services. The process is described through a process model, while the input/output is described through types. OWL adds more vocabulary for describing properties and classes: among others, relations between classes (e.g. disjointness), cardinality (e.g. ‘exactly one’), equality, richer typing of properties, characteristics of properties (e.g. symmetry) and enumerated classes. OWL-S is quite complex for a nonexpert to understand. It needs tools like OWL-S API and WSDL2OWL, which currently have some severe limitations. A more important criticism is that OWL-S focuses only on one ontology; it should allow multiple parameters mapping to one syntactic parameter.

Service Infrastructures

69

OWL-S Process Model

Description logic based types

Atomic Process

Inputs/Outputs

Operation

Message

Bindings to SOAP, HTTP, etc.

WSDL Figure 4.5 Formal mapping of OWL-S to web services (WS) [12]

Web Services Semantics (WSDL-S) WSDL-S, submitted to the W3C by IBM on 1 October 2005 [19], uses the extensibility elements of WSDL 2.0 [20], i.e. semantic annotations are added to the WSDL document elements that have constructs to represent service descriptions like interface, operation, message, binding, service and endpoint. The first three (interface, operation and message constructs) deal with the abstract definition of a service, while the remaining three (binding, service and endpoint constructs) deal with service implementation. The WSDL-S proposal focuses on semantically annotating the abstract definition of a service to enable dynamic discovery, composition and invocation of services. WSDL-S defines a semantic model to capture the terms and concepts used to describe and represent the web service. WSDL-S uses the extensibility of WSDL. Four particular parts of the semantic model are distinguished: ž Input semantics – the meaning of input parameters. ž Output semantics – the meaning of output parameters. ž Precondition – a set of semantic statements that are required to be true before an operation can be successfully invoked. ž Effect – a set of semantic statements that must be true after an operation completes execution after being invoked. Different effects can be true depending on whether the operation completed successfully or unsuccessfully. Semantic annotation in WS descriptions is an obvious first step in linking web services with Semantic Web technologies. It relies on both the WSDL and XML Schema extension mechanisms to reference external semantic models. It is a lightweight approach extending WSDL files with semantic annotations. WSDL-S is independent of any particular semantic annotation language, examples of which include WSMO and OWL-S. The lightweight approach makes WSDL-S easy to handle. However, it does not support mapping of many semantic parameters to one web service parameter, as it is restricted to the principle mechanisms of web services, mainly to achieve a result specific to one service. The

70

Technologies for the Wireless Future – Volume 3

main criticism is that it is too flexible with respect to model, script and rule language, so values of semantic parameters are not always identical to values of syntactic parameters. Semantic Web Services Framework (SWSF) SWSF was submitted to W3C by the National Institute of Standards and Technology (NIST), the National Research Council of Canada, SRI International, Stanford University, Toshiba Corporation and the University of Southampton on 9 September 2005 [15]. SWSF includes the Semantic Web Services Language (SWSL) and the Semantic Web Services Ontology (SWSO) [16]. It builds on the WSDL (v1.1) and extends through supporting the specification of work flows composed of basic services. A promising candidate is the Business Process Execution Language for Web Services (BPEL4WS). The second focus is on the choreography, information exchange and agreed ordering rules to perform a web service based transaction. With respect to registering web services for purposes of advertising and discovery, SWSF will build on UDDI. FLOWS, the ontology of service concepts used in SWSF, is a more comprehensive ontology than OWL-S. In terms of coverage it is distinguished by its axiomatizing of messages, something that was not addressed in OWL-S. Both attempt to provide an ontology for web services, but FLOWS had the additional objective of acting as a focal point for interoperability, enabling other business process modeling languages to be expressed or related to FLOWS. Web Service Modeling Ontology (WSMO) WSMO was submitted to W3C by DERI Innsbruck at the Leopold-Franzens-Universit¨at Innsbruck, Austria, DERI Galway at the National University of Ireland, Galway, Ireland, BT, The Open University and SAP AG on 4 April 2005 [17]. WSMO (Web Service Model Ontology) uses WSMF’s (Web Service Modeling Framework) four elements for describing Semantic Web services [18]: ž ž ž ž

Ontologies, for terminology used by other elements. Goals, which define the problems the web service is to solve. Web services descriptions, which define different aspects of a web service. Mediators, which handle interoperability of other ontologies.

These four are syntactically modeled by WSML (Web Service Modeling Language). WSDL is a form of syntactical contract that works by specifying the format of the messages sent between the web service and the client. In contrast, WSMO describes functionality and behaviour of the web service. The descriptions are for discovering and automatic composition of web services. The UDDI registry for WSDL specifications is based on keywords. WSMO discovery uses the semantics in the WSMO description of a web service. WSMO can be located in UDDI registers, but it is more effective to have a specific WSMO repository, as in the ASG platform. SAWSDL SAWSDL (Semantic Annotations for WSDL) specification [14] defines a set of extension attributes for the Web Services Description Language [20], which allows description of additional semantics of WSDL components such as input and output message structures, interfaces and operations. This specification does not address the annotation of WSDL components that deal with service implementations, e.g. binding, service and endpoint.

Service Infrastructures

71

SAWSDL specification defines how semantic annotation is accomplished using references to semantic models, e.g. ontologies. SAWSDL does not specify a language for representing the semantic models. Instead it provides mechanisms by which concepts from the semantic models, typically defined outside the WSDL document, can be referenced from within WSDL components using annotations. To accomplish semantic annotation, SAWSDL defines extension attributes that can be applied both to WSDL elements and to XML Schema elements. The annotations on schema types can be used during web service discovery and composition. In addition, SAWSDL defines an annotation mechanism for specifying the structural mapping of XML Schema types to and from an ontology. Such mappings could be used during invocation, particularly when mediation is required. Comparison of Approaches The following comparisons will provide a high-level view of the approaches, with specific focus on usability for end users. Further details and in-depth discussions are ongoing in W3C and other fora. The OASIS SEE TC and the W3C SAWSDL WG initiative can be seen as the promising attempts to standardize Semantic Web services technologies. SAWSDL attempts to provide an overlay for semantic annotations, thus allowing various approaches to be compared with each other [21]. ž WSMO versus SWSF – WSMO is a parallel effort to SWSF and bases its work on almost the same fundamental technologies, e.g. on F-logic. Nevertheless, the two groups have pursued complimentary goals. WSMO has focused heavily on the language effort. In particular, on developing a ‘conceptual syntax’ for top-level descriptions of services; this might make the specifications easier to read for the end user. WSMO has also paid special attention to the issue of OWL compatibility. The major distinction between the WSMO effort and the SWSF is with respect to the ontology domain. WSMO is focusing on describing web service choreography through guarded transition rules, while SWSF focuses on extending the functionality of the rule language (SWSL-Rules) that supports meta-reasoning and reification extensions. ž WSDL-S versus WSMO – in WSDL-S, the semantic model of a web service is expected to contain the semantics of input and output parameters and the specifications of preconditions and effects of service operations, plus the categorization of a WSDL interface. WSMO specifies a more detailed model where a web service can have a capability with preconditions, assumptions, postconditions and effects, and an interface with choreography and orchestration; and where data are described using ontologies. WSDL-S is based on the WSDL model of web services interfaces consisting of separate operations, and preconditions and effects in WSDL-S are attached to operations. In contrast, WSMO distinguishes between preconditions, assumptions, postconditions and effects, but these aspects are modeled on the whole web service. WSMO talks about a web service as a whole, without splitting it into operations. Additionally, WSMO models the choreography and orchestration interface(s) of a web service, an aspect not covered by WSDL-S at all. ž SAWSDL – SAWSDL specification is currently a W3C working draft, established in March 2006 by the various actors, in order to not let interoperability problems between the approaches hamper the development of Semantic Web services.

72

Technologies for the Wireless Future – Volume 3

4.2.3 Challenges Despite significant developments and breakthroughs, there are challenges in semantic service realizations, especially in today’s distributed and complex service world. This section is going to illustrate the challenges in service integration through interworking of standards and mediation of ontologies. While service interworking is one issue, user-related aspects like security and privacy of information in distributed ontologies are another aspect. These issues are addressed in Section 4.2.3.1. The growing popularity of mobile services demands attention in the context of semantic service delivery to mobile devices. This is challenging considering the limited resources and capabilities of the mobile environment. Section 4.2.3.2 discusses these challenges from context-aware service delivery, privacy and identity-handling points of view. From the evaluation of the existing and upcoming technologies, WG2 concludes that: ž Evolution to future service provision will happen through semantic services. ž Semantic service descriptions will: ž Allow a dynamic service composition. ž Ease complex services to consist of distributed services components. ž Enable service component interactions. ž Provide service lifecycle support. ž Enable preference description and support adaptation to interests. ž Support personalization and context awareness (cc/pp, UAprof, etc.). An industrial uptake of semantic technologies is hampered by non user-friendly tools, which have mainly been developed in academic groups. Focus for further tool development should be on helping ordinary users to establish both ontologies and service engines. The major challenge in mobile semantic service provision comes from the mobile/wireless environment, which is not a SOA environment. The following features have to be taken into consideration to let semantic services become the platform for mobile services: ž Interfaces between the SOA-based architecture and the mobile phone need to be adapted in order to cope with the radio environment. ž Proximity and phone services need to be integrated into the service architecture. ž Radio and context awareness of service provision are necessary, taking into account the highly variable radio environment. 4.2.3.1 Service Realization and Expectations Future service access is based on interworking between different service modules, a distributed service infrastructure and user aspects like context awareness. While semantic interworking within a part of a system, for example one service platform, is becoming reality, interworking between different users, services and service provisioning engines is still a major challenge. Semantic Service Architecture Interfaces Semantic service provisioning will happen through a variety of systems, and has to take into account user and service aspects. Even though most of them might be represented in a semantic way, there will be service components that are not or cannot be semantically described.

Service Infrastructures

73

Examples of such components are sensors or sensor nodes, which have limited capabilities and thus will only provide a minimum set of communication and sensing capabilities. It is also questionable whether mobile phones will allow memory, CPU and battery-consuming applications to be implemented on the phone. Time-critical applications might also require a specific interface. Thus, we expect that semantic systems will have to work together with nonsemantic systems, and this requires an architecture which is beyond the current developments in semantic services delivery. Mediation of Ontologies Even in a system where all components are implemented in a semantic way, the underlying ontology will not be the same. Ontologies are expected to be developed for specific purposes, with contributions from many different persons and organizations. Because of the distributed and open nature of the web, these ontologies can be expected to contain conflicts and semantic overlap; different ontologies would describe (parts of) the same domain in different ways, because of differences in the points of view of the different people who have developed the ontologies. The worldwide ontologies, or ‘swamp of ontologies’, contain the conflict of interoperability because of the problems of granularity, understanding and trust. Mediation of ontologies is a first step in understanding the concepts and relations, but does not take into consideration the backgrounds of the people who created them. There are different aspects in the interoperability of applications on the Semantic Web. In order to achieve interoperability between applications, data need to be exchanged. These data need to be interpreted by the receiver in the way that was intended by the sender. Let us take the example of the telecom world. From a customer’s point of view, quality of experience means how long it takes to set up a call, or the delay experienced when talking. Service operators mediate these quality of experience requirements through a set of QoS parameters establishing a minimum standard. Finally network operators respond to these requirements by mapping QoS parameters into network parameters through setting priority routes and parameters of handovers. Thus a different understanding of words needs to be taken into consideration when enabling mediation of ontologies. Ontologies can help in the interpretation of data through formal and explicit representation of data, which can help machines to interpret data. Having ontologies is not enough to achieve full interoperability between applications, because of the differences in the ontologies used by various applications. The use of different ontologies and the reconciliation of differences between ontologies is called ‘ontology mediation’. Ontology mediation establishes explicitly the relationships between different ontologies. Applications that use different ontologies will be able to interpret data that have been described in terms of an ontology which is not known to the application, but which have been formally related to an ontology that is known to the application. Industry Standards De facto industry standards have been published by industrial players like IBM, SAP, Microsoft and Oracle. One of the open standards-based integration platforms is the Java Business Integration (JBI)2 architecture. This standard extends J2EE and J2SE with business integration service programming interfaces (SPIs). These SPIs enable the creation of a Java 2 http://java.sun.com/integration/.

74

Technologies for the Wireless Future – Volume 3

business integration environment for specifications such as WSCI, BPEL4WS and the W3C Choreography Working Group. The JBI high-level architecture leaves more freedom to platform providers as there is a limited number of mandatory components. Additional components can be integrated using a generic SPI mechanism. The standard was established at the end of 2006, and is subject to prototypical implementation. Other Semantic Web Service Implementations There are also other frameworks and reference implementations for Semantic Web services available. Most notable of these are WSMX, METEOR-S and IRS-III [22], [5]. WSMX3 (Web Service Modeling eXecution environment) is a reference implementation of WSMO.4 The aim of WSMX is to increase business processes automation in a very flexible manner while providing scalable integration solutions. WSMX internal language is WSML5 (Web Service Modeling Language). WSMO and WSML specifications together with WSMX environment are developed by DERI International under ESSI6 cluster (formerly SDK cluster). METEOR-S is a follow-up project to the METEOR project, which focused on workflow management techniques for transactional workflows at the LSDIS Lab,7 University of Georgia. METEOR-S addresses ‘Web-based business processes with the context of service-oriented architecture (SOA) and the semantic Web technologies and standards. METEOR-S attempts to build upon existing SOA and Semantic Web standards whenever possible (using extensibility features) and where appropriate, propose extensions to or seek to influence to existing standards’. The Internet Reasoning Service (IRS) is KMi’s8 Semantic Web services framework, which allows applications to semantically describe and execute web services. The IRS supports the provision of semantic reasoning services within the context of the Semantic Web. There are currently two implementations, IRS-II and IRS-III, which have been applied within different projects. IRS-II follows the UPML framework while IRS-III is a platform and infrastructure for creating WSMO-based Semantic Web services, building upon the previous implementation, IRS-II. Semantic Support for User-centric Services For mobile service provisioning, service adaptation to user preferences, the user context and the communication capabilities of the user are mandatory. The radio environment might be limited, or the user might not want to be overloaded by content. Enabling the service environment to get the user context will help to just send the information suitable for the current context of the user. One potential approach is to associate roles with the user, and thus adopt the service offer to the role of the user in the current situation. Examples of these roles are a user being ‘project leader’ in a corporate environment, and ‘father’ or ‘soccer trainer’ in a social setting. Based on these roles, the user will receive relevant information and get access to resources. The access provisions to resources might have differential privileges, which can be termed as 3 http://www.wsmx.org/. 4 http://www.wsmo.org/. 5 http://www.wsmo.org/wsml/. 6 http://www.essi-cluster.org/. 7 http://lsdis.cs.uga.edu/. 8 http://kmi.open.ac.uk/.

Service Infrastructures

75

policy. Identity management and privacy assurance can be achieved by semantically defined roles, policies and rules. The introduction of semantics and the representation in RDF and XML allows describing user preferences and relations to characterize the community context. The challenge is to extend Semantic Web services with policies representing security requirements for service discovery and privacy protection of user requests. Secure service interaction requires access control and privacy protection. This is more crucial in a community or group environment. Semantic Web technology can introduce a formal way to represent roles, policies and rules to provide access control and privacy services in identity management of the corporate and social community environment. A semantic service architecture may be composed of distributed ontologies based on these roles, policies and rules. Each role has a certain policy (or policies). A policy represents the privilege reserved for each role in a community and expressed through a set of rules. It is such that ‘when an individual is not being authenticated as a member/leader of a project, then he/she is a visitor’. It is a challenge for the Semantic Web community to handle access control and privacy issues using ontology describing roles, policies and rules. Maintaining security of these ontologies is also a crucial issue in a distributed and open web environment.

4.2.3.2 Mobile Semantic Services Mobile semantic services will have to focus on the commonalities and differences between a user-centric (or I-centric) and service-centric approach. The difference between both approaches is historical, where a service-centric architecture was introduced to let services communicate with each other. The I-centric approach of WWRF is based on the transition of access delivery to user-focused service delivery [1]. Personalization and Context-aware Services The mobile service world has made the move to a SOA-oriented architecture. Most of the mobile services, such as location information, are available through a Parlay X web service interface [1]. Noll et al. established semantic annotations of advanced telecom services to achieve exchange of roaming information on a dynamic basis. The main findings of the approach were the cost reductions in service delivery, due to reduced effort for testing and updating of web services in a semantic service world [23]. Two issues remain unsolved when it comes to the usage of SOA in a mobile environment: first, the variation of the radio quality and, second, specific mobile services in the proximity of the user [24]. Radio is a shared resource, and the quality of the radio link is affected by user mobility, radio environment (user speed and coverage radius), application topology and user terminal requirements. A service-oriented platform builds on reliable, minimum delay and high-bandwidth connectivity, which is not achievable in mobile/wireless environments. A proxy can be set as an interface between service-oriented architecture and the mobile world. Such a proxy might function as a virtual phone in the network, and will just exchange the necessary information between the ‘virtual phone’, as seen by the service world, and the real phone in the rough radio environment. Current rule-based algorithms become too complex when handling user context and preferences, thus requiring new mechanisms allowing dynamic adaptability of services. The

76

Technologies for the Wireless Future – Volume 3

service world of a mobile/wireless user consists of proximity and remote services. Examples of proximity services are admittance services and payment through contactless cards. These services are moved to the mobile phone through near field communications (NFC) and are prototyped worldwide, e.g. by MasterCard in Dallas. One goal of these field trials is to demonstrate interworking between wireless technologies and NFC; another goal is to address security issues like potential threats, identity, privacy and simplicity. Adding NFC capabilities to the mobile phone opens it up for key exchange through near field and through the mobile network, thus providing a principle way of delivering authentication information. Conclusions for Semantically Supported Mobile Services Identity handling is crucial when the mobile service world is moving toward SOA. The user will be available (and traceable) 24 hours a day, and electronic transactions initiated from the phone might become a source of surveillance. Thus protection of the user’s privacy and identity will become a key issue. The SIM card of a mobile phone has the potential to serve as the user’s secure identity handling terminal. Current interfaces between the SIM card, the mobile phone and the supporting mobile network are optimized for speed and minimum consumption of radio resources. Considering the fact that the mobile is optimized from a resource point of view, while semantic services are optimized for service creation, composition and delivery, shows the need for adaptation of semantic services towards the mobile environment. A semantic service has a lot to offer in terms of efficient identity handling and privacy assurance, but it does not satisfy the mobile constraints. Some research suggests having a ‘virtual phone’ in the network might be a potential way of dealing with the constraints of the radio environment. The service-centric world was introduced based on Service Level Agreement (SLA) between trusted partners. In a more dynamic service provisioning world, as envisaged in a Semantic Web services environment, privacy and security become key issues. WWRF WG2 takes advantage of developments in both worlds, using the security and privacy mechanisms of the I-centric world and combining them with the semantic representation of data from the Semantic Web (services) world. In the I-centric world security and privacy are important. Based on user preferences and context handling, connections between different services can be tracked. In current mobile systems, privacy is addressed as network operators don’t send user information like MSISDN to service providers. The current usage of ‘open’ ontologies does not support privacy. With a system which is handling the user’s environment inside its private area (phone, laptop, etc.), a malicious tracking software could easily track the user and thus compromise the user’s online life. 4.2.4 Business Considerations While prototypical implementations of semantic services are well under way, only a limited number of studies have addressed the business qualitatively. ASG project members from DaimlerChrysler, Telenor and Polish Telecom have performed an estimation of the business profitability of Semantic Web services [5], [23]. They used a methodology for measuring the costs of semantic services provision, based on ASG architecture implementations of the mobile traveler scenario. The manpower estimates provided in this example are based on the effort analysis used to implement location-based services in the operational networks of Telenor and TP/Orange.

Service Infrastructures

77

Analysing the business of providing advanced services for the mobile user is based on the following steps: 1. Create a service overview, here addressing services for the mobile user, including location information, routing, points of interest (POI) and booking of a service. 2. Identify real world services, here the location services from Telenor and Orange, and make them available as web services. 3. Establish the methodology to convert the existing web services into Semantic Web services, and make them available in an application. Our methodology is based on three areas: ž Svc – the semantic service creation, including semantic description of the service, service testing and registration at the service platform. ž Dmn – domain ontology development, required for defining the service landscape semantically. ž App – the end user application, which is the interface for real users to address the services. Addressing the lifetime of a product means analysing the costs/effort it takes to establish a service, to maintain the service and to modify/upgrade a service. Our analysis aims to compare conventional services with Semantic Web services. The efforts in person days (pd) per step are estimated based on the experiences from the implementation of semantic-based location services. The usage of semantics for services incorporates the following changes: ž Svc – service creation; we assume that services are available as web services, and thus the main effort is to establish a semantic service description. ž Svc – service testing is easier in a Semantic Web service environment, as a testing suite has to be written only once, and can be used for all testing of all services. ž Svc – service registration requires similar effort in conventional and semantic-based services. ž Dmn – domain ontology development is only needed for SWS, and is a clear cost driver. ž App – application development is not considered now, but is subject to further work. The study used the deducted efforts for semantic service delivery, and applied them to two service providers with up to five services each. As expected, initial semantic service description for SWS drives the costs (see Table 4.1). For one service provider the advantages of using SWS are in the easier testing and upgrade of services, but required efforts in semantic service creation and domain ontology development drive the costs. If we include the domain ontology development and update, we receive the manpower estimations of Table 4.1. The figures show that the development of the semantic service description and the domain ontology are the cost driver for the first provider. A second provider will only see incremental costs for the domain ontology adaptation. They will already see cost benefits when providing five services. When calculating service related costs only between conventional and SWS delivery, a break-even point would be achieved for three providers with five services, or five providers with three services. However, including domain ontology will move the break-even point to higher values (right figures) with 15 services provided from five different providers. When including the update of service, the break-even point will be reached for substantially lower number of providers or services. Thus, service maintenance and service upgrade are the items which make SWS delivery advantageous as compared with conventional service delivery.

78

Technologies for the Wireless Future – Volume 3

Table 4.1 Semantic service provision compared to conventional service provisioning

Description

Provider A Serv. p [pd]Ł Serv. q [pd]

Svc – semantic description Svc – testing Svc – registration Dmn – creation/update App – not considered Ł pd

20 3 3 30 –

4 0 2 3 –

Provider B Serv. p [pd] Serv. q [pd]

5 1 1 4 –

2 0 1 1 –

Conv. Serv.

0 4 2 0 –

D person days; Serv. p and Serv. q D Service p and Service q, two different Services.

4.2.5 Summary and Outlook The Semantic Web is an evolving extension of the World Wide Web in which web content can be expressed not only in natural language but also in a form that can be understood, interpreted and used by software agents, thus permitting them to find, share and integrate information more easily. Concepts and technologies like service-oriented architectures (SOA), web services (WS), Semantic Web (SW) and Semantic Web services (SWS) have gradually grown up to show their viability, especially if they are used in combination. Most approaches like OWL-S, WSMO and SWSF create a SWS description language and then map it to the web service. This section presented the basics of the Semantic Web services approaches OWL-S, WSMO, WSDL-S, SWSF and SAWSDL. WSDL-S, submitted to the W3C by IBM on 1 October 2005, is using the Extensibility Elements of WSDL 2.0, i.e. semantic annotations are added to the WSDL document elements that have constructs to represent service descriptions like interface, operation, message, binding, service and endpoint. It does not support mapping of many semantic parameters to one web service parameter, as it is restricted to the principle mechanisms of web services, mainly to achieve a result specific to one service. SWSF was submitted to W3C by the National Institute of Standards and Technology (NIST), the National Research Council of Canada, SRI International, Stanford University, Toshiba Corporation and the University of Southampton on 9 September 2005. SWSF includes the Semantic Web Services Language (SWSL) and the Semantic Web Services Ontology (SWSO). WSMO (Web Service Model Ontology) uses WSMF’s (Web Service Modeling Framework) four elements for describing Semantic Web services: Ontologies understand terminology used by other elements, Goals are the definition of the problem the web services have to solve, web services Descriptions define different aspects of a web service, and Mediators handle interoperability of other ontologies. SAWSDL (Semantic Annotations for WSDL) specification defines a set of extension attributes for the Web Services Description Language (WSDL) that allows description of additional semantics of WSDL components such as input and output message structures, interfaces and operations. Semantic service descriptions will allow a dynamic service composition, ease complex services to consist of distributed services components, enable service component interactions, provide service lifecycle support, enable preference description, support adaptation to interests and support personalization and context awareness. Semantic descriptions are based on ontologies, which are expected to be developed for specific purposes, with contributions from

Service Infrastructures

79

different people and organizations. Because of the distributed and open nature of the web, these ontologies can be expected to contain conflicts and semantic overlap; different ontologies will describe (parts of) the same domain in a different way, because of differences in the points of view of the different people who have developed the ontologies. The worldwide ontologies, or ‘swamp of ontologies’, are potential sources for interoperability because of the problems of granularity, understanding and trust. De facto industry standards have been published by industrial players like IBM, SAP, Microsoft and Oracle. One of the open standards-based integration platforms is the Java Business Integration (JBI) architecture. This standard extends J2EE and J2SE with business integration service programming interfaces (SPIs). A widespread uptake of semantic technologies is currently hampered by non user-friendly tools, which have mainly been developed in academic groups. Focus for further tool development should be on helping ordinary users to establish both ontologies and service engines. The major challenge in mobile semantic service provision comes from the mobile/wireless environment, which is not a SOA environment. Interfaces between the SOA-based architecture and the mobile phone-based services need to be established. Proximity and phone services need to be integrated into the service architecture, and radio and context awareness of service provision needs to be adapted to the radio environment. We suggest the following three areas of research in order to let semantic services become the ‘glue’ for advances in mobile service provisioning: 1. Semantic challenges. 2. Interfaces between semantics and real services. 3. Mobile-specific adaptation of semantic technologies. In the area of semantic challenges, research should be focused on: ž ž ž ž

Trust, security (only web-based) and privacy. User preferences. Scalability. Dynamic ontologies, protected ontologies.

In the area of interfaces between semantics and real services, we see the need to: ž Ease mediation of ontologies in order to connect the ‘swamp of ontologies’. ž Enhance the usability of tools to create semantic descriptions and ontologies for ordinary developers. ž Integrate semantic services with existing business engines. ž Deal with unstructured data, e.g. results from search engines, social tagging. ž Establish one semantic standard instead of the five now existing ones. Mobile specific requirements should be taken into account, as the mobile environment: ž Needs to deal with a ‘highly varying’ radio interface. ž Is optimized for speed and low resource consumption. ž Consists of mobile and context-specific services, which are not expected to be described as Web Services.

80

Technologies for the Wireless Future – Volume 3

Despite the many areas where future research is needed, Semantic Web services have the potential to be the glue between user requirements, service capabilities and service engines, providing personalized and context-aware services.

4.3 Service Creation In order to understand the challenges and upcoming trends in the field of service creation, a short historic view is given below, outlining the changes in the marketplace that require new ways of thinking about service creation. This historic view is divided into three ‘stories’: 1. The former telecom story: the eldest (20th century) story, which builds on the idea of proprietary network (voice) and the fact that the network itself provides most of the functionality that the envisioned service needs. In this story the difficulty and high cost of creating new services is a major issue, along with the closed innovation model, where third parties have very little control over the range of services that can be provided. In the past years it has been very difficult and a major hurdle to deploy services in INs (intelligent networks) and later in Parlay environments. This is mainly visible with the repeated failure of video telephony, which was technically available several years ago, but has never been introduced successfully into the market. 2. A more up-to-date (2007) story within the telecom-driven world can be called the convergence story: today, the full convergence of Internet and telecom services (like VOIP and IP-TV) is under development and multimode devices will blur the borders between fixed, wireless and mobile services, content, networks, domains, etc. Also, the business models will blur with many actors, strongly supported by third parties and even end users. At the same time, regulators will demand a higher level of security control and communication history control of the involved partners, which will limit privacy. 3. The Internet story: the new, more agile and more dynamic story. The main success is due to the quite ‘stupid’ (IP) network and the open policy, which basically means everybody can add new services to the web, at any time. Most of the service innovation is outsourced to third parties and not done by the big players. The Internet is currently providing a more appropriate innovation model by reducing to a minimum the functionality provided by the network (the end-to-end argument) and lowering the barriers to entry to new innovative service providers, yet there is less control over QoS and security. The telecom and convergence stories tell us of a model with centralized control over service development process, where contracts with third parties are central, where time-to-market is long, and where innovation is limited to what operators can offer in their network and what they decide to share with third parties. The Internet story is about a model that is highly decentralized and leads to niche, specialized and personalized services because of its decentralized control, and to higher competition because the pool of third parties is much, much larger. As the technologies are converging due to trends like FMC (fixed-mobile convergence), a more unified approach to creating new services is desired both for the fixed, PC-driven world and for the mobile world. Furthermore, the use of mobile devices enables end users to participate in a passive or even active role to use, generate and even distribute content and additional services either in a peer-to-peer manner or through portals or other hosted services

Service Infrastructures

81

in the network. This trend can be observed with services like Flickr and YouTube, and the upcoming blogging systems. As soon as users are also able to provide services there will be a much larger landscape of available services, and the Web 2.0 ‘hype’ is just a first impression of what could be possible. This section outlines several technical approaches, available and upcoming business models, and will identify future research issues and topics that need to be investigated in order to enable the ubiquitous service offering, driven by service creation opportunities for everybody. 4.3.1 Terms and Definitions 4.3.1.1 Common Terms Technology-agnostic Service Different from a technical service term, technology-agnostic services are services that are recognized and understood without any technical relation and explanation. The descriptions of such services are usually directly related to the nontechnical enabled features and address a specific problem domain. The focus is on solving a problem without pushing a specific technology. Network-agnostic Service The network-agnostic service supports a seamless service access and usability by using Internet protocols, hiding the access complexity and focuses on the application level. Service Development Platform The service development platform is the platform with which services are created. It presents user-friendly and convenient approaches for the development of services. In a service development platform, service creation tools will be provided along with appropriate documentation and examples. Semantic Service Development Platform A semantic service development platform (SemSDP) provides the possibilities to implement a service from a semantic point of view. Semantics describe services developed using the SemSDP, and enable one to create and model services in a technology-agnostic way. Separating technology of a service from meaning of a service could improve the process of assembling new services. Communication Services in SemSDP Communication services in SemSDP are not defined based on their technical feature set. They are specified based on their role and features, in a technology-agnostic manner that will fulfil a user’s need to conduct acts of communication. Next-generation Internet (Web 2.0) Web 2.0 refers to the second generation of web-based services – such as social networking sites and communication tools – that emphasize online collaboration and sharing among users. According to Tim O’Reilly, ‘Web 2.0 is the business revolution in the computer industry caused by the move to the Internet as platform, and an attempt to understand the rules for

82

Technologies for the Wireless Future – Volume 3

success on that new platform. Chief among those rules is this: Build applications that harness network effects to get better the more people use them [45].’ Service Enabler or Component Service enablers provide certain functionalities that are needed by services to fulfil their task. ‘Service enabler’ refers to a technology that is intended for use in the development, deployment or operation of a service, defined in a specification or group of specifications (e.g. OMG). It contains intrinsic functions, which can interact with other functions, within the domain of the architecture and underlying network resources. Service There is a lot of research on the notion of service, not so much in the IT area, but rather in the economic and business sciences. In a generic way, a service can be defined as any business action or business activity that have a value-added result for a user (a person or a system). This action or activity is offered by a service provider (another person, entity or system), which profits from providing this action [25], [26]. In the telecommunication field, a telecom service is defined by 3GPP as ‘a component of the portfolio of choices offered by service providers to a user, functionality offered to a user’ [27]. Relying on the above service definition, we propose to define a service as: A set of activities that consist of interactions between a user and systems controlled by service providers. These activities have a value-added result for the user, and the service providers profit from providing these activities.

Various actors may play the service provider role: network operator, legacy service provider or even device manufacturer (see Section 4.3.4.4). In this definition we highlight two parties: the user and the systems controlled by the service providers. We propose therefore to distinguish between the service behaviour as seen by the user (service domain) and the technical system architecture (technology domain) designed by the provider. From a user perspective, a service fulfils the user’s needs. From a service provider perspective (whatever actor play this role), a service is implemented with technical functions (e.g. APIs, platforms, etc.). Composed Service A combination of services or service enablers that provides some overall functionality is usually denoted as a composed service. Mashups A web site or application that combines content from more than one source into an integrated experience or service is called a mashup or a mashed-up service. 4.3.2 Service Domain 4.3.2.1 Defining Communication: the ‘Communication Model’ The Generative Communication Model (GCM) is a new information processing model in communication which is poised to increase understanding and help create and develop future communication services.

Service Infrastructures

83

I give a speech to the audience I talk to you Speeches Presentations Written books Explanations

Interpersonal

Talks Ph. calls Mails SMS Suprapersonal

Intrapersonal

I think

Thoughts Post-its Action items © Vodafone

RnD, 2005

Figure 4.6 GCM comprises three levels of communication

As illustrated in Figure 4.6, it distinguishes three different basic levels of communication: thinking (intrapersonal communication), dialogues between two people (interpersonal communication) and monologues to a broader, anonymous audience (suprapersonal communication).9 Each level has its own characteristics, and all levels are interdependent and interwoven due to the fact that GCM is a highly fractal model. Working with GCM helps to gain a better insight into human communication, and as a consequence to create successful communication services to support every type of communication. This approach is both profound and far-reaching, and can be customized to application fields ranging from psychology to sociology to technology. GCM tries to capture the very nature of human communication and helps to clarify the connections between the different levels of communication. Before defining the technical requirements of a communication service, one has to take a step back and have a closer look at communication itself. 4.3.2.2 Activity Theory Widely understood activity theory is a philosophical and interdisciplinary framework to analyse different forms of human activity as a developmental process, combining individual and social levels ([28], p. 25). Building on the awareness that every activity always takes place in a broader context and thus cannot be wholly understood without it, activity theory gives a solution for this problem. 9 In

addition to these three levels, a new upcoming level of group communication may be considered. This is private and public group communication, such as private and public chat rooms, IM-channels or even community portals.

84

Technologies for the Wireless Future – Volume 3

A minimum of meaningful context is being included in the unit of analysis. This unit of analysis is being called an ‘activity’ ([28], p. 26). An activity is seen as a system of human ‘doing’ whereby a subject works on an object to obtain a desired outcome. In order to do this, the subject employs artefacts, which may be external (e.g. a communication service, a computer) or internal (e.g. a plan). Central to the modern activity theory, mainly influenced by the work of [29], are so-called activity systems. These systems, developed by Engestr¨om in 1987, describe an activity graphically, thereby serving as a graphical aid to support the appliance of activity theory to practical problems. As [30] argues: ‘It is obvious, that activity cannot exist as an isolated entity’. Hence the community, rules and division of effort have to be included in the unit of analysis. Communication as a central denominator of every human activity plays an important role within each activity system. Although the very nature of communication and therefore the nature of communication systems still have to be subject to further investigation, the use of activity theory and activity systems can give new and interesting insights into this field of research. To clarify the use of activity theory, a simple activity is described with the help of an activity system in the following (see Figure 4.7): Two partners (subjects) are talking to each other. The desired outcome in this case is an agreement between the two. The object they are working on is the dialogue modulation. As a tool they use a telephone. The community consists of the group of peers; rules are, for example, the mutual listening or the common language; the effort is divided into mutual delivery and receiving of messages. Using activity theory to analyse an activity and thus emphasize the character of a system helps to understand the broader context of the activity. Within this context the utilized artefacts or tools play an important role. They facilitate achieving the desired outcome and are heavily influenced by the social level. They have to be in line with the requirements and demands of the social level. Similarly, tools (or services) can be developed by analysing an activity in a broader sense with the help of activity systems.

Tools & artefacts Telephone

Subject Partners

Rules Mutual listening Common Language

Object Dialog modulation

Activity

Community Group of peers

Figure 4.7

Outcome Agreement

Division of Effort Sender / Receiver: Mutual delivery Receiving messages

Example of an activity system

Service Infrastructures

85

4.3.2.3 Semantic-driven Software Architecture Introduction A domain-specific software architecture (DSSA) is defined as: An assemblage of software components, specialized for a particular type of task (domain), generalized for the effective use across that domain, composed in a standardized structure (topology) effective for building successful applications [31].

Another definition claims that DSSA is: a context for patterns of problem elements, solution elements, and situations that define mapping between them [29].

Both definitions utilize abstractions of technical terms and notions like ‘components’, ‘standardized structure’ and ‘elements’. They miss the point of focusing on the expert language and given concepts arising from a problem domain. DSSA in this regard does not consider the wider context of a possible problem domain, it stems from a technical understanding of service specification – and will prefer a technology-driven implementation. Although no standards exist (though perhaps UML is a minor exception), numerous architecture-driven languages (ADL) have been defined to create and draw DSSA outlines. They have been implemented with a variety of guiding principles and emphasis on different styles, e.g.: ž ž ž ž ž

Aesop deals with the specification and analysis of architectural styles [32]. Rapide focuses on specifying component interfaces and component interaction [33]. Wright is based on communication protocols [34]. MetaH was especially created for the real-time domain like avionics [35]. UniCon addresses packaging and functional issues [36].

Additional languages address further specifics [37], but all lack in making semantics of problem domains a relevant element of their basic design. All these ADLs can be considered as double edged: they reflect knowledge that was abstracted from the problem domain (e.g. MetaH focuses on real-time problems in avionics, Rapide outlines interface and interaction issues). But they still do not seem to really reflect the knowledge domain on a deeper level of understanding. From Domain-specific to Semantics-driven Software Architectures The Missing Link of DSSAs The general idea for specifying a Semantics-Driven Software Architecture (SDSA) is based on the paradigm of almost abstracting technical thinking and prejudice from the problem domain level. We envision a blueprint for future software architectures, which emerge from principles of its inherent capabilities: the wise choice of only a couple of modules which reflect a clear definition should lay the foundation to exploit its capabilities for the implementation of a wide range of possible services. The availability of SDSA-based software architecture would help us to get new meaning and insight into the very nature of service creation and the necessary development processes.

86

Technologies for the Wireless Future – Volume 3

SDSA could become an early sign for an ICT industry undergoing a transformation of seismic proportion from technology- to user-centric development. Today, we still lack the tools, processes and methods. SDSA is even poised for managing classes of services, which today might be difficult to predict precisely. Services might be unavailable from a technical and a practical point of view, but they should be envisioned early on a platform which will best serve as a stable and well-understood tool for a couple of years (take, for example, telepathic interfaces or neuro implants). But changes for software architectures in this regard cannot come incrementally: a shift from DSSA to SDSA requires a phase shift that appears problematic and challenging. Once made, it might reveal a new simplicity without the formulaic solutions of too much technology thinking in software architectures today. Model-driven architecture (MDA), for example, is poised to leverage DSSA and can be seen as a typical standard approach in this regard. MDA claims to split abstraction layers, which help with reusing models. That reuse might be easy if considered from a developer’s point of view. However, it is heavier to reuse, when you want to start from an appropriate understanding of the problem domain and want to utilize the MDA for different purposes. How to Lose Semantics: ‘Take a Seat : : : ’ Take the following example from Tracz [38]. He defined three different problem domains: theatre, airline and library. Any domain comprises specific notions, terms and definitions, which reflect a deep understanding of the knowledge domain. Following the reference design principle of DSSA, Tracz identified generalized notions (see last column in Table 4.2). Thus, for example, ‘seat’, ‘seat’ and ‘book’ became ‘item’. From a DSSA understanding, this was an approach to create a reference architecture model for three separated knowledge domains as suggested by Tracz. But, unfortunately, the additional information of the knowledge domain, which resides in the given notions of ‘seat’ and ‘book’, unnoticeably vanished and was not available anymore. It is thus an effort to reassemble the lost notions when starting from a generalized notion: can you be sure that you find the fitting notions of ‘item’ for the problem domains theatre,

Table 4.2 Comparison of theatre, airline, library and inventory domains [38] Theatre Domain

Airline Domain

Library Domain

Inventory Generalization

Seat Row Section Performance Seating arrangement Tickets sold Tickets remaining Price Performance time Performance date Ticket agent

Seat Row Ticket category Flight number Seating arrangement Tickets sold Tickets remaining Price Flight departure Flight date Ticket agent

Book Shelf Section Title Floor plan Books on loan Books available Penalty for lateness N/A Due date Librarian

Item Room/Shelf/Bin Aisle or Building Description Warehouse Items sold Current inventory Cost/Item N/A Expiration date? Clerk

Service Infrastructures

87

airline and library without any additional information? It would be hard to discover the right definition without a deeper knowledge or additional information of the problem domain and its concepts. Any of the three notions reflect a similar concept with regard to the workflow taking place in the domain. But DSSA would generalize those three distinguished problem domains in a reference design by identifying a term which could be used identically for ‘seat’ (in a theatre context), ‘seat’ (still a seat, but in an airplane context), and ‘book’ (no seat, but just a ‘kind of seat’). The conclusion for DSSA: the more complex a knowledge domain is by its nature, the more challenging it will be to create a software architecture that can become a valuable reference design for many different problem domains without losing important information. The disadvantage of generalization is that information about a domain vanishes slowly and unnoticed. It would be more complicated for nonexperts to get a clear understanding about underlying concepts: DSSA pushes terms and notions toward a technical dimension of perceiving the knowledge domain, not a semantic one. DSSAs are designed around technical concepts like ‘interface’, ‘component’, ‘protocol’ and ‘workflow’. Most of these concepts are usually not helpful to explain a given knowledge domain. Why might this fact be important to think about? From psychology research we know that each single word is strongly related to the personal and cultural background of a person and can create confusion and misinterpretation [39]. Sometimes there is no possibility of directly translating a word between two different languages due to the fact that one cannot identify similar concepts. Requirements engineering often deals with these and similar problems [40]. We see shortcomings of software architecture implementations, which arise from technology, instead made out of the semantics of a given knowledge domain. Model-driven architectures (MDA), for example, possess no implicit capability for using problem domain vocabulary. The New Era of Semantics It has been recognized that semantics are becoming a basic concept in numerous realms of IT. Tim Berners-Lee created some groundbreaking papers on the future of the Internet, which he claimed would become a ‘Semantic Web’ [41], [42]. Real-life implementations are going to debut on the market in the near future [43]. Several teams are currently working on the idea of assembling semantic desktops like Nepomuk,10 Gnowsis11 or BeagleCC, which will bring semantics ideas to the everyday desktop workplace. A possible way to overcome those and other limitations of DSSA might be to create software architectures which would offer service models from scratch, reflecting the archetypical flows of information and communication in a problem domain vocabulary, and making the explanation of the wider context of a service an integral explained part of the software architecture. We call those service architectures semantics-driven software architectures (SDSA). It seems to be the right time to reflect on the impact of semantics on future software architectures. Analogous to Boehm [44], who coined the ‘iconic turn’ for the usage of media, we claim that development of software architectures needs a ‘semantic turn’. We state that a ‘semantic turn’ of dealing with services will replace the former principle of 10 http://nepomuk.semanticdesktop.org/xwiki/. 11 http://www.gnowsis.org.

88

Technologies for the Wireless Future – Volume 3

Table 4.3 Comparing domain-specific and semantics-driven software architectures Domain-specific software architecture

Semantics-driven software architecture

Is dealing with abstractions of technical concepts Wants to be unambiguous

Takes into account notions, concepts, and terms given by the problem domain Defines contradictions and ambiguities as valuable and important design principles which must be reflected by the architecture Is not limited to prerequisites, hidden agendas or assumptions about possible technical implementations Makes the ‘usage’ of the problem domain terminology a central feature of the software architecture Sticks to a specification as far as it better reflects the understanding of the given problem domain (no generalization, if clarity lacks)

Usually sticks in a bias on e.g. device form factors or interaction paradigms Just ‘refers’ to an unambiguous terminology of the problem domain Tries to generalize for the purpose of a reference design

exploiting technology as its best. New and promising software services will be less and less dependent on technical enablers. Instead, the groundbreaking idea behind a service will become paramount. It will be the crucial factor of success or failure. Technical problems will diminish in the future; usability topics will emerge as the top challenge in creating valuable services. Web 2.0 lately presented the idea that there are more technologies like AJAX, RSS and microformats necessary to create business value not yet seen [45]. A software development platform based on a technology-agnostic approach would make it easier to address the paramount goal of the ‘disappearance of technology’ posted by Marc Weiser more than a decade ago. From a software design point of view, his fundamental idea addresses important usability and design issues such as outline, for example, by Donald Norman [46]. The Workflow of Software Service Creation We can distinguish three specific realms within an archetypical software development process to build a service upon software architecture: ž Start with a general vision for a service – is there an exciting idea for a service that can be outlined (beyond one or more technical enablers)? ž Specify the service idea, outlining the service itself and adding a business model and other organizational requirements; understand the broader environment of a service, which will have an impact on its introduction, acceptance, usage, etc. ž Implement the service – specific description of the service, which will become the blueprint of its technical implementation. An SDSA should leverage this basic workflow. An SDSA pattern language should reflect the rules of thumb within any of these stages and could become a guiding principle to create numerous implementations of such a software architecture platform.

Service Infrastructures

89

Figure 4.8 Archetypical steps of service development

4.3.2.4 Scenarios and Requirements for Service Creations Preface As computing and communication technologies are becoming more ubiquitous, more accessible and cheaper to operate, users have become equipped with a set of enabling tools that act as a catalyst to their imagination and creativity and a conduct for their ideas and opinions in the different domains that are pertinent to their lifestyle and interaction with technology. This empowerment of users on the global scale has shifted the decisional power on the design of products and services from the manufacturers and service providers to the end users, giving them the opportunity to make their own films, records or other content and distribute it over the web. This shift is better illustrated by the cover of the December 2006 edition of Time magazine. At the end of every year, Time selects the most remarkable person of the year. Interestingly, the magazine simply nominated ‘YOU’12 as the most important person of 2006, indicating that every end user should be celebrated for their involvement in the generation of content and services and for their intelligent and creative use of technology. Their choice was justified by the re-emergence of the World Wide Web as a powerful new paradigm for community collaboration. After being the ‘property’ of researchers, then commercial companies, the Internet has become a tool for bringing together the small contributions of millions of people and making them matter, and a space where experiences can be personalized and shared. The same holds for end user-driven service creation approaches. First tried in academic environments, then used by early adopters, it will sneak into the mass market in the coming years. Every new technology needs a learning/education cycle for people to get familiar with the concepts, and as soon as they can see and feel the benefits in their daily life it will become more widely used. In the following paragraphs, we will highlight the importance of user-centred content and services to end users, and also discuss why it is crucial and beneficial for companies to encourage such practices, and how this can be done by providing the end users with toolkits to guide them in the design and production of new products and services. 12 http://www.time.com/time/magazine/article/0,9171,1569514,00.html.

90

Technologies for the Wireless Future – Volume 3

Manufacturer/Service Provider Perspective The tailoring aspect of any new product or service by companies is considered to be one of the biggest challenges at present. The marketing mantra today is ‘personalization and individualization’, because the user is special and deserves to be the centre of attention. We argue that it is advantageous for companies not just to enable but also to empower users to create services. First, it is normally very expensive to learn about user needs through conventional market research techniques. To get more insight into user needs, more elaborate techniques like ethnographic studies can be deployed, but they are difficult and time-consuming. The task of finding out what the user of devices and services is set to be becomes more complex as the pace of change is particularly fast in the domain of wireless technologies. Second, if users are enabled and empowered to create services, customization and, in particular, mass customization can occur without extensive investment on the part of the provider. There are just too many contexts and environments (including technical, physical and worldwide) to cover and consider during the development of services. Validating the scenarios of use of services generated by providers is necessary but can be an expensive process. Third, if the service creation function is handed over to the users themselves, providers are tapping into a valuable resource because the market information is ‘coming from a direct source’ if there is a direct channel of communication. Furthermore, the users are able to provide many different ideas and new scenarios due to their diverse needs and levels of experience, and often they are much more creative than business-focused development teams. According to von Hippel [47], the traditional iterative approach to the creation of new services and products that manufacturers and service providers follow (design, build, run and analyse) may not yield solutions that will satisfy the needs of real users within real contexts of use. He argues that users will have better information about their needs (compared to manufacturers) but a less clear and lower quality model of the solution (i.e. they lack the technical expertise to build the solution). He refers to this as ‘information asymmetry’, as users will tend to develop innovations that draw heavily on their own information about need and context of use, whereas manufacturers will tend to develop innovations that draw heavily on the types of information in which they specialize. He suggests that companies (manufacturers and service providers) should outsource key need-related innovation tasks to their users, after equipping them with appropriate ‘user toolkits for innovation’. This difference in the locus of innovation-related information can also help companies generate revenues from user-initiated innovations. While information and early prototypes can be produced by the users and distributed freely, the actual production and diffusion of physical products requires the expertise and processes mastered by the companies for mass production and deployment. The process by which companies acquire the knowledge from the user and convert it into usable products is not straightforward. Von Hippel suggests that companies should learn when and how to listen to users. The recommended approach to facilitating the design and production tasks is to supply toolkits that can serve as ‘platforms’ upon which users can develop, operate and test user-developed modifications. The goal of a toolkit is ultimately to enable nonspecialist users to design high-quality, custom products that exactly meet their needs.

Service Infrastructures

91

Some of the characteristics suggested by von Hippel that we can adapt to wireless services are as follows: ž Learning through trial and error – ability to design, test and run the design, which may reveal errors that the user can resolve. For instance, a mobile service toolkit should allow the user to either emulate the service or deploy it locally on their portable device. ž Appropriate solution spaces – all designs are realizable if they fall within the limits of the solution space, i.e. the possible configurations of the service. For instance, the service provider could limit the number of options for each service component and also restrict the number of possible logical combinations of the components. ž User-friendly tools – taking into consideration the skills required and the language used by the users. Making the toolkit user-friendly means offering different levels of complexity to the users/innovators, ensuring that a suitable semantics is utilized. ž Module libraries – provision of standard modules enables users to focus on the creative aspects of the service production. The toolkits should provide the right balance between service granularity and the intended level of complexity of the composite service to be created. The (End) User Perspective According to von Hippel [47] between 10 % and nearly 40 % of the user base engages in developing new products or modifying existing ones. These ‘lead users’ are at the leading edge of the market with respect to important market trends, and can articulate their own needs and identify where the gaps are in the current services available, and also why their needs have not been met. These are the people that need to be identified early and empowered by companies. The obvious question is why people will be willing to reveal their innovations through the product or service they create for free instead of patenting them. It can be said that, provided with the right environment and tools, the users themselves are motivated to be creative because it is only human to be so, and this is an opportunity for self-expression and sharing. Von Hippel points out that free revealing is often the best practical option available to user innovators as it is difficult to protect from imitations. Users who are willing to protect their innovations should have a monopoly on the innovation-related information, which is difficult to keep for a single individual when there are other people who are likely to reveal this information. The availability of communications means has facilitated the diffusion of information and the wide availability of design, development and testing tools makes it much more difficult and less appealing to hide innovation-related information. In many cases (particularly in the computing area) this revealed information can be exploited by the manufacturers and service providers to create suitable products or improve on pre-existing ones, and make it available to a wider population of users. Young users do not think about patents etc.; they care about becoming visible and gaining their minute of fame within a specific community, e.g. become featured on YouTube, having the most popular Ning service, most diggs on digg.com, etc. To explain the committed contribution of ‘benevolent’ innovators to the creation of new products, let us analyse the Open Source Software (OSS) movement, one of the most

92

Technologies for the Wireless Future – Volume 3

prominent examples of user-centric innovation and innovation communities. The OSS movement aims to create software whereby the source code is accessible for analysis and alteration. Generally, software is first developed by a small team of capable innovators who are willing to voluntarily reveal their innovations, and then a number of interested volunteers who find the information revealed to be of interest join in the efforts. They can contribute by code writing, bug finding, suggestion, documentation writing, etc. The OSS community is supported by a number of tools like mailing lists, forums, version control software, repositories, bug trackers, etc. Hars and Ou [48] carried out a survey with OSS programmers and concluded that ‘external rewards’ such as altruism have greater weight than ‘internal factors’ such as expected future returns. In support of this statement, Linus Torvalds, who created the core of Linux, says that people contribute to OSS not really because of fame and reputation but because ‘it is fun to program’ [49]. Similarly, Eric Raymond views the activity of contributing to free software as a joyful experience and even ‘self-actualization or transcendence need’ [50]. Other research studies find that people contribute to OSS because they identify with the software and have pragmatic motives to improve it, without expecting any individual reward [51]. Finally, two large-scale developers’ surveys [52], [53] found out that the main reason for people to join and continue working on open source projects is to expand and share their knowledge. But many see the benefit of gaining a good name within the open source community to get good job offers. An important open source contribution is often worth more than publications in the IT industry. And as with journals and conferences, some open source projects have more prestige than others. In general, for economical and logistical reasons, companies carry out market segmentation studies to group people with similar (or close) needs. In doing so, they will miss out on those people who have specialized needs, or those who are not completely satisfied. Users may be willing to spend their own resources to create their own solutions, provided that they have the capacity to do so. Those who do not may support others (which may involve financial contribution like donations) to do it for them (examples of such an approach are observed in the OSS movement). Other people (or institutions) with sufficient financial resources are ready to pay great money to do exactly what they need (e.g. customized web site, software or car). Sometimes the accounting considerations do not apply and we can find people who are willing to give their time and efforts to create a custom product because of the reward/reputation/control/self-esteem they obtain, even if the monetary benefits are minimal, as we have seen in the case of OSS. User innovation does not have to be a solitary undertaking; it can, in fact, be enhanced through organized cooperation. Communities of users13 can increase the speed and effectiveness with which users and also manufacturers can develop, test and diffuse their innovations (for example, by using tools and modules already developed by other users). Free and open source software projects are a relatively well developed and very successful form of Internet-based innovation community, but this is not limited to software. Improvements in computer software (e.g. CAD tools) and hardware (e.g. cutting machines) and access to hardware components have made it possible for hobbyists to create richer innovation with hardware and electronics. 13 These are generally online communities, but they can also be supported by paper publications like Make (http://makezine.com).

Service Infrastructures

93

Why Should End Users Develop Services? Telecom operators and service providers offer an always increasing number of services to their customers, but the impossibility of foreseeing all the conditions and needs that a customer can experience in their daily life makes it very difficult to provide an exhaustive set of services. As the end user has, besides services, access to more networks and technology enablers, it seems interesting to investigate the possibility of allowing end users to define their own services whenever they need them. This goal is very challenging and poses a set of constraints mainly related to the simplicity of the way the end user can create their own service and the security issues related. The main idea behind the end user service creation is to shorten the analysis/development/ marketing chain, giving users the possibility of satisfying their specific needs, creating new services by means of a very much assisted assembly process; the services will be the result of a composition of a set of pre-existing components. An assembly process targeting end users can be supported by different tools with different degrees of friendliness and inherent inverse implementation complexity. These tools can range from natural language interpreters to wizard-based rules editors to graphical component assembly tools. What will be described is an approach that is well suited for a very much assisted graphical composition tool that can mediate between implementation complexity and effectiveness of the solution; moreover it can be a core solution to be extended in a stepwise approach with natural language interpreters, for example to propose initial automatic compositions to be manually visualized, verified and corrected, or with wizards to define some details of the composition. The following paragraphs will address the definition of the main characteristics of the notation of a service description language for the end user. All the definitions in the notation for end users will be driven by the main requirement of simplifying and controlling the creation process as much as possible [54]. General Considerations on End User Graphical Notation The service creation for end users will be based on the composition of existing components, each representing a basic piece of service: the end user will be the assembler of some service components chosen out of a set of already available ones, eventually dynamically updated from some centrally provided repository. One of the requirements of the notation is to be able to describe telecom services, which often have the characteristic of being asynchronous and event-based (e.g. ‘whenever I get an incoming message : : : ’); that’s why the service notation will support an event-driven specification paradigm, i.e. service building blocks will generate and process events and the service definition will specify the flow of events inside the service. From a graphical notation perspective, the service control flow will be expressed by means of arcs connecting components. Proprietary, graphical development environments based on service independent building blocks (SIBs) have existed in the telecom industry for many years, often within the intelligent network domain. These service creation environments (SCEs) did not allow reusing service functions present in SIBs because of a lack of underlying protocol dependence, a lack of software reusability, the absence of standards, and a lack of openness by manufacturers and operators. The components selected by the end user in a service design process are general: they provide a standard behaviour that has to be customized for the specific needs of a concrete

94

Technologies for the Wireless Future – Volume 3

service, where they need to be configured to determine their specific functions inside the service; the notation will provide a simple and intuitive support for this task, with a guided graphical interface. The graphical notation must abstract from all protocol details; these will be completely masked by the component implementation, in fact each building block will expose only simple communication concepts, those the user is familiar with: e.g. start/receive a voice call, send/receive messages, and look for information. From an end-user perspective, service composition must be intuitive and self-explaining; no software development skills must be required to develop a new service; moreover the service creation environment for end users will have to feature a service composition assistant: e.g. whenever the user picks a block, the SCE can suggest a selection of other blocks that can be connected to it, or when a connection is being created, the SCE can list a set of syntactically or semantically meaningful endpoints for it. Examples of User-generated Products/Services Software/Content Examples of collaborative efforts to create a service, information or a content include a large library of OSS software, of which we can mention Linux (operating system), Apache (web server) and Mozilla/Firefox (web browser). This unorthodox approach to creating software has been adapted to other activities such as education (e.g. http://www.wikipedia.org, the free encyclopaedia, maintained by a community of writers), music production (e.g. RPM challenge, http://www.rpmchallenge.com, aims to record an album of original recording material in 28 days, collaboratively and using free tools) and movie production (e.g. the Elephants Dream movie, http://orange.blender.org/, which was entirely produced with an open source 3D modelling and rendering application called Blender (http://www.blender.org)). Web There is currently an explosion of user creativity and innovations, which is facilitated by the increasingly accessible multimedia tools and diversity of computing devices that can store and play our content. Most of these tools became accessible on the Internet thanks to a set of technologies that try to provide an experience similar to that of a computer application. The web has literally become the Operating System. There is even a term coined for these technologies: Web 2.0, to indicate the evolution of the web from the static web pages of the last decade, to dynamic media-rich content and complex layouts. An illustrative list of web sites dedicated to facilitate the editing of multimedia content and sharing online is shown below: ž Many Eyes (http://services.alphaworks.ibm.com/manyeyes/home) – provides users with a set of online tools to visualize data sets and enable ‘a new social kind of data analysis’. ž OurStory (http://www.ourstory.com/) – this web site lets people create historical timelines with visuals and text. ž FilmLoop (http://www.filmloop.com) – this web site lets user create a slide show of pictures that scroll across the screen ž iBloks (http://www.ibloks.com) – this web site lets people create 3D interactive multimedia presentations that layer photos and videos with music. ž Tabblo (www.tabblo.com/) – this web site allows the sharing and printing of photos with family and friends.

Service Infrastructures

95

ž Pickle (http://www.pickle.com/) – similar to Tabblo, this allows the sharing of pictures and videos. ž Ning (http://www.ning.com/) – a simple web-based service composition environment to build social network services. Web Mashups Earlier in this section we saw examples of dynamic web applications, where users are able to provide their content. A further step is made by experienced users or developers who are able to combine the content of different web applications in a new application, a mashup. Google offers an API to utilize most of its online services, such as Google Search and Google Maps (http://maps.google.com/), and allow third parties to embed them in their websites, customize the service to their needs and even overlay them with additional functionality. There are even about a dozen real playable games based on Google Maps that can now be accessed. Other similar services are available on the Internet, which people and corporations are freely allowed to remotely access and manipulate. Commercially, online retailers such as Amazon and eBay allow the easy creation of e-business web sites by exposing their technological resources and infrastructure. Through web service technologies, they offer individual developers, entrepreneurs and corporations the ability to easily create and maintain virtual shops. This process of ‘mashup creation’ can obviously be done at the level of a web programming language (e.g. PHP, Java) by developers. A very interesting new trend is the offering of frameworks to ease the creation of mashups by major leaders in the web arena. In fact Google, Microsoft and Yahoo! have proposed Google Mashup Editor, Microsoft Popfly and Yahoo! Pipes, respectively. Here is the description of these frameworks according to their web sites: ž Google Mashup Editor (http://editor.googlemashups.com/) – users can create mashups and simple applications quickly using the Google Mashup Editor. ž Microsoft Popfly (http://www.popfly.ms/) – Popfly is the fun, easy way to build and share mashups, gadgets, web pages, and applications. Popfly consists of two parts: 1) Popfly Creator is a set of online visual tools for building web pages and mashups; 2) Popfly Space is an online community of creators where you can host, share, rate, comment and even remix creations from other Popfly users. ž Yahoo! Pipes (http://pipes.yahoo.com/pipes/) – Pipes is an interactive data aggregator and manipulator that lets you mashup your favorite online data sources. Gaming and Virtual Worlds The gaming industry has been a pioneer in the provision of toolkits for the modification of games (also called ‘mods’). These modifications may include the addition of new content, for example weapons and characters, or the creation of entirely new games. There are numerous games that support mods, such as World of Warcraft, Half Life, The Sims and Quake. These mods are developed by the gamers and freely shared over the Internet. Some of these mods have resulted in greater commercial successes for some games, so much so that game development houses provide extensive documentation and open forums to assist mod makers. There are also some ‘YouTubes’ for games starting up, so the trend of user-generated content is evolving fast into user-generated services. People are also making a living by selling, for

96

Technologies for the Wireless Future – Volume 3

example, virtual furniture in Second Life. At the Nordic Game conference in 2007, there was a great focus on the communities that grow around online games and how the game producers should support them. Hardware DIY User innovation and creativity is not limited to information-centric activities, and can also be witnessed in hardware hacking. The DIY (do it yourself) philosophy ‘questions the uniqueness of the expert’s expertise, and promotes the ability of the ordinary person to learn to do more than he or she thought was possible.’14 The aim could be to customize a product, to add new functionality, to fix a problem or simply to understand the way a device works. Examples of hacked hardware include anything with computers, electronics, robotics, metalworking and woodworking, such as game consoles (e.g. creating a version of Linux to run on Microsoft Xbox), sport equipment, home entertainment systems, etc. The power of hardware DIY is best illustrated by the work of Prof. Neil Gershenfeld (from the MIT Center for Bits and Atoms15 ), who argues that the evolution of computer hardware and software over the last decade, from large, expensive and nonaccessible mainframes to small, cheap and user-friendly portable computers, is replicable in the hardware realms through the use of CAD and fabrication tools. In his vision, we are still at the mainframe age when it comes to machine-building machines. He goes as far as predicting that in the near future, tools will be developed and put at the disposal of users to manipulate material at the molecular and atomic levels. The Center for Bits and Atoms has already set up a number of laboratories (via the ‘Fab Lab’ programme) across the globe, where students and under-served communities research and use tools for personal fabrication of machines. These laboratories are equipped with a variety of tools for measuring, cutting and assembling from the macro to the micro level. 4.3.3 Technology Domain 4.3.3.1 Creating Services from Scratch Nowadays, new services are built by using so-called integrated development environments (IDEs). These environments are typically editors, execution environments and debuggers for given programming languages. Some of these IDEs are open source (like Eclipse), others are commercial applications (like .NET Visual Studio). The different IDEs offer different levels of support for the underlying system. For instance, the .NET IDE has quite a good support for the Microsoft platforms, whereas Eclipse offers a more open approach to support any operating system and even offers different languages like CCC, Java, etc. 4.3.3.2 Service Creation by Composition The service creation workbench (SCW) is an environment designed for service developers to easily create, modify and compose new services. It provides developers with a collection of configured modules that enables various easy compositions of services. 14 http://www.wikipedia.org. 15 http://cba.mit.edu/.

Service Infrastructures

97

The SCW supports the following: ž The design of new services – it enables identification of service criteria and representation of the necessary technologies. ž The chronological design of services – it enables the collaboration of multiple service enablers, technical functions and their information sources and flows (workflow, choreography, feature interactions, composition, etc.). ž The finalization of operation logic and distribution of enablers executed by the service. ž The extension of existing services by the development of new modules. These functions enable developers to: ž Compose new services and extend existing ones quickly, without any in-depth knowledge of all the underlying technologies. ž (Understand and) identify similarities between services. ž Construct applicable synergies or new possible services through combination of existing services. 4.3.3.3 Service Execution Environments (Application Servers and Platforms) Distributed complex services/applications are typically implemented over service execution environments that provide a large set of common facilities (e.g. persistence, lifecycle management, inter-process communication, user management). The developer of a specific distributed service/application can better focus on the characteristic ‘business logic’ of the service/application, exploiting the available facilities. A large class of service execution environments is represented by application servers that implement the Java Enterprise Edition platform.16 In the following we first present the JOnAS application server. Then a service execution environment targeted to the need of telecommunication services – the JAIN SLEE – is presented. All these service execution environments are meant for running application on fixed servers; they do not target the execution on services on mobile devices. In other words, the service logic of the application runs on the server side and the client is typically a web browser. Another class of service execution environment targets the execution of services also on the mobile devices. One example of this class is the MUPE platform described at the end of this section. JOnAS (Java Open Application Server) JOnAS is a pure Java open source application server conforming to the J2EE specification. However, its high modularity allows it to be used: ž As a J2EE server, for deploying and running EAR applications (i.e. applications composed of both web and EJB components). ž As an EJB container, for deploying and running EJB components (e.g. for applications without web interfaces or when using JSP/Servlet engines that are not integrated as a JOnAS J2EE container). ž As a web container, for deploying and running JSPs and Servlets (e.g. for applications without EJB components). 16 See http://en.wikipedia.org/wiki/Java_Platform,_Enterprise_Edition for a list of existing Java Enterprise Edition application servers.

98

Technologies for the Wireless Future – Volume 3

JOnAS is an open source implementation of the J2EE specifications. JOnAS is a pure Java implementation of this specification that relies on the JDK. JOnAS is part of the ObjectWeb open source initiative, which was launched in collaboration with several partners. In Figure 4.9 the general JOnAS architecture is shown. JOnAS 5 is designed with an OSGi-based services architecture to provide a dynamically adaptable application server. JOnAS is the basis of a more global middleware suite delivered within ObjectWeb. It is the cornerstone of the service-oriented architecture, which includes other ObjectWeb projects like the workflow manager Bonita, the BPEL engine Orchestra and the portal eXo Platform. JOnAS provides an execution environment for Java-driven services and applications. In particular, Java beans can be designed and executed for and within a JOnAS server very easily. JOnAS is one of the few freely available application servers that are used in industrial environments. Extensions

Add-ons Java EE

Figure 4.9 The JOnAS execution environment [55]

JAIN SLEE A service logic execution environment (SLEE) is a well-known concept in the telecommunications industry. A SLEE is a high throughput, low latency event-processing application environment. JAIN SLEE is the Java standard for SLEE. JAIN SLEE is designed to allow implementations of the standard to meet the stringent requirements of communications applications, such as network signalling applications. The JAIN SLEE specification is designed so that implementations can achieve scalability and availability through clustering architectures. JAIN SLEE is the only industry standard aimed at portable communications applications, i.e. a communications application can be written once and run on many different implementations of JAIN SLEE. Application portability is made possible by the combination of a programming language API (specified using the Java programming language), an

Service Infrastructures

99

unambiguous technical specification, a reference implementation and a rigorous suite of tests that a vendor must pass before their product is compliant with the JAIN SLEE specification. JAIN SLEE is the point of integration for multiple network resources and protocols. Applications can use many different external network resources from within the JAIN SLEE environment. The JAIN SLEE specification allows developers to write robust components that integrate the well-known ACID properties of transactions into the programming model. Components can be composed to solve more complex problems. MUPE – Multi-user Publishing Environment (http://www.mupe.net) Multi-user Publishing Environment (MUPE) is a client-server type of application platform for mobile multi-user context-aware applications. MUPE is a Java-based implementation designed to support as many standard devices as possible, both in the client and the server sides. The overview of the MUPE application platform is presented in Figure 4.10. Operation The MUPE client subscribes to a MUPE server on the Internet and downloads functionality from it. The MUPE client uses a custom script language which encapsulates J2ME functionality. Most things that can be done with J2ME can be done with MUPE, with the exception of threads. MUPE services are developed with J2SE for server-side functionality and XML for UIs and client-side functionality. XMLs can be changed while the client server system is running, Wireless network

Client

MUPE Core

MUPE Server

End-user devices: mobile phones with J2ME

Middleware: connections

MUPE application functionality

Context External producers

MUPE Client

MUPE Core

Functionality: • All phone related functionality • Extension plugins

MUPE Server Functionality: world content • Service logic

Functionality: • Communications • External

• Virtual

Context

• Utilities

Packaged in: MupeClient2.jar

Packaged in: MupeCore.jar

Packaged in: ContentClasses.jar

Figure 4.10 MUPE application platform

100

Technologies for the Wireless Future – Volume 3

which means the client functionality can be changed at runtime. Due to this, MUPE enables very rapid development of services. The MUPE client uses standard protocols to communicate with the server. TCP is preferred, and HTTP can be used as well. TCP allows the server to push content to the client, which enables a richer set of services. Benefits MUPE allows anyone to easily create their own mobile services, games and applications. To achieve these goals, MUPE uses a plethora of technologies to reduce the complexity of developing mobile services. There are many areas in mobile services that affect the end user: ž It is difficult to install new applications to mobile phones. ž Online services require data transfer, which results in costs. From the developer’s perspective there are even more obstacles: ž Mobile applications are difficult to develop for many reasons, for example the development platform is different from the target platform and on-device debugging is fairly difficult. ž Each change in client software requires a new installation to the device to see how it works. ž Developing mobile online services requires a solid communication framework that is designed for mobile networks. MUPE is designed to overcome these problems as best as possible, using several techniques. For end users: ž MUPE uses a single J2ME client in all of the services, and the client can update itself once new versions are available. End users are required to do a single install only. For developers: ž Only server-side programming is required. Communications framework, client and the basic functionality do not need to be changed. ž MUPE is a complete virtual world, where development can start immediately on the service itself, not on supporting technologies. This saves a lot of time and effort. ž MUPE is completely open source, so the developers are free to modify any part of the platform and have access to all the code that is used (including the parts that do not need changes by developers). ž The developer tool plug-in for Eclipse offer instructions. ž Platform updates are easy – the communication framework is in a single jar file, and the base content is in another. When the platform is updated it is enough to update these files. For all: ž MUPE offers data compression – this does not make the system free to use, but reduces the data traffic, lowering usage costs. ž Graphics are stored in the phone on new service subscription (if the phone supports this), further reducing the need for data transfer.

Service Infrastructures

101

ž MUPE uses only Java (client uses J2ME MIDP 2.0, and server J2SE), making the system usable in most mobile phones and most back-end systems. ž MUPE is designed to operate through phone calls, SMS and other normal mobile phone behaviour. 4.3.3.4 Service Creation Environments Traditionally these environments are proprietary and product driven. They would have to be based on specific technology, like web services, to be generic enough, or to be able to automatically translate between different technologies and protocols. In the following paragraphs some examples are given of existing and used environments or approaches. Eclipse Eclipse (http://www.eclipse.org) is an open source extensible development platform, runtime and application framework for building, deploying and managing software across the entire software lifecycle. Over 60 open source projects are organized around six categories, targeting different application areas, among which is service-oriented architecture (SOA). With reference to our interests, the Eclipse platform can be used to develop service creation environments. There are several features in Eclipse and in the related projects that can be used in this context, primarily the GUI creation capabilities (see for example Visual Editor project (VE), a framework for creating GUI builders, http://www.eclipse.org/vep/). The capability to handle the web services and the composition of web services is provided by the SOA Tools Platform project (http://www.eclipse.org/stp/). Other interesting projects related to modeling are the Model Development Tools (MDT), which aims to provide an implementation of industry standard metamodels (http://www.eclipse.org/mdt), and the UML2 project (http://www.eclipse.org/uml2), an implementation of the UML 2.0 metamodel designed to support the development of modelling tools. The capability to create code for mobile devices is featured in the Device Software Development Platform (DSDP) (http://www.eclipse.org/dsdp), addressing a broad range of needs in the device software development. Graphical Service Creation Workbench The graphical service creation workbench (SCW) developed within the ITEA S4All project supports mobile service developers in putting together their service logic, based on state information and resulting in action triggering in the environment or the mobile device of the end user. service developers can simply ‘draw a service’, which can then be deployed for end user consumption. End users have to subscribe to these services and can personalize the behaviour. The SCW consists of five screens: ž ž ž ž ž

Project – creating new projects, loading saved projects Visually Edit Business Rules Management of State (Delivery) Building Blocks Management of Action (Triggering) Building Blocks Deployment – test and deployment of services within the SEE.

To create a new project, the first step is to discover and configure appropriate service building blocks.

102

Technologies for the Wireless Future – Volume 3

Figure 4.11 SCW State Service Building Blocks Wizard

State service building blocks get information about the context of a user, or about the state of components or hardware in a user’s environment (which can also easily be considered as being context). To add a new state service building block, the State Services tab offers the New State Service Wizard. Service building blocks can be found using a UDDI discovery or by specifying a service directly using its WSDL location. When using WSDL, the URL of any web service can be entered (see Figure 4.11). After this, the state service building block can be parameterized and is then available as a graphical block in the SCW GUI for composing rules, which is shown in Figure 4.12. One example of a service is shown. It is created by dragging and dropping service building blocks, facts, atoms and connectors from the left into the drawing area. The service can be saved using the ‘Save as’ button. Then it will appear in the list of services. The current rules can be tested under the Deployment tab by pressing the ‘Run inference once’ button, shown in Figure 4.13. Through pressing the ‘Deploy’ button, the service is deployed to the service execution environment (SEE), ready for users to subscribe to, personalize and use. This service creation workbench is a first approach where users can draw a service logic based on existing service building blocks that may reside in the vicinity of the user. OMA Support of Service Creation OMA’s high-flown vision is that ‘No matter what device I have, No matter what service I want, No matter what carrier or network I’m using, I can communicate and exchange information’. OMA develops market-driven service enablers by open specifications (requirements, architecture, technical and test) and encourages interoperability at the application level for communication services such as messaging, digital rights management, device management, mobile location services, mobile commerce and charging, etc.

Service Infrastructures

103

Figure 4.12 SCW Rule Editor

The service enablers work across different types of (IP) networks/devices such as fixed/Internet and mobile for different devices such as mobile phones, laptops and traditional PCs, so services should be network agnostic. OMA test specifications and TestFests should ensure interoperability, stability and quality in multi-vendor, multi-operator and multi-content provider environments. The goal is of course that the standards enable rapid deployment of new products and services leveraged on the service enablers. OMA does also support the use of the hidden complexity of underlying access and core networks of IMS to gain the SIP-based IMS capabilities and interfaces for session management, authentication/authorization mechanisms, routing, charging/accounting, mobility management, QoS control, etc. It includes a generic use case that shows how a service provider can develop an application using OMA-specified service enablers that are connected to IMS and use IMS capabilities in an interoperable way. OMA also develops an OMA Service Environment (OSE) [56] that describes the logical interactions between OMA enablers and interfaces to applications, service providers’ execution environments and network capabilities and resources, and a logical architecture for OMA enabler implementations. It is a conceptual environment that provides interfaces to applications (interfaces I0 or I0CP), interfaces to service providers’ execution environments (I1) and interfaces to invoke and use underlying network capabilities and resources for enabler implementations (I2). Of course, many applications use IMS interfaces and capabilities directly or use non-OMA enablers as well.

104

Technologies for the Wireless Future – Volume 3

Figure 4.13 SCW testing and deployment

Application Servers, Service Enablers, API Gateways Dh

Ut http proxy

SLF

Sh

ISC

S-CSCF

Ro OCS

Ut http proxy

CCF

HSS Gm

Rf

AAA P-CSCF

UE

Gm

UE

IP MM Subsystem Core Figure 4.14 OMA enablers utilizing IMS capabilities (http://www.openmobilealliance.org)

Service Infrastructures

105

Applications I0 + I1 Non OMA enab -lers I2

Non-IMS

IMS services

OMA OSE

e.g. - Messaging, - Conferencing

I2 IMS interface IMSCore

Figure 4.15 Relationship between OSE (with OMA enablers), IMS, non-IMS, IMS services and non-OMA enablers (http://www.openmobilealliance.org)

OMA also addresses how the OMA OSE and Parlay/OSA architectures and components (Parlay APIs, Parlay X, Parlay GW and SCFs) can be integrated and coexist for web services. This reference specification [57] should support different OMA service enablers that might use resources such as Parlay and Parlay X components and avoid overlaps of work. Benefits include exposing the interfaces to third parties no matter whether or not they are OMA enablers or Parlay/OSA components, which give more options and flexibility for third parties to develop their own service creation environments. However, there is still more standardization work to do to introduce the specific reference points/interfaces between the different OMA enablers such as messaging, device management, charging and a third party application based on web services as specified in [56] in order to satisfy the Parlay requirements. OMA service enabler constraints include that it takes a long time to develop the global standards and only the enablers are included in test specifications and tested on TestFests, not products or services. Existing Operator Initiatives There are at least three European operators that are experimenting with exposing some interfaces: ž Telenor Research & Innovation has been running PATS (Programme for Advanced Telecom Services) since 2001 in cooperation with the university NTNU, and over 300 services have been built by students, partners, etc. They’re now trying to open up even more by allowing end users to create non-commercial services by registering on their portal (http://www.pats.no). There is still some work left to do on automating the process, especially on ensuring the security and privacy (e.g. related to GSM location) of the telecom network that users have come to expect and which is regulated by law. There is a focus on open, standardized APIs as well as open source. Examples of enablers are SMS, MMS, WAP Push, GSM location, call control and device management. No SDK is available yet.

106

Technologies for the Wireless Future – Volume 3

ž British Telecom have opened up with their Web21CSDK (http://sdk.bt.com/), with the motto ‘Do Less: Achieve More’. They offer messaging, voice, location, authentication, conference call, profile and contacts. It is more a software developer kit approach, where they have tools for .Net, Java, PHP and Python. ž Vodafone has launched a similar initiative called the Vodafone Betavine, which supports SMS and WAP Push (http://www.vodafonebetavine.net/web/guest/projects/api). 4.3.3.5 Conclusion Most importantly, service providers must find ways to expose and open APIs for internal and third party service developers. The developers will find a way as long as the components and enablers are well tested, easily accessible and easy to leverage services on, and are supported by self-explained and good community sites. It will probably become more and more important that components and enablers support the convergence in a multi-vendor, multi-operator and multi-content provider environment. This is the mashup, which is a more informal service composition. Service developers often have strong preferences with regards to their service creation environment. For end users, a more user-friendly environment must exist, but this will of course imply more restrictions. 4.3.4 Business Models for Service Creation 4.3.4.1 Business Aspects Service business will be built around value proposition, customer relationship and business partners rather than around infrastructure and stakeholder activities alone. If the service providers are prepared to enter this path and provide their skills they could also find business opportunities together with other competent, cooperative and customer-oriented players. Anyhow, business models for service provision need to address mobility, network effects and closed wall arrangements. The business model must be attractive to the customer and to all other players. In the current communications markets, companies deliver services to customers in cooperation with other market players using a multitude of business models. Some of the main challenges for service creation are that: ž ž ž ž

Service consumption is subordinate to service development. Knowledge about the human is crucial. Interaction and interactivity are important building blocks. The device alone is under pressure of decreasing value. Customer value is created by the services. ž The backlash to IT and telecom has to do with complex value chains and infrastructures. It is about network logic: ž ž ž ž ž

The customer’s willingness and capacity are the prerequisites. Focus on the development process. Timing and rhythm. Tools for a holistic view on service development processes. Technology maturity.

Service Infrastructures

107

An important differentiation should be made between services and network solutions, which are operator-centric, and solutions, which are end user-centric. Users can have access to personalized applications and services on a basis where they are serviced by network and service operators, as well as by setting up the individualized applications themselves. There could be many different players involved in service creation and they could have complicated relationships. The market potentials for the different players will very much depend on how the value networks are constructed, i.e. the business models developed. For a network-centric model where there are few but financially strong users, infrastructure vendors and service providers will have a higher market potential than the other roles being executed. In a terminal-centric model, however, there could be individually low-intensity but all together high-mass usage. 4.3.4.2 Professional Service Creation The future services will be more personal and focused on individual needs. Customer relations will be the most important factor to the suppliers for their survival and growth. Specialization and focusing on different interest groups in order to increase service usage will be important in the future. Focus will be on service usage patterns. Future Service Requirements Many customers express their needs for better entertainment and information services but also for mobile ticketing and m-commerce solutions. Trust and security are very important features when it comes to financial transactions. This is also relevant when personal data is processed. Financial institutions have to participate to develop processes where these two features are of the utmost importance. A working relationship between infrastructure providers, service providers and security firms will be needed in order to provide secure transactions to users. Efficient customer services and help services, gained from past experiences, should become an integral part of a service provider’s business process. This service can also be seen as a means of channelling the customer’s requirements, needs and preferences to the service provider, which the provider can then use to improve services. Customer Service and Contact Centre New requirements on customer service centres will be caller satisfaction, increased wallet share and increased market share in order to reduce churn and to drive profitability. To create competitive advantage with customer service today, the service providers need to do more with less. So, how can they provide differentiated service at reduced costs? More and more service business interactions are over the Internet and this trend will continue for the foreseeable future. The Internet continues to be a challenge for most organizations. It cannot be ignored; it has already altered the customer communication landscape by adding dimensions such as web self-service, e-mail communication and live web collaboration. Web self-service is a trend that makes it possible to use necessary resources in a more efficient way and thereby drive cost savings and customer satisfaction. E-mail customer service could also help reduce service costs, increase productivity of the contact centre, improve service quality and enhance sales. In order to take advantage of this a special e-mail management system might be needed, since Outlook is not a sufficient tool for this purpose.

108

Technologies for the Wireless Future – Volume 3

General Aspects The acceptance of new services by users is something considered by service providers with the introduction of new applications or services. For new services, the market may be easily segmented into technology-inclined users (techies) and non technology-inclined users (nontechies). And the young-at-heart are an important segment, which might be coupled with the techies, but not necessarily. First, it is essential to form some guidelines as to what makes a technology-inclined user. Technology-inclined users are those who have daily contact with technology and have a reasonable understanding of how the technology works. They know what goes on behind each application or technology they use. They are not afraid to try new technologies and products. A service concept will impact the two groups of users quite differently. The first impact is on perceived usefulness. It is much easier for technology-savvy users to understand and to accept the usefulness of a new application than it would be for non technology-inclined users. The next impact is on ease of use. Nontech users have much higher requirements on simplicity. User interfaces and service application should be intuitive and easy. Behavioural intention to use the system is a major determinant of user behaviour. If users realise that their context has a large influence on their willingness to use the service then it is easier for them to overcome this obstacle. An additional impact on the model is the cost of technology and services. New technologies will require new devices for the users. New services will also add cost for the users. Most users would agree that a comprehensive and thorough examination of service needs and service quality provides an invaluable approach to improving service quality. 4.3.4.3 End User Service Creation Differentiated Market Needs Service and product differentiation are needed to address different corners of the market. Some users find certain services more important and useful than others. Also, to differentiate their service from other providers, it is possible that special or unique services will play a distinguishing role. With so much competition, price is one way to differentiate, and with schemes such as prepay or flat rates with data services, providers can differentiate themselves from others. Of course, other service differentiation methods exist, for example after sales service and customer care, or bundling of services. Individualization is a concept dealing with the integration of individual user preferences, roles, user location, context, and infrastructure and terminal capabilities. The system solution should be tailored to the individual user’s qualities and preferences. End users are concerned about their privacy rights. Aggressive marketing, spam, subscription sales, etc. based on user profile information has made subscribers more concerned about their privacy. Individualized solutions could sometimes offer important benefits over personalization. Critical technology issues for the individual end user that are taken care of by the service individualization concept are security, quality of service, system integration, accessibility and management of user profiles. New Types of Terminals, Networks and Services New terminals and their increasing capabilities make up a possible future threat towards the established value networks and their expected business. Tablet PCs like the Nokia N800, MS

Service Infrastructures

109

Origami, Sony Mylo, etc. are very capable platforms for Internet connections and web service utilization. There might be a need for a new operating system, which could offer a better platform for adding new features like federation support and context management, or graphical user interface (GUI) development. Moreover, to provide compatibility between mobile phones from different manufacturers it is crucial to have a common operating system. The processing and data storage capabilities of new devices are far greater than the capabilities of yesterday’s networks. There is loyalty-building mobile management software in the market that helps users to store and manage personal data, files, graphics, ringtones and to individualize their mobile phones by two-way transmission with their PCs. There is also an evolution of the SIM card, where the storage space can now exceed 1 GB. For many end users, simplicity is the key word for entering into the Internet world. The business potential for individualized services built on new terminal types seems to be very huge and is clearly a threat towards present network-based business models. End User Service Provisioning In the close future there will be more instances where the operator network is bypassed by the user, who will set up their own connection making use of ad hoc or peer-to-peer technologies. End user service creation is something that has captured a major interest during the last few years. Skilled people have started to make private music and TV broadcasting programs. This possibility has created new opportunities for subcommunities to express and develop their important interests and values. In the future it is likely that this development will more or less explode.

4.3.4.4 The Service Value Network In order to survive, the value network player beyond 2010 will have to face key issues such as: 1. 2. 3. 4.

Implications of a converged mobility–broadband environment. Business refocusing. Network and enabler sharing. Finding new and profitable business models.

So far there has been a focus on network assets as the primary competitive advantage, but now there is a tangible shift towards brands, organization and market channels. The main drivers of change for communication services will be to find values around mobile and broadband. The focus of the industry has to shift towards services and there will be an increased competition from players from other important industry sectors. The future stakeholders should derive values from intelligent edge applications and devices. Different kinds of communication enablers will have an increasing role in service provisioning. The Customer-centric Business Model The customer-centric model relates to the prosumer case, where the customer is a producer and a consumer utilizing valuable service and network enablers and aggregators.

110

Technologies for the Wireless Future – Volume 3

Billing & Transaction Operator

Home Equipment Billing and Payment

Revenue

Revenues

Content and Services

Service Enabling/ Provisioning

Customer

Network Provider

Security

Revenue Revenues

Device Manufacturer

Figure 4.16 Customer-centric business model

The Device Manufacturer-centric Business Model Device manufacturers today spend a large amount of their profits on research and development of new technologies and services. Together with mobile operators, the device manufacturers have created a well-established value chain in the provision of mobile services to users. Close cooperation between the device manufacturer and the mobile operator has led to new services being introduced by the network operator (such as the development of payment and multimedia services in Japan and Korea). The device manufacturer has traditionally sold devices mainly through the network operators. This sales arrangement has enabled network operators to package their services with the device and sell these to the users. It has also allowed the device manufacturer to save costs on setting up their own shops. This development is not likely to continue. Game console manufacturers offer solutions today that bypass the existing operator solutions. No subscription is needed, payment is direct and authentication is made by biometrics and other new methods. There are new short lifetime gadgets, dashboard and specialized terminals. The market has gone international. Home networks, residential gateways and wireless networks will be important ingredients of the device manufacturer-centric model. Flat-fee charging is used. The Service Provider-centric Business Model In the service provider-centric business model a network provider offers seamless access on a number of core and access networks. The service provider bundles this seamless access with a number of aggregated services. These services all run on the service-enabling platform in order to be integrated.

Service Infrastructures

111

Device Based Services

Billing & Transaction Operator Billing and Payment Revenue

Content and Services

Network Provider

Revenue

Service Enabling/ Provisioning

Device Manufacturer

Revenues

Security Revenues

Portals and Content Customer enquiries, SLA

Customer

Revenues

Customer Service & Support

Figure 4.17 Device manufacturer-centric business model

Billing & Transaction Operator

Value Added Services

Customer Service & Support Revenues

Billing and Payment Revenue Revenues Portals and Content

Portal & Content

Service Provider

Network Provider

Connectivity

Revenue Utility Devices

Revenues Revenues

Customer Device Manufac turer

Figure 4.18 Service provider-centric business model

Customer enquiries, SLA

112

Technologies for the Wireless Future – Volume 3

Web 2.0 business models belong to the service-centric category. There are two major business models being used for Web 2.0 services: ž Model 1 (Flickr, Skype) – the model is based on technical innovation, offering a compelling value proposition. Value-adding services are being offered as a premium for which the customer has to pay. Network effects help to drive the adoption and value of the service. Technology is important as an infrastructure tool, facilitating the business model. ž Model 2 (MySpace, YouTube) – the model is based on network effects created by a user base and user interaction. A community is built around content such as user profiles for MySpace or interesting blogs for Gawker. Sometimes the service providers syndicate their content to third parties like Google or Yahoo!. The Network Operator-centric Business Model For the network operator of today, one of the challenges is the convergence of different networks: fixed, mobile and wireless. Fixed network operators have been offering a wide range of services from networking to applications and software to clients in order to provide a more comprehensive suite of services and a one-stop-shop option. Vertical integration of services has played a part in the growth of the fixed network operator for some time now. With mobile operators moving into the same service offering, there are several things to consider: ž ž ž ž

ability to integrate different network types simple charging and billing differentiated quality of service trust and security.

Billing & Transaction Operator

Service/ Content Provider

Value Added Services Billing and Payment

Revenues Revenue Content and Services

Revenues Portals and Content

Network Operator

Portal & Content

Device Manufacturer

Devices

Revenue Utilities Revenues

Revenues Customer enquiries, SLA

Customer

Customer Service & Support

Figure 4.19 Network operator-centric business model

Service Infrastructures

113

Most often the network operator has a strong financial position and will be an active player when it comes to finding good positions in the future value networks and business models. With this financial strength, several important attributes for the network operator to further develop could be: ž ž ž ž ž

brand service differentiation simplicity trust and security customer service.

Findings The network operator faces a future full of new opportunities as well as threats, depending on the way they see it. Although the network operator’s core competencies are to build and manage networks, this will not be enough in the future industry where converged services and technologies will be integrated. Whichever area the network operator chooses to move into or to integrate into its present business, these features will continue to be needed. A good brand name not only catches the attention of potential users, it invokes a feeling of trust when users recognize it. Trust and security are particularly important features when it comes to financial transactions. This also applies when personal data is exchanged. A working relationship with security firms will be needed in order to provide secure transactions to users. Efficient customer service and help services, gained from past experiences, should become an integral part of the network operator’s business processes. This service can also be seen as a means of channelling the customer’s requirements, needs and preferences to the network operator, which the operator can then use to improve services. IPX payment, recently introduced by Ericsson, allows mobile subscribers to purchase digital content from a content provider regardless of where they reside or which mobile network provides the mobile service. Consumers can pay for content using their mobile phone and the existing charging relationship with their operator. 4.3.4.5 Business Evaluation of Service Creation According to [58], the business model can be viewed from four different aspects of service creation: ž Service design – description of the service offering, the value proposition and the target group. ž Technology design – description of the technical functionality required to realize the service offering. ž Organizational design – description of the network of different actors that is required to deliver the value services to the end users. Also the roles played by each actor in the network. ž Financial design – description of revenue that is intended to be obtained or earned from the value service. It includes risks, investments and revenue division amongst the different actors.

114

Technologies for the Wireless Future – Volume 3

Business Issues in the Service Domain The central issue for the service domain is value. Value is seen as the perceived benefits and total costs of a product or service for customers in target markets. Services could be delivered on a number of different networks. Examples of business issues that originate from the service domain are customer targeting, creating value, branding, trust and customer retention. An important issue in almost every use case is choosing a profitable target group. Should the service offering be targeted towards consumers or business? Should the service offering be targeted towards a niche market or a mass market? Closely connected to choosing a target group is formulating a compelling value proposition for end users. The added value of a service can be based on value elements such as: efficiency, accuracy, speed, personalization, trust, etc. Another important design issue is how to reach customers. An important mechanism to reach customers is branding. Establishing end user trust is often a balance between security and ease of use. Customer retention refers to marketing strategies aimed to keep customers satisfied with and loyal to the service offering. Business Issues in the Technology Domain The main business issue in the technology domain is functionality. Functionality can be defined as the things a system or application can do for its end users. The technology design describes the technologies involved and the technical architecture in which the technologies operate in the provision of a product or service. There are several ways in which the technologies may be classified, for example according to seven layers of the OSI, or according to technology in hardware or software. The technological design variables are all important components of the technological design that will contribute to the technical functionality of the system. Each component, application, device, service platform, access network and backbone infrastructure makes up the technical architecture and describes a part of the technology design. Important business issues that originate from the technology domain are security, quality of service, system integration, accessibility and management of user profiles. Business Issues in the Organizational Domain Business issues in the organizational domain have to do with how the value network is organized and controlled and how the third parties and end users are given access to network resources and capabilities. Network complexity will imply a need for many partnerships and some conductor will have to manage this partnership network. Figure 4.20 shows the activities that take place in the organization design. Each actor in the value network is different. They will all have their own strategies and goals, and resources and capabilities. These in turn will determine how much they will contribute to the product of service. The different contributions that are required by different actors will determine the value activities that they have to perform in order to deliver the service or product to the end user. Business Issues in the Financial Domain In a multi-vendor supply network the individual partner’s profitability and risk must be dealt with. There should be a fair division of investments, business risks (risk sharing), costs and

Service Infrastructures

115

To participate in

Strategies and Goals

Consists of

Have

Resources and Capabilities

Have

Relations

Have

Actors of the Value Network

Perform

Are used in

May grow into

Consists of

Value Network

Interactions

Negotiate/Enforce

Value Activities

Define

Define

Organisational Arrangements Define

Combine to

Roles

Figure 4.20 Organizational design (adapted from [58])

revenues between the different actors. Pricing of connectivity, network services and end user services are also of common value network interest. The most visible form of revenue is from usage by customers. Revenue, however, can also be generated from government subsidies or tax breaks. Advertisements are another source of revenue. The technology design will result in costs in capital expenditure and this will be another cost input to the finance design. The finance design is the bottom line of the chain, therefore any risks taken or revenue generated will affect the finance design. 4.3.4.6 Conclusion If we compare the different business models we find for the service domain, see Table 4.4. Table 4.4 Business issues in the service domain Business Model

Service Domain Effects

Customer-centric Model Device Manufacturer-centric Model

Many small markets with limited value. The market is defined by the device(s) and the services belonging to them. The service offering defines the market. The market is restricted by access to infrastructure and service restrictions.

Service Provider-centric Model Network Operator-centric Model

116

Technologies for the Wireless Future – Volume 3

: : : for the technology domain, see Table 4.5: Table 4.5 Business issues in the technology domain Business Model

Technology Domain Effects

Customer-centric Model Device Manufacturer-centric model Service Provider-centric Model

Functionality may be restricted by the technical architecture. The user could be locked in by the device technology. Innovative technological solutions like Web 2.0 could be the base. Service platforms ensure interoperability between services. Technology shapes the service offering.

Network Operator-centric Model

: : : for the organizational domain, see Table 4.6: Table 4.6 Business issues in the organizational domain Business Model

Organizational Domain Effects

Customer-centric Model Device Manufacturer-centric model Service Provider-centric Model Network Operator-centric Model

Could be a prosumer society. The middlemen are eliminated. The model opens up for a huge number of partnerships. The network operator has created a lock-in situation where competition determines the customer offering.

: : : and finally for the financial domain, see Table 4.7: Table 4.7 Business issues in the financial domain Business Model

Financial Domain Effects

Customer-centric Model Device Manufacturer-centric model Service Provider-centric Model

Revenues and costs are distributed to the edge. Revenues and costs are distributed to the edge. Brand strength and network effects are essential when creating profitability. Could open up for inefficient long-term investments.

Network Operator-centric Model

4.3.5 Future Research and Development It has been shown in this section that there are already many approaches to service creation available. However, all these approaches are either technology specific or locked down to a very vendor-specific solution. As a matter of fact, service creation is at the moment proprietary and therefore mostly not efficient in a wider sense. Even worse, in many companies new services are often developed from the scratch, as the in-house tools are not up to date or refer to an older technology approach.

Service Infrastructures

117

As can be seen from the approaches mentioned in this section, the listed solutions for service creation do not share a common base or even common definitions. Therefore it is necessary to do further research in the following topics, as this may lead to a more coherent approach to service creation: 1. In a technology-agnostic way, how to specify and describe services – successful services are driven by user needs and not by any specific technology. Therefore it is important to understand and to specify a desired service independent of the given technology, and to figure out in a later stage how to map this specification to selected technologies. Technologies may change over time and the need for specific services remain unchanged. Therefore it is important to specify the service in a technology-agnostic way. The presented approach in this section (based on activity theory) is an initial approach in this direction but is not a complete solution so far. Furthermore, the existing and upcoming tools do not take this view into account, but rather start at a certain technology-driven level. 2. How to enable end users to do service creation – end users are being empowered by content creation mechanisms in the upcoming trend of Web 2.0. With new services in the Internet, everybody is able to create their own personal content and to share it with interested communities. This was unthinkable several years ago, when the Internet was simply a research tool or useable by technology addicted people at best. Looking to upcoming trends like Yahoo! Pipes, and motivated by the business models presented in this paper, it becomes clear that end user-driven service creation is the next logical step that will happen. Of course it will not be as sophisticated as professional services, but it will give end users the possibility to differentiate themselves. Therefore developers should investigate what is needed to support and help the end users to find an easy and intuitive way to create services, not only for and within the Internet, but for all environments which they might use. 3. How to harmonize service creation approaches – looking to the different vendorand project-specific approaches, it becomes clear that no common approaches, or even definitions or enabling frameworks, are used. Therefore interoperability or even cross-technology service creation is not possible at all. In order to achieve this, a harmonization between different tools should be investigated. This can start by using the same kind of enabling frameworks, like the OMA enablers or Parlay X API, and can lead up to the reuse of common semantic definitions. But in order to make service creation tools more widely useable, a certain level of interoperability needs to be achieved. 4. Usability enhancements of devices: easier menu navigation, touch screen support, visual enhancements – different segments have different usability requirements so there must be different software packages for segments based on services (sports, music, TV, video, payment, etc.), features, look and feel, partnerships, etc. 5. Media support: TV, radio, music player, etc. – multimedia applications will be important in the future. Memory will gradually increase in devices and this drives the content business. 6. Web 2.0 support: widgets coming to devices using a web browser core, enabling faster service development using the same tools as on the Internet side – rapid service development is needed. Service development should be enabled with the same tools as currently used on the Internet side. This will also bring several innovations and mashups. 7. Mobile payment and ticketing: remote transactions – user experience and behaviour begin with person-to-person payments (there are already several companies doing this) and secure mobile authentication to bank services.

118

Technologies for the Wireless Future – Volume 3

8. Device and home domain interoperability: ecosystems building around Google, Microsoft, etc. – Google has been building ecosystems around e-mail and IM (Google G-mail), media and entertainment (YouTube), blogging and social networking (Blogger). 9. OS development – Linux and Microsoft Mobile OS are gaining momentum, and platform development should be based on such foundations. Java applications will be strong in the low–mid-range phone, since interoperability is getting better there all the time. This list of research topics is not exhaustive and should be completed as needed. But as service creation is an area of increasing importance for all stakeholders (even the end users) it should be figured out how more consistent and easy-to-use approaches can be developed. 4.3.6 Conclusions Service creation is the next big end user and Internet-driven hype in the technology arena. As the first Web 2.0 services are already enabling technology addicted people to build their own composite services this trend will lead to a huge set of new features and services provided by all stakeholders. In order to support and boost this trend a more harmonized and common approach towards service creation is needed. This holds for the semantic-driven understanding of the services themselves as well as for the tools necessary to make the creation and composition happen. In this section we have highlighted a first approach towards a technology-agnostic way of understanding and describing services as well as different kinds of existing approaches and technologies. Furthermore, it has been clarified which business models can and will drive the service creation process for the different stakeholders. As there is only a minimum of end user-driven service creation (within Web 2.0) and telecom operators are still in the process of finding secure ways of opening their networks to the extent the Internet community is expecting, there are still a lot of issues that need to be researched and harmonized in order to enable easy and seamless service creation for everybody and for every network. However, the benefit of converging networks could be well worth the effort. We propose to start an in-depth research activity for a more harmonized approach to service creation, its semantic and technology-agnostic understanding, and a more user-friendly way of supporting the different stakeholders in their innovative task of building new services. The described interfaces, like Parlay X and OMA enablers, have been standardized over several years and used by vendors and third parties, so it’s due time to get them ‘out there’ and measure the market response of end users.

4.4 Service Architecture for the Wireless World Today’s wireless and mobile services are typically monolithic and centralized in nature, and service access is tied to a single device or delivery channel. There is a growing need for supporting heterogeneous service access and sharing service usage experience. Envisaged new sources of revenue for internet service providers include tailored, personalized and dynamically-composed services. The key requirements for service platforms include achieving faster time to market, cost efficiency and compelling user experience. This mobile service environment can be categorized into two parts: service access and service provision. Service access pertains to a consumer’s device and environment, and the

Service Infrastructures

119

possible interactions with a service. Service provision focuses on platform-side operation and includes mechanisms for service creation, lifecycle management, reconfiguration and composition. Key requirements and scenarios for future wireless networks are presented in earlier releases of the WWRF Book of Visions. Here, we build on these requirements and consider the different concepts and solutions needed to meet them. The key requirements are: heterogeneous networks, mobility, composition, security, privacy, dependability, backward compatibility and migration, network robustness and fault tolerance, quality of service, multi-domain support, accountability, personalization, context communications, extensibility of network services provided and application innovation. Here both service access and service provisioning are addressed, and solutions are considered for the above challenges. This chapter is structured as follows: Section 4.4.1 presents a brief background and motivation for the service architecture. Section 4.4.2 considers the key requirements for the service architecture, and Section 4.4.3 presents the key enabling technologies. Section 4.4.4 examines functional architectures and service delivery frameworks, and Section 4.4.5 considers interfaces. Section 4.4.6 presents groundings and exploitation, and Section 4.4.7, as an example application area, considers scenarios and architecture requirements of vehicular communications. Finally, Section 4.4.8 is a short conclusion. 4.4.1 Background 4.4.1.1 Motivation New sources of revenue for mobile service providers are expected to include tailored, personalized, context-aware and dynamically composed services that are fast to market, cost efficient, and offer compelling user experience [59], [60] to enrich the users’ lives and help to make them more efficient, i.e. empower the users in their activities. Despite the good availability of broadband radio technologies, the business pertaining to mobile services has not yet reached its full potential for a number of reasons, which include the following: ž Very long time to market for new services, due to lack of suitable service creation environments and a vertical design approach. ž Integration and deployment costs are too high due to the inherent complexity and heterogeneity of service execution environments. ž Service provisioning involves more and more parties – telcos, content/service providers, third party networks and service providers, and even end users, thereby increasing the complexity of the environment in which services must live. ž The diversity of the mobile terminal platforms and the capability differences of the mobile devices decrease mobile service penetration. ž Users are surrounded by many access technologies but often cannot handle the complexity of accessing their services. In many companies that develop complex high-technology products, the renewal of the product architecture has been a very painful process. In many cases, successful technology assets that are valuable in the beginning become obstacles later in renewing product development processes. Legacy gradually starts to take over the decisions and real changes are not made until a deep enough crisis forces the action. Many companies have done well so far and

120

Technologies for the Wireless Future – Volume 3

Table 4.8 Shift in device architecture priorities Current

New

Cost and silicon footprint Low power designs Form factor and related mechanics Maximal integration One size fits all optimized performance One size fits all feature list

R & D agility and speed Mastering horizontalization Overall cost efficiency Modularity Form factor Efficient system level power and energy management Performance and feature wise flexibility and scalability

current device architectures are reflecting well the past values. R&D execution has succeeded properly in its task. However, when looking to the future challenges and required changes in device architecture, we need to understand the changes and disruptions in our priorities before jumping to select different technology options (illustrated in Table 4.8). This shift is driven by digital convergence, with an increasing number of different products with growing complexity and varying functionalities. Multiple functions, having different R&D cycles, require modularity and flexibility. Modularity drives horizontalization, horizontalization enables open innovation and finally we end up at some level of industry standard. Indeed, standards play a vital role in defining the structures and interfaces of service architectures. In order to meet the presented challenges, new standards-based mobile service architectures are needed that allow more flexibility in service development, deployment and usage. Currently, the most interesting service platform functions include: ž Converged communications environment. ž Identity provisioning, allowing third parties and visiting networks to assert the identity of a user and their subscriptions while keeping the anonymity of end users. ž Charging and billing, allowing users to easily pay for connectivity, services and goods while allowing the operators to take a small premium from each service. ž Improved service provisioning (e.g. through push technologies, search engines or context-aware service offering) and service access. ž Personalization and adaptation to specific situations, making mobile and converged services easier to use. ž Virtualization, which abstracts hardware functions from software. ž Mediation and brokering, supporting the creation of service bundles. This section addresses service architecture for the wireless world that includes the above functionality. 4.4.1.2 State-of-the-art Analysis New technologies and emerging standards such as SIP [61], IMS (IP Multimedia Subsystem) [62], [63], OMA (Open Mobile Alliance) [64], OSA (Open Service Architecture) [65]/Parlay [66], and web services are crucial building blocks for mobile service architectures that

Service Infrastructures

121

support converged communications. In this section we briefly describe the key technologies and standards and then discuss challenges for the mobile service architecture. The session initiation protocol (SIP) (RFC3261) is an ASCII-based, application-layer control protocol that can be used to establish, maintain and terminate calls between two or more end points. The SIP client is a network element that sends SIP requests and receives SIP responses. Clients interact directly with a human user or a terminal device. Proxies and user agents are clients. The SIP server is a network element that receives requests and sends back responses to requests. Servers include proxies, user agent servers, redirect servers and registrars. The proxy servers handle message routing, authentication and authorization of SIP clients. A location server is typically used to keep track of the current location of SIP clients. The SIP framework supports four different types of mobility [61]: session mobility allows a user to maintain a media session while changing terminals; terminal mobility allows a device to move between IP subnets while continuing to be reachable for incoming requests and maintaining sessions across subnet changes; personal mobility allows the addressing of a single user located at different terminals by using the same logical address; service mobility allows users to maintain access to services while moving or changing devices and network service providers. These different types of mobility support also apply for event delivery and subscribers receive notifications even when they are roaming. Mobility support is realized by updating any client address changes to their respective servers. SIP supports personal mobility by using the forking technique. OMA are one of the key standards organizations for mobile systems and they have defined a reference architecture called the OSE, which includes critical mechanisms pertaining to security and service provision. These mechanisms include the following: authentication, authorization, federated identity, single sign on (SSO), accounting and charging management, and service provisioning and monitoring. The OSE model consists of service enablers, a policy enforcer, the execution environment, and applications. Applications use enablers by using service requests. The model defines how requests are handled and how various security policies are applied. The IMS standard [62] defines the functional architecture for a managed IP-based network. It aims to provide a means for carriers to create an open, standards-based network that delivers integrated multimedia services to increase revenue, while also reducing network CapEx and OpEx. As the recent activities in ETSI TISPAN (Telecoms & Internet converged Services & Protocols for Advanced Networks) have shown, the IMS standard has also become increasingly popular with wireline service providers and is now considered the de facto standard for fixed-mobile convergence (FMC) in next generation networks (NGN) [67], [63], [68]. The IMS architecture has been designed to clearly separate the connectivity, control and service plane. IMS decomposes the networking infrastructure into separate functions with standardized interfaces between them. Each interface is specified as a ‘reference point’, which defines both the protocol over the interface and the functions between which it operates (refer to [69], network architecture). The 3GPP architecture is split into three main planes or layers, each of which is described by a number of equivalent names: service or application plane, control or signalling plane, and user or transport plane. The service plane (application plane) provides an infrastructure for the provision and management of services, and defines standard interfaces to common functionality (e.g. configuration storage, identity management, billing, presence and location). The IMS provides both convergence at the session/control layer by using common session handling mechanisms, and convergence at the application layer by offering all services to any

122

Technologies for the Wireless Future – Volume 3

Figure 4.21 Convergence on application and session/control layer based on IMS

type of device regardless of the used access and core network (see also Figure 4.21). The central components of the IMS system include several SIP servers or proxies called CSCF (call session control function), the HSS (home subscriber server), the MRF (media resource function), BGCF (breakout gateway control function) and application servers (AS). Interfaces are defined between different entities, for example the Gw interface is used to exchange SIP messages between the user terminal and the CSCFs. The control plane comprises network control servers for managing call session setup, modification and release. It is located between the application and transport plane and routes the call signalling, tells the transport plane what traffic to allow, and generates billing information for the use of the network. The connectivity plane comprises routers and switches, both for the backbone and the access network. While the IMS is acknowledged to be a main part of the next-generation network architecture, it is also necessary to improve the service plane to allow fast and easy provisioning of a large set of services. 3GPP has specified together with ETSI an open service access (OSA) framework, which shall give any application easy access to network-related data and functions without the need to know all internal details of an operator network. The API for OSA is called Parlay (or OSA/Parlay) and is developed jointly in collaboration by 3GPP and ETSI and distributed via the Parlay Group. The current version, Parlay 5.0, was developed in cooperation with a number of Java APIs for Intelligent Networks (JAIN) Community member companies. The aim of Parlay/OSA is to provide an API that is independent of the underlying networking technology and of the programming technology. The set of mappings to specific technologies includes CORBA/IDL, WSDL and Java. An important role of the Parlay/OSA framework17 is to provide a way for the network to authenticate applications. The framework allows applications to discover the capabilities 17 http://www.parlay.org.

Service Infrastructures

123

of the network, and provides management functions for fault tolerance and overload. The Parlay service APIs allow applications to make telephone calls, query the location of an entity, and charge for provided services. An IMS OSA-SCS (open service access service capability server) interfaces OSA application servers using Parlay. In addition, a subset of the Parlay API is provided through web service-based interfaces in Parlay X. This newer specification defines a set of simple telecom web services, including third party call control, SMS and messaging functions, charging, and location and user status. The JAIN18 initiative has defined a set of Java APIs for the development communications products and services. The JAIN service logic execution environment (JSLEE) specifies the Java-based service platform for executing event-oriented applications. SIP is supported through the JSIP API and Java Call Control (JCC) API. Web services can be seen as an interoperable and easy way to deploy customized service elements also for mobile clients. The world of mobile platforms, however, is considerably different from the traditional web services execution environment within the fixed network. In the current and forthcoming mobile systems, users are able to access services through different heterogeneous access networks. The service access should be adapted by the service platform and customized based on the terminal, user preferences and network capabilities. Basically, a mobile platform on a mobile client can vary between a thin client, for example a web browser with no additional functionalities and capabilities, and a self-contained, thick client, similar to, for example, a laptop, with quite some CPU power and memory, as well as software capabilities. Both approaches have pros and cons, and different degrees in between are possible. For example, a web browser solution could be an interesting approach to harmonize heterogeneity of underlying platforms such as different operating systems, whereas for many advanced applications, like the ones relying on context information obtained from sensor and other information of the OS, or for disconnected operation, a pure web browser is not sufficient. One important solution for mobile platforms, which is currently being intensively investigated and developed, is IMS, which can be extended to support web services [70]. The integration of the above technologies into coherent service platform architecture is a current topic and an integral part of IMS evolution. A number of key questions regarding the converged communications environment include the role of IMS, how web service access is realized, how security is implemented, and how component services are invoked seamlessly. From the security point of view, a key challenge is how to leverage the existing authentication, authorization and auditing features of mobile service platforms. 4.4.2 Key Concepts In recent years new service platform concepts have been introduced, such as autonomic communications, semantic matching, system reconfiguration and cognitive systems. These concepts are part of service platform evolution from the client-server model towards a more flexible peer-to-peer model. The notion of SOA (service-oriented architecture) has received significant attention within the software design and development community. SOA is a paradigm for organizing and utilizing distributed capabilities that may be under the control of different ownership domains. SOA can be seen as a stepping stone towards flexible and decoupled service management. 18 http://java.sun.com/products/jain/.

124

Technologies for the Wireless Future – Volume 3

In this section, we briefly consider key service platform concepts and aspects. Our focus is on system capabilities and requirements. 4.4.2.1 Autonomic Systems The introduction of autonomic aspects in service architectures is the key to minimizing human intervention, thus reducing management complexity. Autonomic communication systems are adaptive networks, governed by human-specified goals and constraints on the network services behaviour. Furthermore, reconfiguration provides the necessary mechanisms to facilitate for service adaptation, by utilizing service-specific knowledge that may be required. 4.4.2.2 Peer-to-peer Services Support Peer-to-peer services are directly executed between two or more end users acting as service providers, without the need of the core network service infrastructure. Whereas a strict definition of such peer-to-peer services excludes the (centralized) service provider or operator from service provisioning, in some cases additional service provider or operator services are required to ensure service execution, e.g. negotiation, contract enforcement, etc. For next-generation mobile service platforms, it should be regarded as a general requirement to support peer-to-peer services in order to exploit ubiquitous communication networks such as ad hoc networks. 4.4.2.3 Service Adaptation Service adaptation involves certain decisions regarding reconfigurations based on the environment and the corresponding knowledge about the network, the devices and the user. These goals are mapped to appropriate policies at a system level and are subsequently enforced. Autonomic decision making functionality takes into account defined policies and contextual information, which is represented with the use of ontologies and specifies the prioritization of policies in order to define certain reconfiguration actions in an autonomic way. It also caters for the implementation and enforcement of the decided reconfiguration by firing certain actions, such as: ž ž ž ž

Change of the environment in which the application is executed. Change of the content consumed by the executed application. Change of the applications interactions that compose the service. Change of the software components that define one or more applications.

Today, there exist a great variety of mobile devices, such as mobile phones and PDAs, that users can employ to access services. The devices are heterogeneous with respect to their capabilities to handle input and present user interfaces or the media they support. Additionally, network capabilities are also changing when it comes to using mobile devices to access a service. In consequence, the user interface and service adaptation capability is needed to properly handle these discrepancies and provide the best user experience. It provides functionalities to allow service developers to make services available through one or more device using multiple modalities.

Service Infrastructures

125

4.4.2.4 Personalization The personalization capability provides profiles and preferences of users and groups to the services, applications and components, and supports learning of user and group interests as well as user and group preferences. To support user feedback for profile learning, it is directly related to the services and applications. The personalization capability consists of different subcapabilities, such as profile manager and recommender. The former groups all personalization data in terms of the identified data model classes and allows querying of all personalization data that are needed in a specific interaction context. The latter supports learning of situational preferences and interests by means of application-independent learning mechanisms (content-based, collaborative or hybrid approaches) and provides recommendations based on this information.

4.4.2.5 Group Support Applications and services dedicated to groups are supported by providing a description of the operational group context, managing groups and managing group preferences and information. Group support consists of two parts: group management and group evolution. The former supports the managing of groups, i.e. creating, deleting, updating, managing lifetime; stores and dynamically updates templates for groups; and provides mechanisms to ensure trustworthy and private communication within and towards the groups against threats. The latter monitors group behaviour to learn new or modify existing group templates, which are used for the group creation.

4.4.2.6 Context Awareness The context awareness [71], [72] capability takes care of raw, interpreted and aggregated context data. This capability handles context data related to individual users and to groups of users. It supports the service developer by providing users’ and groups’ current context information through well-defined interfaces. New context information and changed context information can be notified to interested components and application services, and context information can be requested from the context awareness capability. The context awareness capability contains, for example, the following sub-capabilities: personal context, group context and context management. Personal context receives raw data, user behaviour, service discovery data and application requests from the user interface and service adaptation capability and directly from services and applications. Group context produces interpretations of the environmental context of the group, the possible context the group is in and the possible contextual personalized control actions to be performed. Finally, context framework specifications are needed to describe the representation, exchange, interpretation and prediction of context, as well as upper-layer context reasoning.

4.4.2.7 Privacy and Trust Services and applications deal with data related to the user, which raises the issue of trust and privacy of the personal user data.

126

Technologies for the Wireless Future – Volume 3

Trust can be defined as a set of assumptions about the integrity, strength, ability, surety, etc. of a person or thing with respect to some set of sensitive actions. It is assumed either that this party will do some actions or that they will refrain from doing some actions. The nature of these assumptions varies depending on the context in which trust comes into the picture. For example, in the case of a commercial transaction, the trust assumption would be that the other party in the transaction will pay for the service, while in a computer communication setting, the assumption would be about the identity, reliability and availability of the system with respect to some expected criteria. Privacy, like trust, is an abstract term that defies an easy definition. Privacy is a personal, subjective condition that varies from person to person and hence any system that purports to implement privacy features for the user must have features that let the user define and control how their privacy is handled by the system. One reasonable definition of privacy is that the user is able to control the personal identifiable information released about themself according to their own interests and values, within the limits of existing regulations. It is often not just one user of the service sharing personal data, as is the case with traditional single user-application interaction; often groups of users can be involved. The trust and privacy management of personal user data needs to be supported by a suitable policy system. The privacy and trust capability contains, for example, the following subcapabilities: privacy policy storage and management, and trust engine. The former deals with the management of user- and group-related privacy policies. The latter ensures that the system can be trusted to perform as is declared in the policies. The ability of a person to exercise fine control over their personal identifiable information is the feature that a privacy-preserving system must offer. But such a feature or guarantee is rarely offered by current platforms, and in the form discussed here it is still not required by regulation – even by the telecom regulation. The problem is that as privacy-sensitive data becomes electronically available, it is technically easy for it to become available to other individuals and organizations. ‘Data privacy’ or ‘information privacy’ is provided if, where data is electronically processed by another party, the individual is able to exercise a substantial degree of control over that data and its use. This is most important when a communication system offers users services that make use of their personal information. 4.4.2.8 Service Usage and Provisioning The service usage capability covers all aspects related to the service usage; in particular it covers every step in the timeframe between service discovery and service offering. The service provisioning capability holds a repository of services known to the system, their descriptions and properties and offers functionalities for service discovery, proactive service provisioning and service composition. These two capabilities have a very strong relation to each other. The situations in which users are provided services are derived through analysing the current context of the users. This way, a certain location can trigger services, for example a bus stop triggers a service to buy tickets. This functionality also has to take care of selecting the services that best match the user’s needs in certain situations. 4.4.2.9 Operational Management and Service Lifecycle The operational management capability supports the management of the whole lifecycle of services. To perform this, it needs access to information about services stored in the service

Service Infrastructures

127

provisioning capability, and access to applications and application services, and to the user agents running on users’ devices. To manage the privacy aspects, it also needs access to the privacy and trust capability. 4.4.2.10 Charging and Billing Accounting is the process of collecting details of resource usage for the purpose of billing a customer. The resource usage consists of users’ activity while accessing networks’ resources, like the amount of time spent in the network, the services accessed and their QoS while there, how often specific content is accessed, and the amount of data transferred during the session. After charging-related details have been collected, the accounting data is processed in order to charge the relevant business entities. OMA defines two charging interfaces: offline and online. Offline charging is a mechanism in which charging information does not affect the service rendered. The online interface allows realtime charging. Both charging interfaces can be categorized into two distinct charging submodels, namely the event-based charging model and the session-based charging model. In the event-based charging model, each service usage is reported with a single charging record or a resource usage authorization procedure. In the session-based charging model, the same service usage occurs within an end user session. The service usage is reported with several charging events and the creation of one or more charging records in offline charging, or performance of credit control session in online charging. One of the challenges is how to support electronic payment, especially micropayments in P2P environments. The network-centric charging and billing enablers may not be available in this case. 4.4.2.11 Service Layer Mobility Support Service layer mobility capability comprises the following five sub-capabilities: terminal, personal, service, profile and session mobility support. Terminal and personal mobility are already available in today’s networks; the realization of service mobility, profile mobility and session mobility is still a prerequisite for a next-generation service platform in order to provide services in heterogeneous access networks and on diverse devices. 4.4.2.12 Negotiation Support Multi-domain, multi-provider environments (ranging from BAN to Internet) will require advanced mechanisms to negotiate about required QoS and respective service parameters, cost models, permissions, etc. Consequently, negotiation can be found on all layers as part of end-to-end reconfigurable systems. 4.4.2.13 Role-based Service Management The concept of roles is frequently used in modern information management systems for structuring data and controlling access to it. In what follows, we shall first briefly describe the general concept before moving on to describing its application to service lifecycle.

128

Technologies for the Wireless Future – Volume 3

A role can be defined generally as being associated with a specific set of tasks as well as a specific viewpoint to information. A role does not need to have 1:1 mapping to job descriptions. A specific role can be participated in by multiple job descriptions, and – conversely – a job description may participate in multiple roles. It could be said that a role provides a task-oriented abstraction layer between job descriptions (subject to organizational changes) and information that is managed. From the viewpoint of organization of information, a role can be associated with a specific view, including both static data and dynamic aspects such as activity diagrams and processes. This principle is illustrated in Figure 4.22. The concept of roles lends itself well to be used in describing service lifecycles. Such an analysis has been carried out by TeleManagement Forum (TMF) for the creation of telecommunications services [73]. Even though the TMF work covers only a part of the entire service lifecycle, it contains useful classes of role, some of which have a wider scope of applicability as well. These include: ž ž ž ž ž ž

business analyst role solution designer implementer connectivity engineer project manager customer care

Role-specific views of information can be defined for an instance of the role using suitable frameworks. In the case of TMF, the enhanced Telecom Operations Map (eTOM) process framework [74] and Shared Information/Data (SID) information ontology [75] have formed the basis for this. 4.4.3 Enabling Technologies In this section, we present a number of enabling technologies that realize concepts presented in the previous section. First, we consider the distributed communication sphere, which deals

Role

View

View Role

Data

Figure 4.22 An illustration of role-specific viewpoints of information

Service Infrastructures

129

with the facilitation of spontaneous ad hoc communications. We then consider service composition, semantic enhancement and intelligent enablers. Later, we present open and controlled access, a privacy management mechanism, and IMS-based identity management. Finally, we consider device-specific aspects. Several of the enabling technologies described below were devised in the framework of the IST-FP6 SPICE (Service Platform for Innovative Communication Environment) project. SPICE addresses the problem of designing, developing and putting into operation efficient and innovative mobile service creation/execution platforms for networks beyond 3G. The purpose of SPICE is to research, prototype and evaluate an extendable overlay architecture and framework for rapid creation and deployment of intelligent and personalized mobile communication and content & information services. 4.4.3.1 Distributed Communication Sphere It appears more and more that mobile communication should not merely target one end user using one mobile terminal connected through one network/radio interface. Instead, the user will be moving in a sphere in which multiple access technologies will provide the connectivity for multiple devices that together deliver the service the user may wish to use. This involves many devices and many distributed service-enabling elements and forms a person’s distributed communication sphere (DCS). To deliver the complex services of tomorrow, the DCS will need to have the inherent possibility of accessing/connecting via a number of different access technologies and access network umbrellas. In a nutshell, a person’s DCS will contain: ž All terminals, gateways and devices that can participate in and contribute to a person’s mobile connectivity and communication means (e.g. cell phones, PDAs, car displays, laptops). ž All available and permitted communication technologies and channels (access network). ž The whole direct communication environment, including people, devices, services, resources, etc. ž All services as well as all system and other information that can be used in the context of a communication act. The DCS is subject to frequent changes as the user moves. This basic feature implies that communication devices may be only temporarily associated with a user and the connection types as well as devices and their capabilities are changing all the time, thus making the configuration of the DCS highly flexible and changeable over time. Some of the most important enablers are discovery, analysis and optimization of the direct user environment in order to represent its abstraction in the form of a communication model (see Figure 4.23). 4.4.3.2 Service Composability in a Loosely Coupled Approach During service creation, services will be composed of a collection of service enablers or any other existing services. A service itself can be realized as a component that in turn is composable again. This property will enable the creation of complex services based on previously created ones. It will ease and speed up the service creation process by providing reusable service components that can be embedded into new services. Advanced service description languages will allow the service architect to refer during service creation to actual, well

130

Technologies for the Wireless Future – Volume 3

User User context & preferences

Terminal & device

Gateway

Terminal & device

Gateway

e.g. UMTS

e.g. ADSL

Access Network: Bluetooth, WiFi, …

Core Network: UMTS, IP backbone, …

Distributed Communication Sphere

Core Network

User rules

Communication model

Buddies

Preferred services

personal Information/ user profile data

Relevant context information

SPICE Services

SPICE PLATFORM

Figure 4.23 DCS (distributed communication sphere) and its abstraction (communication model) as viewed in SPICE. (Reproduced with permission from SPICE project)

identified and running components, without having to worry about where they are located. It will also allow the same architect to refer to abstract services (those described within service ontologies) that are not tied to actual service instances already. These actual instances are searched out (using semantic discovery) and the whole set is dynamically orchestrated at runtime. This a posteriori mapping between service blocks and actual instances explains the term ‘loosely coupled’. Loosely coupled service components have the flexibility to cooperate dynamically with each other, even across heterogeneous middleware technologies, when a composition of their functionality is needed to reflect, for example, changing business roles, adaptation to new system and service demands, etc. As these components will potentially exist in many different administrative domains, it will be necessary for them to coordinate in order to set up mechanisms for translating and mapping security and privacy policies between them. Figure 4.24 illustrates dynamic service composition and the monitoring and controlling aspects of the process. Monitoring is vital for assessing the current and expected future status of services. The orchestration engine creates and sustains composed services with the help of information from the monitor component. Status information is disseminated in the form of asynchronous events. The orchestration engine controls and tunes components when necessary. In some cases it may be necessary to replace a component or distribute functionality. Remote operation is realized through the exposure layer. 4.4.3.3 Knowledge Management A shared understanding both of semantics and of syntax is a prerequisite in order to communicate, cooperate, etc. This also holds for software systems: two applications cannot interact with each other without common understanding of terms used in the communication. This is where ontologies come into the picture. An ontology is a formal specification of

Service Infrastructures

131

Remote Components

REMOTE EXPOSURE LAYER LOCAL EXPOSURE LAYER Local Components

1. Composed Service 2.

4. 3. Service monitor

Orchestration Engine

Service Execution Environment

Figure 4.24 Service composition with monitoring and controlling

concepts, their relationships (expressed in a logic) and specific instances, somewhat like an enriched thesaurus. This enables ontologies to support the sharing and reuse of formally represented knowledge among applications. Ontologies are mainly used in service architectures for publishing, discovering, (re)using and combining content, services, multimedia modalities, resources and devices. The goal of the knowledge management framework (KMF) is to provide an easy-to-use infrastructure that standardizes the exchange and discovery of knowledge. To have a shared understanding of the meaning of the information that is delivered and exchanged by the KMF, the sources of knowledge exchange their information as instances of a shared ontology. 4.4.3.4 Intelligent Service Enablers Besides various network, signalling and service platform enablers (enabling services, service functions), advanced service architecture solutions should employ the intelligent service enablers, aimed at assisting the existing enablers in providing the necessary information and knowledge for all the phases in service lifecycles. The intelligent service enablers are aimed at enriching the service platform with intelligent personal information and knowledge provisioning. Such functionalities allow a service – in a multi-domain environment – to access and meaningfully interpret the end user’s situation and to behave accordingly. There are a number of key functions and mechanisms needed in adaptive, mobile middleware that support ubiquitous, attentive and context aware distributed computing. Intelligent enablers retrieve and process information collected from various information and knowledge sources, e.g. heterogeneous context, user profile and service profile sources. The information is processed with advanced context-based reasoning methods targeting plausible and usable results. Scalable access mechanisms are needed, as well as exchange of information between platforms and domains. Prediction techniques, mechanisms anticipating (foreseeable) changes, and integration mechanisms yield proactive service enablers and extend the services with intelligence. These anticipatory service enablers allow for alertness and

132

Technologies for the Wireless Future – Volume 3

responsiveness to changes in the environment, proactively triggering mobile services in advance of changes having actually occurred. These approaches are grouped as follows: ž Personal information management – focuses on service platform provisioning of relevant and meaningful information and services based on user profile management and matching and profiling techniques. ž Distributed context information interpretation – incorporates advanced inference engines that are able to use dynamic context information, user profiles and application-specific knowledge in order to infer relevant knowledge for either the service provider, the application or the end user. ž Attentiveness – provides mechanisms that notify services of events (such as location or network QoS changes) that are relevant for them, and that can be specified by the service itself.

4.4.3.5 Open and Controlled Access New service architectures introduce a model where services are rapidly created and deployed through an open access to basic components and services. Those basic components include network resources, like location and presence information, which through the service platform will be easily integrated by service providers. Service architectures should aim at opening but controlling the access to all the services and basic components they handle to any actor from any domain. This control aspect includes the respect of users’ privacy. Actors who access service platform components and services will be either third parties or the users themselves. This principle implies that new service architectures will ease the emergence of new business models, between telcos, users and service providers, where telcos will act as service platform owners and provide access to internal components to service providers. The latter will then act as ‘third parties’ from the telcos’ points of view, while users will consume the services in a controlled manner. The service architecture will consequently provide mechanisms to manage the business and technical relationships with third parties, through the management of service level agreements, as well as the management of user privacy preferences. 4.4.3.6 Privacy Management Framework The knowledge management framework (KMF) consists of an ad hoc network of distributed knowledge sources, sinks and brokers. The knowledge sources (KS) retrieve information from sensors carried by the user, from databases, by reasoning, etc. The main objective of the KMF is to enable the generic discovery and exchange of information. This information consists of different types of data fragments, such as context information, user preferences, location, mood, recommendations, rules and facts. This information is more commonly called knowledge and is used by services which require specific user-related information to function. Thus, it becomes imperative for the user to have control over who gets access to the knowledge about them. The user should have the possibility of defining their own privacy policies, which control the access to knowledge about the user. As an example, we study a policy language to express

Service Infrastructures

133

End-user

Service Platform

Identity Management

Third party platform

Service

Service

Authentication engine

Authorisation engine User Privacy

SLA management

Service Consumer

Policy Enforcement Point

Service Provider

Policy Management

Figure 4.25 Service platform open and controlled environment

the permissions/restrictions on user-related knowledge. To achieve this, the operators assign ‘sensitivity labels’ to the different types of knowledge (‘knowledge names’) and the user may determine who is enabled to access the knowledge associated with them for a given sensitivity label. Figure 4.26 exemplifies this approach and shows how a user can control access to their knowledge. In the scenario shown, a knowledge consumer (KC) wants to access a particular user’s knowledge information from a knowledge source (KS). The KC (which can be a user, a service or another knowledge source) sends a request to the KS in the platform, which consists of a credential, the knowledge name (KN) required and the username of the user whose knowledge is requested (Target_UserName). The request is first intercepted by the policy enforcement point (PEP) of the service execution environment (SEE). If the KC is allowed to access services, the request is allowed to proceed to the KS. The PEP in the KS intercepts this request and forwards it to the policy decision point (PDP). The PDP checks with its policy database to see if the target user has allowed that particular knowledge to be given to that requesting user. If yes, the PDP sends back a response saying ‘Yes’ to the PEP, which then forwards the request to the KS. If the answer is a ‘No’ from the PDP, then the request for knowledge is denied. 4.4.3.7 Identity Management Framework Given the growing number of IT platforms and applications, a coherent management of identity-related information such as user profiles or authentication/authorization data is becoming an important factor to the attractiveness, efficiency and cost effectiveness of a telecommunications operator. For example, users do not want to remember many different

134

Technologies for the Wireless Future – Volume 3

Knowledge Consumer Sensitivity Lattice PEP of SPICE SEE

AAS

Critical Standard Query Language (e.g. WS-Policy, XACML)

{...,credential, KN,Target_UserName,...}

High

(credential, S,Target_UserName)

Personal

Low

Common

PDP

PEP of KS

Public

Yes | No

Buddy Lists

Policy Database

KS Group KAT (User Specific) KMT (Generic)

Target-User KDT (User Specific) Mood Happy

Mood

Personal

Location.City

High

Location.City

Munich

TimeZone

Low

TimeZone

GMT +2

KnowledgeAttribute Sensitivity Name(KN) Label (S)

High

Buddys Bob Locate Me

Common

Office

Sensitivity

To Whom

Individual

Service

KnowledgeAttributeValue

Figure 4.26 Privacy management

usernames and passwords to access services. Here single sign on comes in to relieve the user from this burden, thereby making the operator’s services more attractive to its customers. This section shows how an IMS operator can provide a single sign on service based on a smartcard application associated with the IMS subscription. An IMS operator may, in this way, also turn its large customer base and the strong authentication method into a new asset by offering authentication as a service to application providers. The IMS operator can also offer a manifold of services in a secure way without having to issue additional security credentials. Within IMS, the identity of a user is represented by the user’s private identity (IMPI) and associated public identities (IMPUs). The IMPI corresponds to an application on the smartcard inside the user’s mobile phone and is used for authentication by means of the AKA protocol. The IMPUs are the user’s public identities as seen by others. The user can choose which IMPU to use for an actual request. The chosen IMPU is sent with all service requests inside the operator’s IMS network and may be trusted by application servers. This implies a single sign on functionality because on reception of a request the service need not verify the user identity any more. In particular, the service does not need to ask the user to authenticate again for this specific service. This IMS solution to single sign on does not work for non-IMS voice over IP solutions based on SIP as specified by the IETF. Here the recent RFC 4474 on SIP identity describes how authenticated identity information can be integrated into an additional header of IETF SIP messages. On interworking of IMS with IETF SIP-based networks, the IMS may insert such a header with an authenticated identity into signalling messages towards non-IMS SIP-based networks. The originating proxy (S-CSCF in IMS) verifies this identity information based on

Service Infrastructures

135

the IMS authentication by AKA protocol. Of course, it still has to use signatures to assert this identity information to the terminating non-IMS SIP proxy. For user self-administration of HTTP-based supplementary services for IMS, e.g. presence list management, a single sign on solution is standardized in 3GPP. It is based on the same user identity and credential as access to SIP-based IMS services. Authentication is based on the generic bootstrapping architecture, as described in the following section. In order to leverage the IMS authentication for non-SIP-based services, two solutions currently exist. One can use the generic bootstrapping architecture (GBA) as defined by 3GPP, or one can combine GBA with Liberty Alliance protocols. 3GPP Generic Bootstrapping Architecture (GBA) The 3GPP generic bootstrapping architecture (GBA) is a way to establish a shared secret between a user device and an application server, using the security association between the user’s smartcard and the IMS operator’s (or any mobile network operator’s) home subscriber system. This shared secret can be used, for example, for single sign on or to encrypt media streams. Applications using GBA can be based on HTTP, for instance. On the user side, GBA works with 3G UICC smartcards as well as with 2G SIM cards. Another feature of GBA is that user security settings (GBA-related user profiles including identities and authorization flags) can be stored in the HSS and evaluated by the network operator or by the service provider. This may be used, for example, to exercise access control or to give the user the opportunity to use only certain identities with certain services. Since its foundation in 2001, the Liberty Alliance Project has developed several protocols for web-based single sign on solutions: the Identity Federation Framework (ID-FF), the Identity Web Services Framework (ID-WSF) and the Identity Service Interface Specifications (ID-SIS). ID-FF provides protocols for single sign on and identity federation, which can be

HSS

Application Server BSF

UE

Figure 4.27 GBA architectural overview

136

Technologies for the Wireless Future – Volume 3

used with normal web browsers. ID-WSF specifies protocols for message exchange between SOAP-based web services and ID-SIS-specific data services. Liberty Alliance deliberately does not specify authentication methods used between the user and the identity provider. Thus the Liberty Alliance framework is ideally suited to extend the authentication provided by network operators based on their customer relation to single sign on for any services within Internet and other networks. For this purpose, the Liberty Alliance Identity Provider plays the role of a GBA application. In this way, authentication with the Liberty Alliance frameworks is based on the smartcards issued by mobile network operators or IMS operators. 4.4.3.8 Service Roaming A mobile service is typically executed in a single execution environment, such as an IMS application server. It is expected that in the near future services will no longer be executed within a single server or domain, but rather their lifecycles will span multiple servers and platforms. A service can be distributed over multiple administrative domains and servers for a number of reasons. One reason is simply that a particular service is not available natively on the local platform. Many different information retrieval, search and access services can be realized using well-specified external interfaces, for example access to Google search API using SOAP invocations. Another reason is service localization or service usage optimization due to terminal or user mobility. Service roaming support can be applied to the following situations: ž Access to the chosen set of home services and information. ž Access to suitable services and information from the foreign domain (corresponding to the user profile), preferences and service context information). ž The possibility to combine home services/service enablers with services/service enablers from the visited domain to new services.

HSS Identity Provider

= GBA Application

Service Provider

BSF

UE

Figure 4.28 Combination of GBA and Liberty Alliance

Service Infrastructures

137

In order to carry out those use cases the service roaming manager enables the inter-platforms mechanisms so that home functionalities can be performed through the visiting domain. In order to be able to make meaningful service activation and execution plans, each platform must have models of the components and composed services. The models must be expressed in a language that allows comparison and combination of the models. 4.4.3.9 Device Aspects Modularization and platformization are seen as the two main elements of the future mobile device architecture, addressing current architectural challenges. With respect to the service architecture, modularity is encapsulating and classifying services, whereas platformization is providing additionally unified, persistent interfaces. The level of services can vary of course, from end user services down to platform services, but concepts should be applied in a similar way. Both elements are introduced in the following. Modularization of the Device Service Architecture Modularization of the device architecture has two cornerstones, namely interconnectcentricity and services. Both have strong positive implications to modularity and performance, as well as security. Modularized device service architecture consists of three logical foundation element types, as shown in Figure 4.29, namely application nodes (AN), service nodes (SN), and interconnect (IN). Communication between any nodes, regardless of their physical location, always takes place over the interconnect. Service nodes provide services in a black box manner, hiding the underlying implementation. The higher the service abstraction, the looser the service coupling when used. Services are accessed through a message-based interface, for example. All information exchange related to this interface is called service communication. Application nodes can use services, whereas they cannot provide any services. However, application nodes may possess data objects as well as access data objects owned by other nodes in the network. All communication related to data objects is named data communication. The device platform architecture is viewed as a set of subsystems connected together via an interconnect, as shown in Figure 4.30. Each subsystem contains all the resources needed to implement the service and application nodes mapped to the corresponding subsystem. Typically, the interconnect consists of two parts, namely high interconnect (H_IN) and low interconnect (L_IN). High interconnect provides mechanism for service and data discovery and access. Low interconnect is responsible for reliable information transport. It is important SN AN

SN

Interconnect

AN

SN Service Communication

SN

Data Communication

Figure 4.29 Modularized logical architecture

138

Technologies for the Wireless Future – Volume 3

Device Y

Device X

AN AN

SN

SN

H_IN L_IN

Switch

Switch

L_IN H_IN

SN

SN

Sub-System

SN

H_IN L_IN

AN SN

Sub-System AN SN

Sub-System

Sub-System

SN

AN

SN

SN

AN

SN

H_IN L_IN

Switch

Switch

L_IN H_IN

L_IN H_IN

SN

SN

SN

SN SN Sub-System

SN

SN Sub-System

L_IN H_IN SN

SN Sub-System

SN

H_IN L_IN

AN

L_IN L_IN H_IN

Sub-System SN AN SN

SN SN

Sub-System Device Z

Figure 4.30 Modularization of the device service architecture

to keep in mind that subsystems can be separated either within the hardware domain (either on-chip of off-chip) or within the software domain. In the first case the interconnect solution is hardware based (e.g. MIPI UniPort), whereas in the latter case the interconnect solution is software based (e.g. IPC). In both cases, the specification of interconnect stack has been done keeping in mind complexity and performance. Platformization of the Device Service Architecture Within the mobile device architecture, platformization is a key enabler in managing architecture maintenance and its further development. The concept of platforms or platformization tries to provide a solid environment which developments and innovations can start building on, without having to understand how the platform itself is implemented. Ideally, a platform separates properties of the functionality below the platform from functions above the platform. Taking this forward, functionalities provided from a platform are considered as platform services. This implies that these platform services are encapsulated in such a way that there is no dependency on functions (or better services) external to the platform. The separation into platform dependent and independent functionality does not refer to any specific layer. Indeed, as the mobile device architecture spans from low-level implementations details up to end user applications and services, the abstraction level varies in a similar way. Typically in the mobile device architecture, the hardware abstraction layer (HAL) and the application programming interface (API) are the two most prominent levels, as shown in Figure 4.31. From the platform security, component integration and platform performance point of view, the lower platform layer is of specific interest and importance. In this, the concept of persistent and standard HW interfaces (e.g. MIPI, PCI) is extended into the software domain through the definition of hardware abstraction layers. They are typically stacked, encapsulating and abstracting specific functionality of the device. Platform resources are thereby presented in a convenient form to the operating system and via this to the end user applications. Taking the example of a hard disk, hardware abstraction provides a convenient way to read/write data without having to handle (and to know about) disk sectors and tracks.

Service Infrastructures

139

Services and Application Framework API Operating System Kernel HW abstraction Hardware resources

Figure 4.31 Platformization of the device service architecture

In order to achieve full platformization across a fleet of devices, hardware abstraction is not sufficient, as real machine compatibility and hardware resource constraints cannot be circumvented. Therefore virtualization is introduced, which maps visible interfaces and resources (at a given level of abstraction) onto interfaces and resources of an underlying, possibly different, real system. Coming back to the example of the hard disk, virtualization transforms, for example, a single large disk into two independent virtual disks, each of which appears to have its own tracks and sectors. Virtualization also enables hosting of multiple domains and domain separation. Isolating domains means that one domain does not necessarily notice that it is running beside another on the same device, enabling modularization concepts in the software execution environment. Typically these are hosted from a virtual machine monitor, being responsible for resource management and context switching. A potential setup is shown in Figure 4.32. 4.4.3.10 Visualization of Resources There are two different important tasks concerning virtualization: The virtualization of hardware and the management of virtualized hardware. Another approach to virtualization (e.g. Application 1

Application 2

Application 1

Application 2

Application 1

Application 2

Execution Environment 2

Execution Environment 2

Execution Environment 1

Virtual Machine

Virtual Machine

Virtual Machine

Virtual Machine Monitor Hardware Platform Figure 4.32 Virtualization of the device service architecture

140

Technologies for the Wireless Future – Volume 3

Guest Guest OS 1 OS 2 Virtualization Host OS Hardware

Figure 4.33 Full virtualization

programmable networks) is provided by execution environments, which are hosted by the network. Virtualization of Hardware Currently, many projects and companies are developing virtualization solutions, e.g. VMWare, XEN, KVM (kernel-based virtualization), Microsoft Virtual PC, Parallels Desktop, Virtual Box, Qemu, Virtuzzo, OpenVZ (based on Virtuozzo), Linux VServer, etc. These solutions are based on different kinds of virtualization. Forms of virtualization are, for example, full virtualization, paravirtualization and OS-layer virtualization [76]. When offering full virtualization, full-featured hardware is virtualized (e.g. virtual bios, CPU, graphic, storage, network adapter, etc.). Unmodified guest OSs can be run within a VM, being isolated from each other. Typically, many VM instances can be run concurrently, sharing the available hardware resources among them. This kind of virtualization produces high overhead, slowing down VM performance. In Figure 4.33 full virtualization is shown. The virtualization layer (virtual machine monitor (VMM) [77]) is located above the host OS; this means the virtualization software runs as a normal application on the host OS. An example of full virtualization is the VMWare workstation. To avoid drawbacks of full virtualization, paravirtualized VMs are similar but not identical to the underlying hardware [78]. XEN 2.0 is an example architecture using this approach. The virtualization software (hypervisor) of XEN runs as OS directly on the hardware. Every guest OS has to be modified (hard to achieve for closed source OSs) to be paravirtualized. Instead of performing system calls, the guests have to perform hypercalls on the hypervisor. Real hardware access is possible for the guests, under the control of the hypervisor. The hypervisor has to support all of the underlying hardware. In XEN 3.0 [79] guests can be virtualized without modifying them, using virtualization support of X86 CPUs. Figure 4.34 shows virtualization done by a hypervisor without an additional host OS. To overcome the problem of providing different hardware drivers, XEN uses a trick. The hypervisor only manages central elements, involving CPU and memory, while other (less critical) hardware is accessed by a so-called domain 0 guest. This domain 0 OS is a privileged guest OS having access to the hardware. The operations of other guests of domain 1 are relayed to the domain 0 OS. In Linux OS, a new form of virtualization technique has recently been proposed, the KVM (kernel-based virtualization). Here, also, a hypervisor is used, but it

Guest Guest OS 1 OS 2 Hypervisor Hardware

Figure 4.34 Hypervisor

Service Infrastructures

141

is integrated as a kernel module within a full-featured Linux OS. The OS level virtualization approach, in contrast to the other virtualization solutions, does not virtualize a full-featured hardware for every VM. Instead, all of the guest OS environments share the same kernel. Nevertheless, the VMs are isolated from each other. An example of this kind of approach is the Linux V server. Management of Virtualized Hardware When hardware is virtualized in several VMs, they need to be managed. More specifically, the available hardware resources have to be distributed among the VMs running on it (e.g. bandwidth, CPU-power, memory, storage space, etc.). The allocation of resources depends on the services which are provided by the VMs (e.g. their SLAs, current and future access of users to these services, importance of the service, etc.). Additionally, VM management needs to perform, particularly if more than a single hardware is available to run VMs on. VMs have to be created, terminated, cloned, or moved from hardware to hardware. This VM management needs to consider the available resources of all involved hardware. VMs can be moved from hardware to hardware while they are offline or while they are online (live migration). None of these management functions are provided directly by the virtualization software; they are provided by special management solutions. Figure 4.35 shows the management solution of VMWare, the Infrastructure 3 [80]. Programmable Networks Programmable networks have been proposed as a solution for the fast, flexible and dynamic deployment of new network services. Programmable networks are networks that allow the functionality of some of their network elements to be dynamically programmable. These networks aim to provide easy introduction of new services by adding dynamic programmability

VirtualCenter Management Server

Manage

Virtual Machines

Virtual Machines

Virtual Machines

App

App

App

App

App

App

App

App

App

App

App

App

App

App

App

os

os

os

os

os

os

os

os

os

os

os

os

os

os

os

ESX Server

ESX Server

ESX Server

Physical Servers

Figure 4.35 VMWare Infrastructure 3

142

Technologies for the Wireless Future – Volume 3

to network devices such as routers, switches and applications servers. Dynamic programming refers to executable code that is injected into the network element in order to create the new functionality at runtime. The basic idea is to enable third parties (end users, operators and service providers) to inject application-specific services (in the form of code) into the network. Applications may utilize this network support in terms of optimized network resources and, as such, they are becoming network aware. Programmable networks allow dynamic injection of code as a promising way of realizing application-specific service logic, or performing dynamic service provision on demand. As such, network programming provides unprecedented flexibility in networking. However, viable architectures for programmable networks must be carefully engineered to achieve suitable trade-offs between flexibility, performance, security and manageability. The key question from the public fixed, mobile operators’ and Internet service providers’ points of view is how to exploit this potential flexibility for the benefit of both the operator and the end user without jeopardizing the integrity of the network. The separation between services (applications) and network infrastructure results in some loss of interaction between the two layers; the programmable network technology offers some remedy to this inefficiency. Active or programmable networks can be used in conjunction with overlays for the benefit of applications. Active networks are ‘active’ in two ways: routers and switches within the network can perform computations on user data flowing through them; and users can ‘program’ the network, by supplying their own programs to perform these computations. Active or programmable networks can rely on active packets (i.e. those carrying executable code), active nodes (the code resides in the nodes, but it is initiated by commands in the packets) or both. The FAIN system (http://www.ist-fain.org/) relies on code that is sent by applications or service providers to network nodes to implement customized management of their data flows. Code is executed on designated routers and can affect resource management through a control application-programming interface. Full state-of-the-art review and research in programmable networks can be found in [81]. 4.4.4 Functional Architecture and Service Delivery Framework Telecommunications systems typically have a layered structure. The motivation for layering is that it separates functionality and APIs, which promotes ease of development, portability and manageability, with the price of potentially increased overhead. Protocols are organized into a protocol stack, for example the TCP/IP stack. Above the core networking stack, we have various middleware, which has become one of the cornerstones of modern distributed systems development. The middleware layer can be split into sublayers, each highlighting a specific domain of functionality. In this section, we consider a functional architecture that has four clearly defined sublayers focusing on the core aspects of the distributed service execution environment, namely the capabilities and enablers layer, the component service layer, the knowledge layer and the value-added service (VAS) layer. Figure 4.36 presents an overview of the layered architecture. The terminal platform is presented on the left of the figure and the server-side distributed platform on the right. Both the terminal and the server-side system have a layered structure: ž The capabilities and enablers layer consists of external services and enablers that are used by the platform. Entities on this layer are not part of the core platform. ž The component services layer provides facilities for component-based development and deployment. This layer includes services such as the message router, various managers,

Service Infrastructures

143

Service Execution Environment Terminal Platform SPICE Service Execution Environment Value added services layer Composite components and orchestration

Value added services layer Knowledge layer Component layer Capabilities & Enablers IMS client Browser Basic OS support

Knowledge layer Brokers, Mediators, Reasoners Component service layer SPICE components and component Capabilities & Enablers IMS System Legacy systems

Third party components

Repositories, profiles, ACLs, SLAs

Figure 4.36 Overview of the layered architecture

controllers, watchers and resource adapters. The resource adapters are used to integrate legacy components with the platform components. ž The knowledge layer supports the discovery, delivery and transformation of information, such as context and presence variables. This layer introduces several fundamental building blocks for intelligent components, such as the knowledge brokers, reasoner and recommender services. ž The value-added services layer exposes services to the outside world. Typically, these services are composite components, built from basic and composite components. Runtime discovery and orchestration support the creation of ad hoc consumer services. Components are accessible by the outside world through the exposure and mediation layer. A third party SEE (service execution environment) can use the publishing capability of the platform to publish and advertise services or components in the overlay systems of different platforms. SCE (service creation environment), presented on the top of Figure 4.36, offers a set of integrated tools which support the service creation process. Separate toolsets are created to support the professional users and end users in service creation, service emulation and the service deployment process. ACE (automatic service composition engine) produces automatic service compositions based on formalized service requests. Formalized service requests include a semantic description of the desired service or service composition and, optionally, nonfunctional properties or constraints. The KMF (knowledge management framework) consists of a network of distributed knowledge sources, sinks and brokers. The main objective of the KMF is to enable the generic discovery and exchange of information, such as context information, preferences, recommendations and rules. This objective will be achieved by providing a common set of interfaces for knowledge discovery and knowledge exchange within the knowledge layer of the platform. The use of IMS technology ensures a solid basis for application and service signaling; however, the development of added-value services on top of the system

144

Technologies for the Wireless Future – Volume 3

is crucial. Due to its success, the World Wide Web is host to vast amounts of distributed, machine-processable data covering all aspects of human life. Reuse, common interpretation, combination and shared understanding of these data are necessary for the mobile services to be context-aware. To address the interoperation challenge of the web and mobile service environment, a wide-ranging mobile ontology (http://ontology.ist-spice.org/) is being developed. This comprehensive mobile ontology covers the mobile communications domain, specifically addressing persons, terminals, services and networks. The mobile ontology is a machine-readable schema specified in RDF/S and OWL formats; these technologies have been chosen due to their relative maturity. The ontology enables sharing of knowledge and exchanging of information both between people and across services/applications in mobile environments within any domains. Security and access control in the platform is provided by the access control and ID-management framework. This framework provides a single sign on feature to the user by integrating the credentials, GBA (generic bootstrapping architecture) from the mobile world and Liberty Alliance from the IT world. In addition, the framework also takes care of enforcing policies related to privacy, access control and service level agreements (SLAs). The framework enables third party service providers to access services and enablers in a secure manner and in a way that is consistent with the SLAs negotiated for such usage. This will help third party service providers to offer innovative services by leveraging existing service in the platform, thereby reducing the service development and deployment time. The multimodal delivery and control system (MDCS) is responsible for delivering a particular end user service to a particular user, utilizing the resources currently available in that user’s distributed computing sphere (DCS, SPICE deliverable 3.2: Specification of the service enablers provided by the user equipment management, dynamic desktop and resource utilisation in the DCS, (http://www.ist-spice.org/documents/D3.2_v1.0-140108.pdf)).The MDCS is composed of three main sub-components: ž The Resource Coordinator – the entry point to the MDCS, it takes decisions on how to render a service, in terms of multimodal adaptation and session mobility, based on recommendations from the KMF. ž The Modality and Adaptation Enforcement Point – responsible for enforcing the decisions of the resource coordinator regarding multimodal adaptation, i.e. realizing the multimodal interface for the user. It relies on renderers and activators on the devices of the user’s DCS to deliver adapted service output and gather user input. ž The Session Mobility Enforcement Point – responsible for enforcing the decisions of the resource coordinator regarding session handover between the devices of the user. The content protection framework specification work has been performed with the aim of maintaining independence from specific media objects, formats, operating systems and runtime environments. In order for the end user to render the content on a device, the user must have both the protected content and the associated rights object. The license issuer is the entity responsible for granting permissions to render the protected content on the end user device. 4.4.5 Interfaces Service interfaces with respect to the device platform will need to distinguish between the low-level and the high-level platform services. The former will be linked to the API and

Service Infrastructures

145

the device service layer (e.g. video, storage, camera service, audio, communications, etc.), whereas the latter are related to the hardware abstraction layer (e.g. device driver services, memory access, scheduling, etc.). The service architecture must also interact with the capabilities and resources offered by the underlying network, a generic means for such an interaction being provided by resource adaptors. Such resource adaptors act as gateways between the underlying network capabilities and the service platform, allowing events received from such networks to trigger some service logic in the platform, as well as enabling applications to interact and use capabilities provided by such networks. An example of service-to-networks interface (i.e. ambient network service interface) is presented in this section.

4.4.5.1 High-level Platform Service Interfaces Each service has a formal XML-based service interface specification (SIS) describing the messages the node understands and generates. SIS also specifies which services the corresponding service node uses. SIS can be expressed in a form following the Web Services Description Language (WSDL) specification. Service identification and addressing methodology is an important part of the service architecture. In the simplest case, a unique string containing the name of the service, vendor, release and version could identify a service. Identification information is used by a service to register itself into the device. As an outcome, the service node gets a device-context unique service identifier (SID). Using this, clients may later on access the corresponding service. In SOA and web services-based service environments a number of SOAP intermediaries may act on a message. A message path comprises the nodes that process the message from the sender to the ultimate receiver. The SOAP header system is a flexible basis for building mediation systems; however, an intermediary may also need to inspect the content of a message realizing content-based routing. In content-based routing the destination is determined by an active intermediary on the basis of the content of the message and the subscriber interests. In addition, to maintain security policies and to tag messages with security tokens, mediators may be used for various transformations, augmentations and handling special situations, such as detecting spam and denial-of-service attacks. The service broker is a key component in the architecture that is responsible for connecting different components together and for targeting messages to their proper receivers. The component connections necessary to serve a request are calculated by the service broker when a request is received by the service platform. The selected components and their connections are referred to as a service component network. A component model should allow component networks to be adapted to the conditions of the environment. This is supported through the notion of an abstract component network. These networks allow connections between components that are not known beforehand at the time of designing the composite system. The networks have adaptation rules that control how the concrete components and their connections are selected to best adapt to the environment. When a broker is used to compose a service, the service composition network is saved to the internal storage of the broker and a session identifier (ID) is assigned to the network. This session ID is included in the request to the components of the component network. The broker is therefore responsible for session management pertaining to composite components.

146

Technologies for the Wireless Future – Volume 3

HAL Monitor

Data base

LLDD

Plug & Play

Kernel

Interrupt

LLDD

LLDD

Peripheral

DSP

Sub system

DMA transfer

Resource Manager

Scan

LLDD

Proccessor

MMU

DMA

Memory

Figure 4.37 HAL (hardware abstraction layer) elements

4.4.5.2 Low-level Platform Service Interfaces The hardware abstraction layer separates the software environment into (hardware) platform-independent and a (hardware) platform-dependent parts. The four elements can be described in more detail in the following: ž The hardware components, which will be abstracted, contain the (application) processor(s) together with its main peripherals (DMA, interrupt, memory), platform peripherals, other programmable components (e.g. DSPs) and complex subsystems (e.g. video/image, communications). Hardware components may be optional with respect to different platform instantiations; pluggable (e.g. headphones) and component parameters are dependent on the platform instantiation. ž To cover and encapsulate specifically the variations and configuration of platform components, a capability database, containing all relevant platform implementation-specific parameters, is considered. It provides a single point of access through a monitor interface, which allows applications and services to adapt themselves to the existing platform configuration, without having this information specifically built into the application itself. The capability database might also support basic resource management (e.g. binding and preallocation of platform resources to requesting services). ž Abstraction of platform services is done through low-level device drivers. These provide basic services to access peripheral components and subsystems (e.g. read from/write to a hard disk). Device drivers hide the respective implementation details and settings. The device drivers may use the capability database to optimize their performance (e.g. take use of DMA). A management interface is required to enable, for example, loading/unloading of a device driver module. ž A basic kernel offers basic OS-like services, which enables the hosting of different execution environments in a virtual environment (virtual machine monitor). It therefore has to deal with domain protection and isolation of the different domains, as well as enabling inter-domain communication. Additionally, the kernel offers basic services like memory

Service Infrastructures

147

allocation, context switching, interrupt handling and scheduling. The service level is heavily dependent on the used kernel implementation. Currently these interfaces (i.e. the functionality, available parameter, name space, etc.) are heavily dependent on the operating system, the target platform and the component provider. Reuse across different implementations is limited and always needs a certain amount of adaptation and porting work. Therefore interoperability and maintenance of different components is necessary, otherwise R&D efficiency and the ability of platform management are not given within a modular, multiprovider, heterogeneous environment. 4.4.5.3 Ambient Network Service Interface This section describes the ambient network service interface, defined by the Ambient Networks project as a possible instance of service-to-network interface. The purpose of the Ambient Networks (ANs) project [82], [83] is to develop enabling techniques to support distributed management of new services within and across dynamically changing and heterogeneous wireless ad hoc networks. An AN consists of potentially many heterogeneous AN nodes, which are dynamic in the sense that they may rapidly join and leave the AN. AN nodes share a common control space, known as the ambient control space (ACS) [82], [84], which consists of several functional entities (FEs), e.g. context FE, mobility FE, security FE, overlay management FE [84], that interact within/across ACS(s) to support current and new services in ANs. Due to the service-oriented architecture of the ACS, the natural solution is to construct ASI as hub-and-spoke server, which exposes internal ACS services. The ACS architecture is shown in Figure 4.38. The ACS possesses three types of interface: ANI, ARI and ASI. The ambient network interface (ANI) [83], [85] connects the ACS of different ANs. This interface is used for negotiation of network composition agreements and for transferring control information between the networks. The ambient resource interface (ARI) [84], [85], [86] lies between the ACS and the physical connectivity network. The ARI abstracts the specific control techniques of the underlying connectivity network and allows control functions of the ACS FEs to operate seamlessly over a heterogeneous network infrastructure. Finally, the ambient service interface (ASI) [85], [84] interfaces with the external services and clients, and allows them to issue requests to the ACS concerning the establishment, maintenance and termination of end-to-end connectivity between functional instances in the AN. 4.4.5.4 Ambient Service Interface Concept The main task of ASI is to make available the ACS functionality for external users and applications in a possibly transparent and uniform way. Though many attempts have been made in order to unify the aspects of attachment operator services to the telecommunication networks in terms of APIs (like the aforementioned Parlay/Parlay X), we propose another solution that concerns only the communication layer, which can be used in situations where there is a need for distributed, flexible and programmable interface, which hides the complexity of the internals of the considered system. The proposed solution was successfully deployed in an early implementation of ACS in AN project; however, it can be used in any distributed environment, especially in one implemented around the SOA paradigm.

148

Technologies for the Wireless Future – Volume 3

Client1

Client2

AS

Legend: ASI – Ambient Service Interface ANI – Ambient Network Interface ARI – Ambient Resource Interface

nt Control Spa ce bie Am

AR AN

FE1

Ambient Connectivity

FE2

FE6

AN

FE5 FE3

FE4

Figure 4.38 Ambient network interfaces: ARI, ANI, ASI

4.4.5.5 ASI Switchboard As a high-level interface, the ASI is responsible for handling the high-level messages, then decomposing them into smaller invocation primitives, and sending them to the final destination (to the FEs that carry out the required operations, like set up of the SATO – service aware transport overlay, delivery of ACS context information, etc.). Such a task concerns the advanced functionality such as that found in ESB systems, where many event-driven entities communicate over standardized message bus, and where the routing of the messages is based not only on their primary destination, set in the address headers (like destination endpoint in SOAP header), but on the message content, performed operations, their parameters and context. The main concept of ASI is its design as a distributed interface, available in each ACS and their selected ONs (overlay nodes, the main part of a key element of ANs – the SATO). When the external client sends the ASI primitive (related to the specific FE; in our case identified by the name of the primitive) towards the ASI, the ASI contacts the ACS registry (formerly the ACS framework component) and finds the location of the requested FE. After successful FE lookup, the ASI forwards the primitive to the destination FE, which carries out the requested operation. A full sequence diagram of all invocations while communicating via ASI components is depicted in Figure 4.39. The main component of the ASI is the ASI switchboard mediator (ASB mediator), which receives all the ASI primitives destined for the particular ACS, and makes a decision according to the logic included in the mediator (or a chain of mediators). In our case, the mediator routes the messages according to the information found in the ACS registry, where all the FEs register after their startup and deregister when they are being stopped. An additional helper element of ASI, the ASB CC (ASI switchboard configuration component), is used to store all the configuration parameters required by the ASB itself, since the ASB may implement a stateless logic due to its role as a relay. Such an approach ensures location transparency of internal components of the ACS (the FEs), since the external client does not have to know their real location (the client has to know only the address of the ASB). This fact has a big consequence during the AN composition process, when two or more ACSs are composing into one ACS, and the ASI has to handle these changes. The problem of AN compositioning itself is handled by another ACS interface – the

Service Infrastructures

ASI Client

149

Bootstrap Mgr

ASB Mediator

ASB CC

ACS Registry

configureASB()

FE

register() configureASB()

ASI primitive findRegistry()

lookupFE() ASI primitive ASI primitive ASI primitive

Figure 4.39 ASI operation

ANI. It uses the DEEP protocol (destination endpoint exploring protocol) [87] and solves the problem of changing the content of composing registries in order to reflect the current state of registered FEs. After the composition on the network level, the ASI uses the new ACS registry content to route the messages, so all the changes are handled transparently from the client point of view. 4.4.5.6 ASI Flexibility Each ACS constitutes a certain type of domain with a list of FEs and their relations. The domain can also be seen as a security domain, which influences the operation of the ASI as an ACS access element. Communication of external clients with ACS via ASI has many consequences compared to the direct communication with FEs themselves (see Figure 4.39). Besides the base ASI functionality in the form of internal lookup and routing to the given FEs, it could enrich the interface by certain nonfunctional features, such as simple authentication and authorization or realization of certain security policies in general in ANs. The flexibility also means the ASI extensibility during runtime. We can imagine that the ACS functionality is extended by the addition of a new FE, which registers in the ACS registry. After this, the new functionality is automatically available via ASI. The ASI always makes lookup in the ACS registry before forwarding the message (the lookup is based on well-defined criteria, like message context, operation signature, etc.), so the new functionality can be easily added, without changing the ASI itself. Another possibility is to exchange the FE after its failure or to use the ASI as a load balancer in a transparent way from the client point of view. Another flexible ASI feature is the possibility to implement the mediating component in the form of a protocol gateway, which can handle many protocols from the outside of the ACS and then translate them into internal FE calls. Such a solution makes the ASI open for different network protocols.

150

Technologies for the Wireless Future – Volume 3

4.4.5.7 ASI Configurability and Programmability ASI configurability is the ability to change its functionality before the ASI interface is started. The ASI configuration involves specification of the sequence of mediators used during the message processing. Next, it can define rules that decide which mediators are used in which order. The configurability concerns more the initial ASI parameters, which can be stored in a configuration file or a database. Programmability is another ASI feature that allows us to put an additional logic to the ASI interface, compared to the logic performed by the FEs. However, the ASI logic should concern the highest level of communication with ambient network, i.e. the coarse-grained operations, like establishment of the SATO (service aware transport overlay), attachment of external service provider or aggregated context dissemination. This programmability is achieved thanks to the ASI mediator architecture, where each ASI request is handled by the specialized mediators, which sequentially process the request, until it reaches the final destination. The process of mediation can range from decomposition of the high-level ASI messages into the fine-grained second-level ASI requests suiting the functionality of each FE, through the process of message-switching, based on the FE database (the ACS registry), and finally to the validation of incoming messages against some message schemas, rules or security policies. Such defined programmability should be achievable during the system runtime, i.e. it should be possible to inject a new mediator into the chain of existing ASI mediators. For this reason there should be a method of packaging the mediator functionality into a well-defined and distributable type of component, which can be developed and then deployed in the ASI framework to carry on a required functionality. ASI programmability differs from configurability because it introduces functional adaptation into the ASI operation. Since configurability concerns only the possibility of setting some initial parameters, sequences and mediators for the ASI, the programmability allows changing of the processing path and defining of new mediators during the runtime, depending on the perceived environment properties, such as services, etc. Based on that information we can change, for example, an internal message routing or the processing rules. This facility can be called a logic injection and can be performed thanks to the possibility of loading and deploying new mediators during the runtime (the full potential of the proposed ASI architecture may be achieved when runtime reconfiguration and programmability are supported by the software environment). The mediators used can either be new switching components or can contain an additional logic concerning the security or message validation. ASI mediator can also discover any modifications of the ACS registry (the ACS framework component) and trigger an action on FE registration or deregistration. In some cases leasing of FEs should be also considered, since we do not know whether the registered FE is still active and achievable. In case of an access attempt to the inexistent FE, ASI switchboard can return an error message containing the error description. This dynamic of ASI reflects well the dynamics of the ACS registry. 4.4.5.8 Ambient Service Interface Design ASI design and implementation is based on the new WS 2.0 (web services 2.0) specification. The technological solution proposed for AN is the synapse-based ASI framework, which allows deployment of programmatic functionality and mediates between users, external applications and service providers, and the ACS internals, i.e. the FEs.

Service Infrastructures

151

ASI Switchboard

External Services

ACS

ASB Configuration Component Service Container (c.2) (c.3)(2) (3) ACS Client

(1)

ASB Mediator

(6) FE SAP

(7)

(8) ASI ReferencePoint

(r.1) Mediator Framework

FE

(r.4) (c.1) (c.4)

(r.2) (r.3) (4) (5)

ACS Bootstrap Manager

ACS Registry

Legend: ASB ConfigurationPhase: c.1-c.4 ACS Registry Configuration: r.1-r.4 Communicationvia ASB: 1-8

ACS Framework Component Service Container

Figure 4.40 ASI switchboard

We assume for the purpose of prototyping that all FEs are represented as WS. However, this does not influence the type of protocol used outside the ACS, where we can plug in other types of transport, or even try to implement other types of API and then, using a mediator, translate them into the API used by the ACS FEs. The scheme of the ASI switchboard (the ASI mediation framework with routing mediator deployed) is depicted in Figure 4.40. In order to operate properly, ASI requires some steps of initialization. Firstly, the ASB configuration component, which holds all the ASI switchboard configuration parameters (like location of ACS registry) is set by the ACS bootstrap manager (messages c.1–c.4). Secondly, we have to be sure that all FEs are registered with ACS registry (r.1–r.4), since then communication can be established between the ACS client and the given FE (messages 1–8). During the communication, the switchboard dispatches the messages and routes them to the particular FEs based on some predefined rules and the logic included in the ASB mediator. The shown ACS registry is responsible for FE lookup performed by the message content (it could be a name of the invoked method, for example), while the FE SAP is simply a FE service access point, which mediates between the WS protocol and the FE internal protocol. Because the constant contacting of the ACS registry could become a bottleneck, authors are considering additional improvements to the presented architecture, which include the caching capabilities to store the address of ACS registry and the locations of recently requested FEs. We are also aware that the simple process of FE registration and deregistration seems not to be a reliable solution in case of FE failure or communication problems, and as such should be complemented by some kind of leasing mechanism. 4.4.6 Groundings and Exploitation Convergence, fixed-mobile convergence and IP multimedia subsystem (IMS) have become buzzwords in the telecom society. The term ‘fixed-mobile convergence’ (hereafter referred to

152

Technologies for the Wireless Future – Volume 3

as FMC) is currently used in different contexts. The common use of networks and, above all, of their services is known as ‘fixed/mobile convergence’ (FMC). Its use spans from interworking between mobile networks and the PSTN domain, to convergence in dual mode devices, which enable WLAN/2G/3G voice roaming, and to FMC as a service delivery platform common to fixed and mobile networks. 4.4.6.1 Expectations from End Users End users prefer robust and secure communications, attractive prices and attractive services that make today’s life easy and comfortable. They want to enjoy new communication services that can be accessed anytime, anywhere, on any device independent of the access network. They expect services/service profiles that support their mobility at home, during travel and on business, such as one-number service, one (voice) mailbox, one address book, etc., and which enable them to switch easily between their private and business roles. They want to use different types of end devices, and they expect that the offered services are automatically tailored to the end devices’ capabilities. Moreover, end users need devices which support a user-friendly handling of the services and which enable easy and comfortable subscriber self-administration of the capabilities of the terminals and the subscribed services. A network-based address book would be an example of such a service. Everyone is familiar with different kinds of address book: one made of paper, a second one in the portable DECT phone, a third one located in the cell phone, and a fourth one stored on a PC. All address books have to be manually synchronized and they seem to be always out of sync. One additional fundamental element of FMC is the support of ‘one-stop shopping’, by the provision of a single customer interface, i.e. a unique point of contact for the user dealing with every customer need. Besides appropriate customer care functionalities, customized billing information, for example a unified bill for all services with every requested level of grouping, simplifies the cost transparency for the user. To make FMC a success, customized tariff packages, for example reduced tariffs at home zones, separation of business and private tariffs, etc., are important as part of the marketing of FMC. 4.4.6.2 Challenges/Benefits for the Operators As technology develops, people are using an ever broader range of devices and services. This results in complexity, which, among other things, service providers and network operators have to deal with. Service providers and network operators should aim to provide any service at any access, any location and any time, with the same look and feel on any device. Supported by automatic adaptation of services to terminal/network characteristics and user preferences, this would result in a greater acceptance and use of service by end users. This would be even more supported if automatic customization of user access to services and the network were provided. First, however, operators have to overcome several obstacles (for example, extending already existing infrastructure to support fixed-mobile convergence, building a personalized environment of attractive services to increase customer loyalty). In the end services will be experienced by the user in the same lively, colourful, close and personal quality as natural communications.

Service Infrastructures

153

Another important aspect for the introduction of FMC, especially for large operators, is the expected reduction of OPEX and the service differentiations in relation to other operators. 4.4.6.3 Drivers in the Standardization The FMC target network architecture closely follows the next generation network (NGN) release 1 standards defined by the ETSI Telecoms & Internet converged Services & Protocols for Advanced Networks (TISPAN) group. ETSI TISPAN R1 selected the IP multimedia subsystem (IMS) as specified by the 3rd Generation Partnership Project (3GPP) R5/6/7 as the heart of its network. IMS is based on the session initiation protocol (SIP) and is responsible for registration, session setup and teardown, routing, etc. The focus of ETSI TISPAN NGN is to provide a standard for converged services from the fixed network operator’s (FNO) point of view, whereas the 3GPP standards were driven by mobile network operators (MNO) and equipment vendors. In addition, the 3GPP specifications for IMS were taken as the basis for the PacketCable 2.0 standards elaborated by CableLabs. FMC is essentially an integration of already existing solutions especially targeted at mobile and/or fixed networks. The big new thing introduced by FMC is the convergence of services, and the convergence of the service delivery platform (based on IMS) for any access. 4.4.6.4 Categories of Convergence Fixed-mobile convergence is a multifaceted topic. There are a number of convergence levels at which FMC can be realized, e.g. one could distinguish between service, product and network convergence. Service convergence comprises requirements by the end user, like one bill and number, and moreover same-service-offering spanning for all kinds of access networks. Service convergence leads therefore to a merger of voice, data and multimedia applications/services to provide customized service sets accessible from any type of terminal in any environment. Network operators may take their already existing identity, charging and billing infrastructures and act as providers for these service-enabling features. Other starting points for product convergence are common application servers and common session control. On the network side, a common core network comprising control, user and transport plane supporting any type of access network will offer the basis for service and product convergence. 4.4.6.5 IMS as Common Service Delivery Platform for FMC The IMS is a powerful set of network elements to process and execute any kind of customer connectivity requests independent of the terminal or access network technology, only requiring IP connectivity and an IP-based signalling protocol. It is the key technology that will help network operators and service providers to build common applications and bridge different types of access networks, such as GSM/UMTS, WiMAX, cable modems and DSL. Because IMS is access-network agnostic it is able to bring together the different types of access network mentioned above and can be used within carrier networks, enterprise networks and mobile networks. Therefore it is the appropriate basis for a common service delivery platform for FMC.

154

Technologies for the Wireless Future – Volume 3

Within the IMS world there is no major difference between voice and data calls, due to all sessions being initiated and controlled by the common SIP signalling protocol using common IP networking strategies. Based on this, IMS-controlled IP networks could be deployed for both voice and data sessions within: ž ž ž ž

mobile 2G/3G networks wireline xDSL and CATV networks new alternative wireless networks with WLAN, WiMAX or Flash-OFDM technology corporate network applications.

With the use of key mechanisms like session negotiation, session management, support of QoS and mobility management, IMS enables: ž various real-time multimedia user-to-user services, like basic IP-based access-agnostic voice (VoIP) and video services ž non real-time applications like chat, instant messaging, user-to-server and server-to-user services ž combinational services, by supporting a cooperative control of CS and PS domain-based services, e.g. support of the exchange of pictures (via an IMS-based service) during a CS call. 4.4.6.6 Unified Service Creation Environment (SCE) as an Enabler for FMC One expectation of next generation networks would be the ability to open the creation of end user services to service providers and, finally, the customers. This would enrich the number and variety of services. Another topic on the way to NGN is service creation environments (SCEs) based on propriety software. SCEs would produce a huge barrier for third-party applications and service creation by other SCEs. Service providers will expect services and service creation to happen independently of any proprietary software. A prerequisite for this is that the vendors provide open APIs based on widely implemented standards like OSA and Parlay. 4.4.6.7 Discussion and Outlook By reusing 3GPP IMS to allow carriers managing and operating networks to deliver services, which are access-technology agnostic and effectively interworking between mobile and fixed-line networks, ETSI TISPAN has defined an architecture which fulfils the requirements of FMC. This type of architecture will open the door to new access-independent services that will increase market stimulation and raise the stage for collaborative interworking between the fixed and mobile managed networks. In addition to new services, the common IMS architecture and application deployment will contribute cost savings on operation, development and network management. Both FNOs and MNOs, as well as hybrid operators, will benefit from these services and generate new revenue streams, reducing customer churn, increasing ARPU, or both. Two different levels of integration between SPICE and IMS were considered the most opportune candidates for converged architecture guidelines: tight and loose coupling with IMS.

Service Infrastructures

155

The rationale of the tightly-coupled approach is to create a service platform on top of OMA/IMS, very closely following standardization efforts. This approach would result in the following platform features: ž User identity would be handled together with the IMS system, so that OMA/IMS services could be easily used and build into the service platform. Hence, users are always IMS subscribers. ž Access to the platform would always require users to be registered with IMS, besides only using IMS interfaces Ut (XCAP (XML Configuration Access Protocol), limited to self-management of services) and Gm/ISC (SIP). ž Application servers would be implemented as IMS application servers. ž OMA/IMS service enablers would be the preferred choice. The loosely-coupled approach would result in a service platform with some support of IMS. This approach would have the following platform features: ž User identity would not rely solely on the IMS system, although IMS could be used for bootstrapping SPICE users’ identities. ž Access to the platform would not require users to be registered with IMS and would allow access by any means supported by the service, e.g. HTTP/WAP/iMode. ž Application servers could utilize a plethora of implementation technologies and interfaces, including e.g. the ISC interface for an IMS application server. ž OMA/IMS service enablers would be treated in a similar way to other systems. Their corresponding data models would be mapped to respective concepts of the platform through adaptors to guarantee the link with the related enablers. Neither solution described above fulfils completely the requirements for a flexible converged service platform, either by restricting the applicability of the platform to IMS only or by not supporting the IMS enablers in an optimal way. One solution is a compromise between the tight and loose integration, which would have the following features: ž User identity – if the user is already an IMS subscriber (IMS registered or not IMS registered), the platform would reuse the well-defined 3GPP/OMA mechanisms for user-specific activities (e.g. it reuses 3GPP GBA architecture methods for user authentication). For non-IMS users, enhancements to 3GPP/Liberty Alliance could be used. ž Application servers based on the platform would mainly behave like IMS application servers. These servers could be accessed by the users and other application servers via SIP-based and non-SIP-based (HTTP, XCAP, XML) mechanisms. In case the AS did not conform to the current 3GPP IMS specification, it would have to be ensured that the IMS was optimally supported. ž Service enablers from OMA/IMS (such as presence, messaging, POC) would be accessed by the capability enabler layer of the platform (e.g. to build new services). This reuse would ensure compatibility with recent and upcoming IMS versions. ž OMA and IMS oriented data models would be reused to achieve an optimal integration with the IMS. These data models should be used on the terminal as well as on the application server side for defining, modeling and exchanging data.

156

Technologies for the Wireless Future – Volume 3

4.4.7 Scenarios and Architecture Requirements of Vehicular Communications Vehicular communication comprises a wide array of communication scenarios ranging from today’s applications, such as voice and telematics (emergency call, breakdown call, concierge services, etc.) [88], [89], to future safety and traffic management applications. In general, this is summarized as car-to-x communication, ‘x’ denoting communication partners, situations and application domains. To only name a few, there might be car-to-enterprise and car-to-home applications scenarios comprising applications of a mostly commercial nature, but also car-to-car and car-to-trafficinfrastructure encompassing more safety- and mobility-oriented applications. The latter two can also be considered to belong to the domain of machine-to-machine (m2m) communications. Currently envisaged applications include: ž Critical safety applications requiring a dedicated frequency band, such as intersection assistance, traffic merging, forward collision warning/avoidance. ž Mobility applications, such as traffic flow improvement. The future may also hold applications from other car-to-x domains to coexist with these, leveraging synergies by sharing communications infrastructure at least on the vehicle, if not on the infrastructure side: ž Commercial applications, such as infotainment, generic Internet connectivity or vehicular-specific Web 2.0 types of service. ž Proprietary applications, such as telemetry and telediagnostics, of a more vendor-specific nature. While the latter category closely resembles ‘classic’ telecommunications services and frequently borrows from or directly applies the same technologies or standards, the former safety/mobility category of applications poses some stricter requirements. Specific characteristics, other than in wireless telecommunications, are: ž For ‘safety’ applications: ž high reliability ž low latency ž ad hoc message exchange ž trustworthy info exchange directly between peers to ensure action only based on sensor data received from certified systems. ž Privileged operation of specific applications which are prevalently public applications wielding sovereign control (like traffic light control and traffic flow management). Commercial and proprietary application requirements typically do not differ from those of other mobile applications on handheld devices. In fact, there is a growing demand for applications on mobile devices to be integrated with and made usable in an in-vehicle environment. This creates a whole new set of challenges for car-to-mobiledevice communications standards, ranging from very short range but broadband communication technologies to data representation and application layer protocols. Frequently, applications may be directly bound to the

Service Infrastructures

157

vehicle (telediagnostics) in a similar way to FOTA (firmware update over the air) or require access to critical vehicle sensors and hence require privileges such that they can only be offered by the respective vehicle manufacturer. To harmonize concepts and fill the standardization gaps for car-to-car and car-totrafficinfrastructure communication, vehicle manufacturers, first tier suppliers and research institutions joined forces in the CAR 2 CAR Communication Consortium (C2C-CC, http:// www.car-to-car.org). C2C-CC is building on the efforts of a European frequency allocation at 5.9 GHz and liaising with IEEE and ISO DSRC standardization activities. A reference model for the service architecture is given in Figure 4.40. The reference entities ‘vehicle’ and ‘roadside unit’, which could be a traffic light or sign, designate the two building blocks quite specific for safety/mobility applications, whereas the other blocks bear relevance for the operational and support-subsystem side of the architecture, up to and including interfaces to the ‘classical’ telco type of service provisioning and operational/management platforms. In the following, we present a list of requirements for the vehicle side of the service architecture, which poses some specific and unique challenges: ž It requires tight integration with vehicle sensor and driver support systems (especially for safety applications), currently an area of proprietary solutions. For service developers, this implies a heavy effort in platform-specific adaptations of the vehicle-side service artefacts; an alternative could be the open standardization of appropriate APIs. Which APIs can be made available and which architectural concepts have to be standardized is an open issue. ž To meet efficient, minimum latency realtime communication thresholds, safety features do not use the IP protocol at the network layer, whereas for other services IP is a likely candidate. Among the service definition and execution challenges are: ž Openly defined, trusted service execution runtime environment and clearly specified APIs. ž Service-specific application protocols vs. runtime artefacts; while the former allows assurance of interoperability while targeting a wider choice of implementation platforms more easily, the latter offers greater flexibility and swifter service development cycles responding to current market needs. ž Compatibility with CE (consumer electronics) service platforms and ambient intelligence approaches. Frequently, services running on a mobile device will have to adapt to the in-vehicle context. Service logic may be distributed between the CE device and the vehicle. Moreover, future vehicles may connect to service provisioning platforms currently limited to mobile devices and interact with their environment with very much the same ambient intelligence technologies as will be found in CE devices. ž HMI (human machine interface) design/control is car-centered, not application-centered. The integrity and graceful performance of all in-vehicle systems can only be ensured by a rigid engineering approach. In particular, driver distraction and other considerations call for one consistent user interface design. Service developers will have to integrate with the respective vehicle concept. It is yet unclear how this can be supported in the future by appropriate abstractions and platform technologies. ž Standardized mechanisms for service management.

158

Technologies for the Wireless Future – Volume 3

Currently ongoing activities/observations for the service architecture that start to address these requirements in research and product development are: ž Some limited exploration of open service platforms such as OSGI, for example in the GST project [90]. Definition of trust verification mechanisms and federation technologies, such as the use of Liberty Alliance and SAML standards in telematics architectures [91], [92]. ž Industry-standard operating systems start making their way into vehicles (see for example Microsoft Blue&Me). For the roadside unit, the issues and requirements are quite similar to those just discussed. One differentiator is that service announcement is a key functionality here. Specific access control and charging mechanisms will also have to be considered. Another growing challenge, implicitly mentioned above, is the integration of CE devices inside the car. Currently confined to integration for hands-free telephony, phonebook access and audio playback based on Bluetooth, this is already today an area plagued by only partial standardization and it remains challenging to ensure interoperability. Further integration would require a ‘semantic’ matching of HMI functionalities, service platform integration and federation, functional abstractions and harmonized application architectures. Moreover, semantic intelligence for largely automated information processing relieving the driver of many routine tasks (e.g. finding traffic and other information relevant to his current itinerary, presenting advance hazard notifications with sufficient relevance to the current driving situation), as well as semantic capabilities enabling new entertainment functionalities (such as adaptive music playback based on mood and atmosphere) will contribute greatly to a future context-aware vehicle service platform acting with sufficient ambient intelligence to enhance comfort and safety of future mobile travelers.

I3 Vehicle

I2

C2C Service Infrast.

I7 Road-Side Unit I8

I1

I10

C2C IPv6 Backbone

I5

I9

I12

C2C OSS I11 Telco SPP

I6 TelcoIPv6 Backbone, incl. heterog. accessnetwork

Vehicle

I4

Other Service Infrast.

Figure 4.41 Reference model for the service architecture of car-to-car and car-to-x

Service Infrastructures

159

Conclusions In the section above, we identified communication scenarios and service architecture requirements for several vehicular communication scenarios, including but not limited to car-to-car as well as car-to-trafficinfrastructure. A substantial overlap with ‘classical’ wireless scenarios does exist, although here some additional and specific requirements have been pointed out and motivated. Mobile services and applications call for increasingly integrated platforms and solutions, which are not only location and context aware, but intelligently adapt to and integrate with different usage environments and situatively interoperate with peer units (cf. CE device and vehicle service platform as one example). Therefore, one strategy for research and product solutions proposed here is to broaden the exchange and cooperation between the to date rather separate developments in the automobile, consumer electronics and telecommunications domains. Synergies can be achieved by leveraging available technologies, hence building on today’s and prospective wireless solutions and adding solutions which address the vehicle’s specific requirements. This, among automotive requirements such as reliability, should specifically address those gaps that are critical to service platform interoperability but not addressed in current standardization, such as vehicle-specific APIs, runtime environments and architectural concepts. This approach suggests a closer cooperation between the car industry and the wireless industry, in an exemplary way the cooperation between the CAR 2 CAR Communication Consortium and the WWRF, one example being the inclusion of the text above in this Book of Visions of the WWRF. 4.4.8 Conclusions In order to meet the current challenges in mobile service development, deployment and management, new standards-based mobile service architectures are needed. In this section we have covered several service platform functions that are currently seen as interesting and as enablers for a new mobile service ecosystem. These functions include the following: ž Converged communications environments supporting spontaneous ad hoc interactions between devices. ž Identity provisioning allowing third parties and visiting networks to assert the identity of a user and their subscriptions while keeping the anonymity of end users. ž Charging and billing, allowing users to easily pay for connectivity, services and goods while allowing the operators to take a small premium from each service, improved service provisioning (e.g. through push technologies, search engines or context aware service offering) and service access. ž Personalization, context-awareness and adaptation to specific situations thus making mobile and converged services easier to use. ž Virtualization, which abstracts hardware functions from software. ž Mediation and brokering, supporting the creation of service bundles. The integration of the above technologies into a coherent service platform architecture is a current topic and an integral part of IMS evolution. A number of key questions regarding the converged communications environment include the role of IMS, how web service access is realized, how security is implemented, and how component services are invoked seamlessly. As basic enablers for a converged communications environment, we have discussed reusing

160

Technologies for the Wireless Future – Volume 3

the IMS platform by using mechanisms such as session negotiation, session management, QoS support and mobility management. Additional enablers include the functions described in Section 4.4.4, including IMS-based identity management. Mobile service platforms are expected to borrow features from Internet services and so-called converged communications platforms are becoming available in the near future. As mentioned in this chapter, the open nature of the Internet and web enables rapid service creation and deployment, and, more importantly, supports rich community-driven content and service creation. These features are also vital for mobile platforms, and current platform research and development aims to bring this functionality into converged communications platforms. In addition, we started to identify communication scenarios and service architecture requirements for several vehicular communication scenarios, including but not limited to car-to-car as well as car-to-trafficinfrastructure. A substantial overlap with ‘classical’ wireless scenarios does exist, although here some additional and specific requirements have been pointed out and motivated. From the mobile platform viewpoint a crucial challenge is trust. Current mobile platforms can be trusted by the end users and service-level agreements determine different parts of service access and usage. This notion of trust is very different for current Internet services, which typically have different business models than mobile services. A large part of the revenues of Internet companies stems from advertisement fees rather than actual service usage. The elusive nature of trust on the Internet makes it difficult to create pay-by-usage mashups and to provide guarantees on service usage. It is also not easy to identify customers across services while maintaining privacy. Several service-level technologies have been developed to alleviate this issue, namely the Liberty Alliance and OpenID for enabling single sign-on on the web. Therefore, it is crucial to find ways to maintain the mobile service platform as a trusted entity for users and, at the same time, to be able to offer the flexibility of Internet services. Technologies such as 3GPP’s GBA pave the way for identity management in this kind of converged environment. A key challenge is how to develop and maintain an interoperable and federated mobile service ecosystem that meets user expectations while retaining manageability and easy charging and billing. Moreover, new tools are needed to support end users in creating new content and services. Tools for this kind of end user- and community-driven effort should be easy to use, but still flexible enough to support creativity. The number of connected devices is going to increase vastly. The mobile service platform should support different device groups and usage environments, such as smart spaces and ad hoc groups.

4.5 Acknowledgements The following individuals contributed to this chapter: Contributors to Section 4.2 are Josef Noll (University Graduate Center, UniK, Kjeller, Norway), Matthias Wagner (DOCOMO, EuroLab, Germany), Massimo Paolucci (DOCOMO, EuroLab, Germany), Erik Lillevold (University Graduate Center, UniK, Kjeller, Norway), Anna V. Zhdanova (Forschungszentrum Telekommunikation Wien, Austria) and Mohammad M.R. Chowdhury (University Graduate Center, UniK, Kjeller, Norway), with external contributions by Kashif Iqbal (DERI, National University of Ireland, Galway, Ireland) and Dumitru Roman (DERI Innsbruck, Austria).

Service Infrastructures

161

Contributors to Section 4.3 are Olaf Droegehorn (ComTec, University of Kassel, Germany), Henning Olesen (Technical University of Denmark, Denmark), Alberto Baravaglio (Telecom Italia, Italy), Anne Marte Hjemås (Telenor, Norway), Babak Farshchian (Telenor, Norway), Bernd Mrohs (Fraunhofer FOKUS, Germany), Stefan Holtel (Vodafone Group R&D, Germany), SianLun Lau (ComTec, University of Kassel, Germany), Andreas Pirali (ComTec, University of Kassel, Germany), Niklas Klein (University of Kassel, Germany), Rune Roswall (TeliaSonera, Sweden), Henrik Thuvesson (TeliaSonera, Sweden), Karim Yaici (University of Surrey, UK), Amy Tan (University of Surrey, UK), Stefano Salsano (University of Rome, Italy), No¨el Crespi (GET-INT, Institut National des Telecommunications, France), Emmanuel Bertin (France Telecom R&D, France) and Jari Porras (University of Lappeenranta, Finland). Contributors to Section 4.4 are Christophe Cordier (France Telecom, France), Josip Zoric (Telenor, Norway), Sasu Tarkoma (Nokia, Finland), Vilho R¨ais¨anen (Nokia, Finland), J¨org Brakensiek (Nokia, Germany), Robert Seidl (Siemens, Germany), Erwin Postmann (Siemens, Germany), Jorge Cuellar (Siemens, Germany), Hariharan Rajasekaran (Siemens, Germany), Matthias Franz (Siemens, Germany), Guenther Horn (Siemens, Germany), Wolf-Dietrich Moeller (Siemens, Germany), Alex Galis (University College London, UK), Kazimierz Bałos (University of Science and Technology Krakow, Poland), Hans J¨org V¨ogel (BMW, Germany) and Klaus David (ComTec, University of Kassel, Germany).

References [1] W. Kellerer, M. Wagner, R. Hirschfeld, J. Noll, S. Svaet, J. Ferreira et al. “Systems beyond 3G – Operators’ vision”, Eurescom Project P1203, December 2002. [2] Kagal, T. Finin, M. Paolucci, N. Srinivasan, K. Sycara and G. Denker, “Authorization and Privacy for Semantic Web Services”, IEEE Int. Systems, 19(4) pp. 50–56, 2004. [3] S.A. McIlraith, T. Cao Son and H. Zeng, “Semantic Web Services”, IEEE Int. Systems, 16(2) pp. 46–53, 2001. [4] UK Technology Strategy Board, “Information and Communication Technologies”, http://www.dti .gov.uk/files/file27990.pdf, April 2006. [5] V. T¨orm¨al¨a et al., “Business analysis, deployment strategy and evaluation of ASG based services”, ASG deliverable D7.IV, 2006. [6] “Semantic Web”, http://en.wikipedia.org/wiki/Semantic_Web. [7] RDF, http://www.w3.org/RDF/, 2006. [8] OWL, http://www.w3.org/2004/OWL/, 2006. [9] Web Ontology Language for Web Services, http://www.daml.org/services/. [10] Korpip¨aa¨ , J. H¨akkil¨a, J. Kela, S. Ronkainen and I. K¨ans¨al¨a, “Utilising context ontology in mobile device application personalization”, ACM International Conference Proceeding Series:Volume 83 Proceedings of the 3rd International Conference on Mobile and Ubiquitous Multimedia, ACM Press, pp. 133– 140, 2004. [11] A.V. Zhdanova, M. Boussard, P. Cesar, E. Clavier, S. Gessler, C. Hesselman et al., “Ontology Definition for the DCS and DCS Resource Description, User Rules”, EU IST SPICE IP deliverable (D3.1), 2006. [12] D. Martin (ed.), “OWL-S: Semantic Markup for Web Services”, Technical Overview (associated with OWL-S Release 1.1), W3C Member Submission, available at http://www.w3.org/Submission/OWL-S/, 22 November 2004. [13] Ontobroker 3.8. Ontoprise, GmbH. [14] Semantic Annotation of WSDL, http://www.w3.org/TR/sawsdl/. [15] Semantic Web Services Framework, http://www.w3.org/Submission/SWSF/. [16] Semantic Web Services Ontology, http://www.w3.org/Submission/2005/SUBM-SWSF-SWSO-20050909/. [17] Web Service Modeling Ontology, http://www.wsmo.org/. [18] J. de Bruijn, C. Bussler, J. Domingue, D. Fensel, M. Hepp, M. Kifer et al., “Web Service Modeling Ontology (WSMO)”, DERI Technical Report.

162

Technologies for the Wireless Future – Volume 3

[19] “WSDL-S: Adding semantics to WSDL”, white paper, http://lsdis.cs.uga.edu/library/download/wsdl-s.pdf. [20] R. Chinnici, J-J. Moreau, A. Ryman and S. Weerawarana (eds), “Version 2.0 Part 1: Core Language”, World Wide Web Consortium, http://www.w3.org/TR/wsdl20, 27 March 2006. [21] E. Dorner, J. Kopecky, C. de Saint Marie and M. Stollberg, “Revision of DIP standardisation strategy and results of standardisation efforts”, EU FP6 project DIP, D7.8, Nov 2006. [22] “METEOR-S: Semantic Web Services and Processes”, http://lsdis.cs.uga.edu/projects/meteor-s/. [23] J. Noll, F. Kileng, R. Hinz, D. Roman and M. Pilarski, “Estimating business profitability of Semantic Web Services for Mobile Users”, in S. Schaffert, Y. Sure, “Semantic Systems, From Visions to Applications”, Proc. ¨ of the Semantics, Osterreichische Computer Gesellschaft, pp. 195– 204, 2006. [24] J. Noll, “Services and applications in future wireless networks”, Telektronikk 3–4, pp. 61–71, 2006. [25] C. Lovelock, “Services Marketing, People, Technology, Strategy”, 4th edition, Prentice Hall, Englewood Cliffs, NJ, 2001. [26] C. Gr¨onroos, “Service Management and Marketing: A Customer Relationship Management Approach”, 2nd edition, John Wiley & Sons, Ltd, Chicester, UK, 2000. [27] 3GPP, “Vocabulary for 3GPP Specifications”, TR 21.905, release 7 v7.3.0, 2007. [28] K. Kuutti, “Activity Theory as a potential framework for human-computer interaction research”, Context and Consciousness: Activity Theory and Human Computer Interaction, B. Nardi. Cambridge, MIT Press: pp. 17–44. [29] Y. Engestr¨om: “Learning by Expanding”, Helsinki, Orienta Consultit, 1987. [30] L. Bannon, “Activity Theory”, http://www-sv.cict.fr/cotcos/pjs/TheoreticalApproaches/Actvity/ ActivitypaperBannon.htm, Zugriff, 1997. [31] R. Hayes-Roth and W. Tracz, “DSSA Tool Requirements for Key Process Functions”, Technical Report ADAGE-IBM-93-13B, Owego, Loral Federal Systems, 1994. [32] D. Garlan and R. Allen, “Formalizing Architectural Connection”, 16th International Conference on Software Engineering, Sorrento, Italy, IEEE Computer Society Press, 1994. [33] D.C. Luckham et al., “Specification and Analysis of System Architecture Using Rapide”, IEEE Transactions on Software Engineering, 21, pp. 336– 355, 1995. [34] D. Garlan, R. Allan et al., “Exploiting Style in Architectural Design Environments”, SIGSOFT Software Engineering Notes, 19, pp. 175–188. [35] S. Vestal, “Languages and Tools for Embedded Software Architectures”, http://www.htc.honeywell .com/projects/dssa/dssa_tools.html, 1996. [36] M. Shaw et al., “Abstractions of Software Architecture and Tools to Support Them”, IEEE Transactions on Software Engineering, 21(6), pp. 314– 335, 1995. [37] N. Medvidovic and R.N. Taylor, “A Classification and Comparison Framework for Software Architecture Description Languages”, University of Southern California, pp. 1–24, 1997. [38] W. Tracz, “DSSA (Domain-Specific Software Architecture) Pedagogical Example”, ACM SIGSOFT, 20(3), pp. 49–62, 1995. [39] P. Farb, “Word Play: What Happens When People Talk”, 1993. [40] D.C. Gause and G.M. Weinberg, “Exploring Requirements: Quality before design”, New York, NY, Dorset House Pub., 1989. [41] T. Berners-Lee, http://www.w3.org/History/1989/proposal.html. Information Management: A Proposal. 1990. [42] T. Berners-Lee, J. Hendler, et al., “The Semantic Web.”, In: Scientic American, 284(5), pp. 34. 2001. [43] N. Spivack, “The Third-generation Web is Coming”, http://www.kurzweilai.net/meme/frame.html?mainD /articles/art0689.htm, 2006. [44] G. Boehm, “Was ist ein Bild?”, M¨unchen, Fink, 1994. [45] T. O’Reilly, “What Is Web 2.0?”, 2004. [46] D.A. Norman, “The Invisible Computer: Why good products can fail, the personal computer is so complex, and information appliances are the solution”, Cambridge, Mass., MIT Press, 1998. [47] E. von Hippel, “Democratizing Innovation”, MIT Press, 2006, also available from: http://web.mit .edu/evhippel/www/books/DI/DemocInn.pdf.

Service Infrastructures

163

[48] A. Hars and S.S. Ou, “Working for free? Motivations for participating in open-source projects”, International Journal of Electronic Commerce, 6(3), pp. 25–39, 2002. [49] L. Torvalds and R.A. Ghosh, “What Motivates Free Software Developers?”, First Monday, 3(3), 1998. [50] E. Raymond, “Homesteading the Noosphere”, http://catb.org/¾esr/writings/cathedral-bazaar/homesteading/, 2000. [51] H. Guido, S. Niedner and S. Herrmann, “RP Special Issue: Motivation of Software Developers in Open Source Projects: An Internet-based Survey of Contributors to the Linux Kernel”, http://opensource .mit.edu/papers/rp-hertelniednerherrmann.pdf, 2003. [52] K. Lakhani, R.B. Wolf, J. Bates and C. DiBona, “The Boston Consulting Group Hacker Survey”, 2002. [53] R.A. Ghosh, B. Krieger, R. Glott and G. Robles, “Survey of Developers”, 2002. [54] SPICE, http://www.ist-spice.org. [55] Courtesy of the Bull JOnAS team, http://jonas.objectweb.org. [56] “OMA Service Environment (OSE 1.0)”, http://www.openmobilealliance.org. [57] “OMA Parlay in OSE (PIOSE 1.0)”, http://member.openmobilealliance.org/ftp/Public_documents/ARCH/ Permanent_documents/. [58] E. Faber et al., “Designing Business Models for Mobile ICT Services”, 16th Bled Electronic Commerce Conference, Bled, Slovenia, June 2003. [59] R. Tafazolli (ed.) “Technologies for the Wireless Future: Wireless World Research Forum (WWRF)”, John Wiley & Sons, Ltd, 2005. [60] A. Sch¨ulke, D. Abbadessa and W. Winkler, “Service Delivery Platform: Critical Enabler to Service Providers’ New Revenue Streams” World Telecommunications Congress, 2006. [61] H. Schulzrinne and E. Wedlund, “Application-layer mobility using SIP”, Mobile Computing and Communications Review, 1(2), 2000. [62] “3GPP TS 23.228”, Technical Specification Group Services and System Aspects, IP Multimedia Subsystem (IMS), Stage 2 (Release 7). [63] G. Camarillo and M.-A. Garc´ıa-Mart´ın, “The 3G IP Multimedia Subsystem (IMS): Merging the Internet and the Cellular Worlds”, John Wiley & Sons, Ltd, 2004. [64] Open Mobile Alliance, http://www.openmobilealliance.org/. [65] “3GPP TS 23.198”, Technical Specification Group Core Network and Terminals, Open Service Access (OSA), Stage 2 (Release 7). [66] The Parlay Group, http://www.parlay.org/. [67] A. Cuevas, J.I. Moreno, P. Vidales and H. Einsiedler, “The IMS Service Platform: A Solution for Next Generation Network Operators to Be More Than Bit Pipes”, IEEE Communications Magazine, August 2006. [68] A. Akkawi, S. Schaller, O. Wellnitz and L. Wolf, “A Mobile Gaming Platform for the IMS”, Proceedings of 3rd International Workshop on Network and System Support for Games (Netgames 2004), Portland, USA, August 2004. [69] “3GPP TS 23.002”, Technical Specification Group Services and Systems Aspects, Network Architecture (Release 7). [70] A.V. Zhdanova, M. Boussard, P. Cesar, E. Clavier, S. Gessler, C. Hesselman et al., “Ontology Definition for the DCS and DCS Resource Description, User Rules”, EU IST SPICE IP deliverable (D3.1), 2006. [71] D. Raz, A. Juhola, J. Serrat and A. Galis, “Fast and Efficient Context-Aware Services”, p. 250, John Wiley & Sons, Ltd, April 2006. [72] A. Galis, R. Giaffreda and T. Kanter, “Ambient Networks ContextWare”, in “Ambient Networks: Co-operative Mobile Networking for the Wireless World”, pp. 288, John Wiley & Sons, Ltd. [73] “Service Framework”, GB 924 v1.9, TeleManagement Forum, November 2004. [74] “Enhanced Telecom Operations Map (eTOM)”, v6.1, TeleManagement Forum, December 2005. [75] “Shared Information/Data (SID) model”, v6.0, December 2005. [76] P. Siering, “Realit¨atsverschiebung – Virtualisierungstechniken im Vergleich”. [77] A. Singh, “An Introduction to Virtualization”, http://www.kernelthread.com/.

164

Technologies for the Wireless Future – Volume 3

[78] P. Barham, B. Dragovic, K. Fraser, S. Hand, T. Harris, A. Ho et al., Xen and the art of virtualization, Proceedings of the Nineteenth ACM Symposium on Operating Systems Principles (Bolton Landing, NY, USA, 19– 22 October 2003), SOSP 03. ACM Press, New York, NY, pp. 164–177, http://doi.acm.org/10.1145/945445.945462. [79] I. Pratt, Xen 3.0 and the art of virtualization, XEN – Computer Laboratory Architecture, University of Camebridge, http://www.cl.cam.ac.uk/netos/papers/, 2005. [80] VMWare, Brochure of VMWare Infrastructure 3, http://www.vmware.com/pdf/vi_brochure.pdf. [81] A. Galis, S. Denazis, C. Brou and C. Klein, “Programmable Networks for IP Service Deployment” , p. 450, Artech House Books, June 2004. [82] The Ambient Networks Project, http://www.ambient-networks.org. [83] N. Niebert et al., “Ambient Networks: An architecture for communication networks beyond 3G”, IEEE Wireless Communications, 11, pp. 14–22, IEEE, 2004. [84] X. Gu et al., “QoS-Assured Service Composition in Managed Service Overlay Networks”, Proceedings of 23rd IEEE International Conference on Distributed Computing Systems (ICDCS 2003), 19–22 May 2003. [85] Z. Li and P. Mohapatra, “QRON: QoS-aware Routing in Overlay Networks”, IEEE Journal on Selected Areas in Communications, special issue on Service Overlay Networks, 2003. [86] D. Xu and K. Nahrstedt, “Finding service paths in a media service proxy network”, Proceedings of ACM/SPIE Conference on Multimedia Computing and Networking, pp. 171– 185, San Jose, 2002. [87] M. Johnsson, A. Schieder and R. Hancock, “Ambient Network System Description”, AN project deliverable, FP6-CALL4-027662-AN P2/D07-A2, 2006. [88] http://www.bmw.com/com/en/insights/technology/connecteddrive/overview.html. [89] CAR 2 CAR Communication Consortium (C2C-CC): “Manifesto – Overview of the C2C-CC System”, version 1.0, http://www.car-to-car.org, 21 May 2007. [90] http://www.gstforum.org. [91] H.-J. V¨ogel, M. Smirnov, J. Kane, I. Kulp and A. Petrou, “Aspects of End-to-End Security for Open Telematics”, Proceedings, IST World Congress, San Francisco, Nov 2005. [92] http://www.projectliberty.org/liberty/adoption/travel_transport/bmw.

5 The WWI System Architecture for B3G Networks Edited by Andreas Schieder (Ericsson GmbH, Germany), Elias Tragos (National Technical University of Athens, Greece), Andrej Mihailovic (King’s College London, UK), Jukka Salo (Nokia Siemens Networks, Finland) and Jan van der Meer (Ericsson Telecommunicati, The Netherlands)

5.1 Introduction The Wireless World Research Forum aims to shape the global wireless future, in which the development of a future network structure plays an important role. The purpose of this chapter is to present such a network architecture vision based on the contributions made by and agreements achieved between five large-scale integrated research projects funded within the European Union’s 6th Framework Programme. The contributing projects are WINNER [1], Ambient Networks [2], MobiLife [3], SPICE [4] and E2R [5], which gather in the Wireless World Initiative (WWI) [6]. Developing a reference architecture for future networks requires carefully identification of guiding design principles as intrinsic aspects of the reference model. These architectural principles represent invariants, which need to be carefully chosen to provide sufficient structure to implementations of future network architectures. On the other hand, the model needs to provide enough flexibility and freedom to ensure its future applicability and extensibility. In general, we need to acknowledge that the fixed and mobile networking businesses are becoming mature industry segments, which results in specific requirements being put on the architectures of the telecommunication networks deployed. The most important aspects address the efficiency, extendibility, feature richness, openness and usability. These requirement areas move into focus when the technology has matured such that the intended services are provided to the customer in a basically satisfactory manner. This is the case for fixed and mobile networking, which is a technology area that has moved from its revolutionary phase into an evolutionary phase. Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

166

Technologies for the Wireless Future – Volume 3

These high-level considerations on a future network architecture have led to a couple of architectural principles, which are shared by the five projects gathered in WWI. These principles are discussed in the following section.

5.1.1 Key Architectural Principles A general trend observable in many modern network architectures is a horizontally layered structure adopting the principles proven successful in the design of telecommunication protocols. A layered structure ensures a decoupling of functional areas and allows the reuse of components as well as a shared usage. While a layered structure is generally acknowledged to be an appropriate design choice, the number of layers and their scope is an important decision to be made for the network reference model we are heading for. The WWI has agreed on four distinct functional layers, complemented by a vertical reconfiguration plane as shown in Figure 5.1, which coexists with the four layers and enhances the control and management functionalities. The reference model assumes different radio access technologies to be realized in the lowest layer as well as a network control layer, a service layer and an application layer placed on top of it. Moreover, the architecture enables the dynamic reconfiguration of each layer through the vertical reconfiguration plane spanning all four layers of the reference model. In general, WWI aims at integrating various radio access technologies in a common framework, ensuring the cooperation of diverse access and core networks and hiding the heterogeneity on the network level to services and applications. The focus of the WWI model is restricted to a scenario that assumes the presence of all project functions contributed to these layers. We thus assume the presence of a RAN as specified by the WINNER project, an ambient control space as specified by the Ambient Networks project, a service layer as detailed by MobiLife and SPICE, as well as the end-to-end reconfiguration enablers and functionalities in the system architecture contributed by the E2 R project.

Application

Application Interface

Cognitive Service Provision

Service Layer

Ambient Network Interface

Decision Making and Reconfiguration Management Legacy Control

Ambient Service Interface Ambient Control Space

Ambient Resource Interface

Reconfigurable RANs WINNER RAN legacy RAN

…

E2R Reconfiguration Plane (Management and Control)

Application

Application

WWI component

…

WINNER RAN legacy RAN SelfConfiguration and Selfnon-WWI/legacy component Management

Figure 5.1 Layered structure of the WWI reference network architecture

The WWI System Architecture for B3G Networks

167

A specific consideration in the WWI architectural principles is given to the Ambient Networks and E2 R projects in the following text because their scope and relevance extend beyond a single layer in the WWI reference model. Hence, their solutions provide a broad range of interoperability and enhancing features related to the coexistence with and synergy between the WINNER, SPICE and MobiLife projects, as well as with the underlying legacy systems. In the following text we lay out the essential architectural approaches of AN and E2 R in their relations to the WWI reference architecture. 5.1.1.1 Reference Points Connected to the layered structure of the WWI reference architecture is the presence of some major reference points located between the functional elements. The ambient network interface (ANI) connects the ambient control spaces (ACS) of different networks (inter-ACS communication), and may also be used to facilitate interaction between functions residing in the same ACS (intra-ACS interaction). The ANI is a reference point either located between different networks or between functions belonging to the same ACS. In the former case, this reference point is made out of a set of interworking interfaces, which will be selectively used based on the purpose of the interworking between the networks. When used between domains the reference point is prescriptive, i.e. it mandates the way in which the objects providing the interfaces are built. When used intradomain, the ANI can be declared to be prescriptive, e.g. for efficiency reasons, however this is irrelevant for the outside of that domain. Otherwise, the ANI is descriptive and thus does not constrain any implementation. The ambient service interface offers connectivity and control functions for use by upper layer applications within a node operating in a network, or for use by functions residing in the service layer. It allows applications and services to issue requests to the ACS concerning the establishment, maintenance and termination of end-to-end connectivity between functional instances connecting to the ASI. The ASI also includes management capabilities and means to make network context information available to applications and services. The ambient resource interface is the reference point through which functions of the ACS control and use services of the underlying connectivity. The ARI is located inside a node and also provides an abstraction of the resources in the underlying connectivity infrastructure. The ARI encapsulates the capabilities of the underlying infrastructure into abstracted information. Because these abstractions are not tied to any specific network technology, the ACS functions are portable between different network types; this decoupling enables a simple migration path from current systems to the WWI system. A main requirement for the ARI is to offer a standard functionality independent of the nature of the underlying technologies. In the context of the E2 R project, reconfigurability, defined as denoting the capability of a system that can dynamically change its behaviour, tackles the changeable behaviour of wireless networks and associated equipment, specifically in the fields of radio spectrum, radio access technologies, protocol stacks and application services. E2 R features are abstracted in the E2 R reconfiguration plane in Figure 5.1, with management and control functionality involved with all four layers of the WWI reference architecture, with specific focus on RAN reconfiguration aspects, as depicted. The Cognitive Service Provisioning Interface in the overall WWI reference model abstracts the E2 R system architecture functionality of the cognitive service provision as the interface to the service layer. The cognitive service provision module undertakes

168

Technologies for the Wireless Future – Volume 3

the necessary content control and service adaptation procedures; finally, after a) sensing the environment and acquiring contextual information and internal status, b) negotiating and deciding the concrete reconfiguration actions, and c) implementing the equipment reconfiguration along with possible adjustment of radio network resources and network elements, the service and content provided to the user may need to be adapted as well. The Decision Making and Reconfiguration Management Interface represents the E2 R interactions with control layers where the actual decision making and reconfiguration management module in E2 R produces and evaluates dynamic policy rules prescribing a set of high-level limitations defining the system behaviour in terms of users and applications requirements, resource availability and business objectives. In addition, it exploits self-knowledge and contextual information in order to formalize the autonomic behaviour of the reconfigurable equipment or network element and produce definitive reconfiguration decisions. Finally, it orchestrates mobility-related signaling, controls reconfiguration sessions, and participates in end-to-end negotiation signaling in use cases involving reconfiguration. Regarding the interfacing of E2 R reconfiguration plane interactions with RAN layers, the Self-configuration and Self-management Interface abstracts the functions of the self-configuration and management (SCM) module, which performs protocol and cross-layer reconfiguration-mode switching, undertakes the self-optimization of local resources, and includes self-healing capabilities. Besides, it orchestrates software upgrade procedures: specifically, it undertakes the actual transfer of downloaded software, the reliable mode switching from many-unicast to multicast and broadcast download modes, as well as local post-download procedures. In addition, it accommodates access and security control mechanisms and generates reconfiguration charging records. The Application Interface makes it possible for applications to access resources on the service layer. Resources are services providing either the required functionality (e.g. sending of instant message and handling related charging) or access to network resources. Both methods are subjected to the access control policies defined for the user of the application interface and the accessed resource. When accessing a resource of the type service, all functions for SPICE service are available as brokering, composition, etc. When the application interface is used on behalf of an end user, where end user-specific data is required on the SPICE platform, the end user must log in to the SPICE platform to handle the appropriate authentication. 5.1.1.2 Main Functional Elements This overall general WWI reference architecture is complemented by a set of three specific functional architectures addressing key functions foreseen for B3G networks. These functions are: ž Heterogeneous radio resource management (HRRM) – refers to the coordination of several parallel-deployed radio access networks to best utilize the resource usage. ž Mobility management – refers to the techniques that handle the mobility of the user as he/she is moving between the cells of one network or as he/she is moving between different networks. Traffic balancing and handover are the main techniques (related to the mobility of the users) used in order to efficiently exploit the resources of a network. ž Context awareness – a crucial feature for the services, applications and networking functions of the future wireless communications. According to Dey and Abowd [13], context is

The WWI System Architecture for B3G Networks

169

any information that can be used to characterize the situation of an entity, where an entity can be a person, place or object that is considered relevant to the interaction between a user and an application, including the user and application themselves. These functions are described in the following sections.

5.2 Heterogeneous Radio Resource Management (HRRM) in the WWI System Architecture 5.2.1 Introduction and Motivation Wireless communications utilize radio resources, which typically are limited by physical properties, regulatory requirements (for example spectrum bandwidth and emission powers) and economical constraints. Radio resource management (RRM) concerns control mechanisms to best utilize these resources. For efficiency reasons, RRM is usually fairly tightly coupled to the design of the wireless system it controls. The heterogeneous radio resource management (HRRM) functionality in the WWI architecture targets resource usage efficiency across multiple accesses and faces additional challenges arising from the heterogeneity of the access technologies (in terms of e.g. data rates, coverage, services and existing RRM functionality) and the range of business models to be supported. 5.2.2 Synthesis of a WWI Architecture for HRRM Realizing HRRM in a network scenario built on the assumption that radio access is provided by various heterogeneous access networks leads to the need to structure the overall network into an access-independent stratum and an access-dependent stratum. This conclusion is supported by all WWI projects, where HRRM represents the fusion of radio resource management functionalities in each involved project (AN, E2 R and WINNER [10], termed as MRRM, JRRM and CoopRRM, respectively). In addition, one can identify a set of functions which need to be present in these two domains to jointly implement the HRRM feature. Our analysis has led to an identification of 12 functions. The purpose of each and whether they are allocated to the access-dependent or access-independent stratum are discussed below. 5.2.2.1 Access-independent Functions ž Interaccess selection – this is a generic function that enables the selection of an access resource for scheduling, handover or load balancing purposes. ž Intersystem access control – this is a collection of functions for control of the heterogeneous network usage. These functions include load balancing and congestion control amongst networks, intersystem admission control, intersystem handover decision, scheduling and QoS management. ž Interaccess handover execution – provides the toolbox to release the current resource and move the call on the target-selected resource within a different access network. ž Reconfigurability – coordinates the resulting end-to-end actions in reconfiguration scenarios and, in addition, facilitates network adaptations and spectrum allocation issues.

170

Technologies for the Wireless Future – Volume 3

ž Path selection – takes into account the end-to-end characteristics of user plane sessions and extends the decision about the access links by functions to select suitable core network paths. The selection takes into account QoS capabilities and security information such as knowledge about the presence of firewalls or private networks. 5.2.2.2 Access-dependent Functions ž Radio monitoring – this function enables reporting on the state of current and neighbouring access networks. ž Access advertisement – this function enables advertising of the identity and the capabilities of the networks over beacons or other means. ž Handover execution – provides the toolbox to release the current resource and move the call to the target-selected resource within the same access network. ž Access discovery – this function enables the terminal to discover the available networks. ž Spectrum management – this function allows spectrum resource sharing and reassignment between access technologies and operators targeting at a joint optimal resource usage. Benefit for the overall system is the trunking efficiency and load balancing. ž WINNER deployment mode selection – as mentioned in [11], WINNER has different operational deployment modes. This function is an internal WINNER function responsible for selecting the target WINNER deployment mode for the users making a handover from legacy networks to WINNER or for new user admission in the WINNER RAN. ž Reconfigurability – performs specific reconfiguration actions in the access and terminal parts involving context, QoS, configuration analysis and invoking, profile and policies processing. All functions mentioned above jointly implement HRRM and ensure the cooperation between heterogeneous radio access networks. In summary, a picture as shown in Figure 5.2 can be constructed to visualize the allocation of the different functions to the two stratums and the radio access network types considered in the WWI context. Access Independent Stratum Path selection

Reconfigurability

inter access selection Inter-system access control Handover Execution

RANx Radio Monitoring

WINNER WINNER deployment mode selection Radio Monitoring

Access Advertisement Access Advertisement Handover Execution Handover Execution Access Discovery Access Discovery Spectrum Mgmt. Spectrum Mgmt.

reconfigurable RANx Reconfigurability Radio Monitoring Access Advertisement Handover Execution Access Discovery Spectrum Mgmt.

Figure 5.2 First synthesis of a WWI system architecture for the provisioning of multi-radio resource management

The WWI System Architecture for B3G Networks

Bearer Mgmt. Functions

Policy Related Functions Security QoS …

171

HRRM

Radio Monitoring

Access Detection

Access Discovery / Monitoring

Detected Set DS of Accesses Policy Constraints

Access Validation Validated Set VS of Accesses

Data Bearer

Access Control

Request Policy Constraints

Candidate Set CS of Accesses

Access Performance and Characteristics

Access Selection Active Set AS of Accesses

Figure 5.3 Managing different sets in the HRRM

5.2.3 HRRM Decision Making HRRM is responsible for the selection of an appropriate access network or access link to either satisfy a new connectivity request or to ensure maintenance of an already established connectivity session. The decision on what access to choose is made by executing a sequence of selection steps as depicted in Figure 5.3, which was originally published in [11]. Each of the steps in this sequential process leads to a different set of potential access networks or access links and eventually to the selection of one final access network. The Ambient Networks project has defined four different sets, which are adopted by HRRM. According to [12], these sets are defined as: ž Detected set (DS) – this set includes all accesses detected by the mobile terminal either through scanning for available accesses or by listening to a pilot channel broadcasting information about all available accesses. The detected set is maintained by each mobile terminal individually and independent of any current request for connectivity. ž Validated set (VS) – the access options maintained in this set are a subset of the accesses present in the DS. Only access networks satisfying certain policy constraints are accepted into the VS. Such policies can include security, QoS requirements and cost restrictions. ž Candidate set (CS) – the CS contains a subset of the access options present in the VS. The candidate set is only created after a connectivity request is received and the requirements for the characteristics of the wanted connectivity are known. Access networks capable of satisfying the requirements and possessing sufficient free resources are accepted into the CS. Candidate sets are maintained for each end-to-end connectivity session. ž Active set (AS) – the AS is constructed from the CS by applying policy constraints and considering load and performance characteristics of the accesses in the CS. The accesses in the AS are the ones eventually used to implement a particular end-to-end connectivity session.

172

Technologies for the Wireless Future – Volume 3

5.2.4 Use Case Examples We can distinguish two fundamentally different approaches to implementing HRRM. First, functions in the access-independent and access-dependent stratums may coordinate the usage of different radio access systems to finally ensure that a particular user is served by the best-suited access network. The access networks considered in this cooperative scenario are supposed to deploy different technologies, but are aligned to such an extent that they can interwork and cooperate on the control plane. On the other hand, reconfiguration allows the access networks and/or user terminals to ensure that they always deploy the best-suited technology for the current usage scenario and/or adaptation of networks’ operating and/or service delivery conditions. Variants of the combinations and integration options are explained in the following text by focusing on the explanation of relevant use cases. The terminal and network cooperation in selecting access networks can be summarized to be provided by the WWI access-independent functions of the HRRM (see Figure 5.4). We distinguish four basic types of access network for the selection process: ž Legacy RANs such as the 3GPP UTRAN or IEEE WLAN and WiMAX networks. ž Reconfigurable legacy RANs, which allow the deployment of different radio access technologies on the same physical hardware and in a shared spectrum. ž Future RANs, represented by the RAN specified by the WINNER project. ž Future reconfigurable RANs. The reconfiguration of a WINNER RAN is not yet specified, but it is included in the figure as an optional case or as an example of a possible reconfigurable future wireless network. The implementation of the HRRM functions and the interactions between the access-independent functions (HRRM and RCM) are described in the following section by means of a representative set of procedures covering the following use cases: ž HRRM selecting and invocating the WINNER RAN (case 1). ž WINNER RAN and reconfigurable RAN HO request to HRRM for intersystem HO (case 2). ž Spectrum coordination between WINNER RAN and reconfigurable RAN (case 3). ž HRRM/RCM radio reconfiguration of reconfigurable RANs (case 4). ž HRRM/RCM request for access to additional operator resources (case 5). ž Reconfigurable terminal roaming with reconfiguration from one type of network to another (e.g. from Winner RAT to a reconfigurable UTRAN) (case 6). 5.2.4.1 HRRM Selecting and Invocating the WINNER RAN HRRM is assumed to coordinate the general access network evaluation and selection of radio accesses, while the WINNER RAN also includes radio resource management functions itself to directly coordinate with legacy RANs. When a new radio access is selected, requiring movement of an active session to/from a RAN that is not under direct control of the radio resource management functions in the WINER RAN, the radio access will have to rely on the HRRM control function. The WINNER RAN is treated as any other RAN and included in the process of access selection, as discussed in Section 5.2.3. The access selection procedure

The WWI System Architecture for B3G Networks

173

WWI Access Independent Functions RCM

HRRM

WT

UT

Reconfigurable WiMax

WiMax

Reconfigurable WLAN

WLAN

Reconfigurable UTRAN

UTRAN

Legacy RANs

Reconfigurable Winner RAN

Winner RAN

Future RANs

Spectrum owner

RT

RCM: Reconfiguration Control and Management WT: WINNER Terminal UT: UTRAN Terminal RT: Reconfigurable Terminal

Figure 5.4 Range of HRRM use case deployments

is triggered by a flow setup request that is, for example, received from a client application running on the user equipment. The access selection function in the core network decides on the access network based on the information provided by the mobile terminal in the validated link set (see Figure 5.5). In the scenario presented here, the WINNER RAN is selected and the access control and access selection functions in the WINNER RAN are triggered to decide upon the best suitable target mode to be applied (see [1] for a discussion of WINNER RAN modes). When the WINNER RAN mode is selected the radio resources are seized, the active link set is updated and the access flow is finally established. 5.2.4.2 WINNER RAN and Reconfigurable RAN HO Request to HRRM for Intersystem HO The second use case targets the handover of an active session from a WINNER RAN to a second RAN (which is not a WINNER RAN and may be reconfigurable or not). As this is a typical intersystem scenario, the functions in the common core network are mostly concerned

174

Technologies for the Wireless Future – Volume 3

User Equipment Access Selection

Common Core Network

WINNER RAN Access Control

Access Control

Access Selection

Access Selection

Access Control

Flow Setup Request Access Flow Setup Request Request validated Link Set Validate Detected Link Set

Updated validated Link Set Access Resource Request

Build Candidate Set

Target Mode/BS request Target Mode/BS response Access Resource Response

Seize Access Resource Active Link Set Establish Access Flow Attachment Request Attachment Reply Establish Access Resource Access Resource Established Access Flow Established Flow Setup Response

Figure 5.5 Access selection procedure selecting and invocating the WINNER RAN

with the realization of this procedure (see Figure 5.6). The handover scheme employed in the WWI network model relies on the assistance of the mobile terminal to provide information about the access quality (e.g. C/I ratio) to the access selection function in the core network. Based on this information and the core network’s knowledge about the end-to-end path, the access selection function in the core network eventually initiates a handover request. The first step in the execution of the handover is the selection of the appropriate handover tool, as we operate in a multitechnology environment and different access networks typically employ different solutions and protocols to execute a handover. The following actual handover execution follows a make-before-break scheme controlled by the handover execution function of the core network. 5.2.4.3 Spectrum Coordination between WINNER RAN and Reconfigurable RAN HRRM also includes functions to reconfigure radio spectrum usage. Two reconfiguration cases need to be considered: first, spectrum can be assigned to the WINNER RAN to satisfy an increased demand in the WINNER RAN; second, spectrum can be reassigned between reconfigurable legacy RANs to optimize spectrum usage for a particular usage scenario. Both spectrum configuration cases could target the same spectrum and thus lead to contention. There is thus a need for the spectrum owner/regulator to specify resolution rules and perhaps also include a third-party spectrum management provider. The resolution activity is indicated in Figure 5.4, showing HRRM interacting with the spectrum owner (case 3).

The WWI System Architecture for B3G Networks

User Equipment Access Selection

175

Common Core Network

WINNER RAN Access Control

Access Control

Access Selection

Access Selection

Access Control

Flow Setup Request Access Flow Setup Request Request validated Link Set Validate Detected Link Set

Updated validated Link Set Access Resource Request

Build Candidate Set

Target Mode/BS request Target Mode/BS response Access Resource Response

Seize Access Resource Active Link Set Establish Access Flow Attachment Request Attachment Reply Establish Access Resource Access Resource Established Access Flow Established Flow Setup Response

Figure 5.6 Intersystem handover from a WINNER RAN towards a second (reconfigurable) RAN

The spectrum sharing use case can be split into two cases, when the involved RATs in the shared frequency band have equal regulatory status (horizontal sharing), i.e. no system has priority over the other(s) in accessing the spectrum, or when the spectrum sharing is performed with clear established priorities (vertical sharing). In the second case, the primary RAT has a preference in accessing the spectrum and the secondary RAT(s) may only use the spectrum as long as they do not cause harmful interference to the primary. The horizontal sharing can be split into two cases, namely without or with coordination. In the case where there is no coordination, radio access systems using two or more RATs operate in the same frequency band without the possibility of jointly coordinating their spectrum access. This means that neither system is aware of the location and the current state of the other systems. No signaling is possible between the involved systems, and without the signaling it is generally not possible to prevent interference. However, the only possible way of restricting the effects of mutual interference consists in a high attenuation between the interfering transmitter and the victim receiver and a sufficiently low transmit power. This can be achieved by spatial separation and/or directional antennas. In the case of the horizontal sharing with coordination, the involved radio access systems coordinate their spectrum access based on a set of predefined rules (i.e. spectrum etiquette) that all RATs adhere to. This requires capabilities for signaling or at least detection of the other RATs.

176

Technologies for the Wireless Future – Volume 3

The vertical sharing can also be split into two more cases. In the first case, the WINNER RAT is the primary RAT and the spectrum is dedicated to the WINNER RAT. In this case, no special requirements for the WINNER RAT to coexist with other, secondary RATs exist. The other RATs are allowed to be used only as long as they do not generate interference to any WINNER receiver. The WINNER system can (but is not obliged to) assist the secondary RATs by signaling the free spectrum resources via its broadcast channel while keeping the control of the spectrum utilization. In the second case, the WINNER RAT is the secondary RAT. The WINNER RAT has to control its emissions (at the base station (BS) and for all UTs) in order to avoid interference to the primary RAT. This requires considerable knowledge of the deployed primary (legacy, non-WINNER) RAT. In this case, some parts of the WINNER system (i.e. short-range cellular, peer-to-peer and feeder links) operate in a frequency band which is assigned to a primary RAT, possibly a legacy RAT. The considered WINNER mode has to operate in such a way that the interference to the primary RAT is prevented. The difference from the horizontal sharing scenarios is that the primary (legacy) RAT has preference in accessing the spectrum and the secondary (WINNER) RAT is not allowed to generate interference, whereas the primary RAT is allowed to interfere with the secondary RAT. The challenge for the secondary RAT, i.e. the WINNER RAT, is to make best use of the ‘white spaces’ in time, frequency and space, and to transmit in these white spaces only, i.e. without generating harmful interference to the primary RAT. In order to do so, the white spaces have to be identified first and then the transmission parameters have to be adapted accordingly. While the adaptation step is a complex but nevertheless well-defined optimization problem, the identification step is much more difficult to handle and can possibly consist of several components: 1. 2. 3. 4.

download from a database (e.g. regulator or another authorized entity) information retrieval from a central radio controller information offered by the primary RAT real-time measurements

Components 3 and 4 could come from HRRM, and components 1 and 2 could be provided in cooperation with E2 R. A combination of these components would give more reliability to the identification process. Naturally, this reliability depends strongly on the properties of the primary RAT. It is much easier to identify the white spaces if the primary system is static in time, frequency and space (i.e. spatial location). 5.2.4.4 HRRM/RCM Radio Reconfiguration of Reconfigurable RANs When available radio and network resources are limited or specific access is required by user preferences, there are two approaches to increase performance or to meet the user preferences: either evaluate access options, i.e. use HRRM and the reconfiguration control management (RCM) to find additional spectrum (case 4), or launch dynamic roaming agreements through the ambient networks composition process to find additional networks and access options (case 5).

The WWI System Architecture for B3G Networks

177

5.2.4.5 Reconfigurable Terminal Roaming with Reconfiguration from One Type of Network to Another (e.g. from Winner RAT to a Reconfigurable UTRAN) In case of additional radio resources are needed to satisfy the real-time service delivery levels, operators can provide them via the reconfiguration functionality. The actions can include spectrum allocations, dynamic network planning, dynamic upgrading of network components, self healing actions, etc. The use of reconfigurable terminals further provides for the uninhibited roaming of reconfigurable terminals to arbitrary networks (case 6). More specifically, when a terminal wishes to enter into a new network and it lacks a specific set of protocols, it is provided with all the mechanisms to identify which protocols are needed, where to find them, how to download them, how to become reconfigured in real time and in a seamless way, and how to verify the reconfiguration actions.

5.3 Mobility 5.3.1 Introduction and Motivation Mobility management techniques support user mobility, including the traffic balancing which is essential for a network to use efficiently the resources of the system. In the future, wireless networks will be able to cooperate in order to provide the user maximum QoS, and mobility management is one of the key mechanisms for that. In the WWI system architecture, mobility management will ensure user mobility in heterogeneous access environments. There are many issues related to mobility that are investigated in the scope of the WWI architecture. The main goal is to define a global WWI functional architecture for mobility management by analyzing and combining the architectures of the different projects, which each deal with mobility from their own perspectives. This will be done by the unification of the mobility requirements and related issues, or to put it another way, by extraction of common and ubiquitous mobility management requirements applicable and present in all three projects, with the ambition of promoting them as essential mobility issues for emerging telecommunication environments. 5.3.2 The Functional Architecture Mobility management in the WWI concept is related to mobility management between different access networks, since all the projects are working on enabling cooperation of networks, but each from a different perspective. For example, WINNER is working towards making the new network able to cooperate with the legacy networks. Some of the common requirements for mobility management in all the projects are: ž Handover – fast and seamless handover execution and need for seamless reconfiguration for the execution of the handover. ž Triggers – either for handover and/or for general mobility actions, triggers have been acknowledged in all the projects. The actual triggers vary by and large and can include physical layer-related triggers, context information, user preferences and profiles, location

178

ž ž ž

ž

Technologies for the Wireless Future – Volume 3

information, more implicit cases such as adaptation/monitoring of service to system changes, and intelligent triggers. Context and location awareness – the availability of context and location information to the mobility management processes is another requirement that, if utilized, expedites and improves important performance criteria. Planning – this relates to the actions that are performed in response to determination that a handover is imminent or in an execution stage. The requirement also demands a fallback to unplanned handovers in case a planned handover fails. Coping with legacy systems and multiple operator deployments – support for different technologies for vertical handover could relate to seamless execution of reconfigurations for terminals attempting to match to the spectrum and network standard. One extreme case is network reconfiguration due to mass handover requests from terminals with the same radio mode. For example, terminals should be able to become reconfigured when entering into a new access network that belongs to either the same or a different operator. The need for reconfiguration arises from the fact that different networks, especially those belonging to different operators, may use a different set of mobility management, RRM and QoS protocols. Thus, the user equipment needs to be able to be adjustable to the new environment in a seamless way. Obviously, this requires the ability to discover early any incompatibilities between the current configuration and the protocols used by the new network, to locate and download the appropriate set of protocols, and to reconfigure the protocol stack appropriately. Mobility as diverse architectural component – this includes all of its aspects: session/application, traditional, resource expenditure, load balancing, multi-homing, security, all variants of mobility: endpoints, sessions, flows, interfaces, network/groups, flexible spectrum utilization, group and signaling management [7].

A WWI functional architecture for the handover process part of the mobility management is shown in Figure 5.7. The different functional entities of all the projects related to inter-RAT handover are presented, as well as the connections between these entities. This figure is separated into layers, starting (from bottom to top) at the physical layer and ending at the application and service layers. It provides a good example of the inter-RAT handover process and how the entities of the projects could cooperate in a unified way to handle an Inter-RAT handover. The central entity is the global RRM entity (named HRRM, as it was described in Section 5.2) that is responsible for the RRM decisions, the RAT selection, the load balancing, etc. This figure also shows many triggers for an inter-RAT handover, either from the network layer (from SRRML or SRRMW) or from the service layer (from service or context information). The newest addition is a vertical cross-layer entity related to fuzzy logic, which is described in the next section. 5.3.3 Fuzzy Logic-based Heterogeneous Mobility Management In this section, the adaptive multicriteria handover algorithm in the WWI concept is presented [7]. The intersystem FLC is responsible for the handover decision. It applies the predefined rules to the current system conditions. A neural network learns the FLC parameters from the resulting handover quality indicators.

The WWI System Architecture for B3G Networks

Sends terminal capabilities

Application open XML-based Interface

Context Info.

Service Info.

Fuzzy Logic based Triggering - Mgmt

Consults for handover triggering

Handover Indication Req. Handover Initiation Service Adaptation of reconfiguration Handover Decision Making

Service Layer Autonomic Decision

Service Enabler

Making handover decisions, Admission Control, RAT selection, QoS, etc

Common Control Layer

Context Info.

Consults for IIO

179

TC

HRRM

Sends reconfiguration on RC capabilities Reconfiguration Control Actions

RCM

Consult

Handover Ctrl. Performance Measurement Load Measurement Handover Req. from SRRMW, SRRML Handover Exec. Order from WWI-RRM

Access & security mgmt

Consults for security

UTRAN

WINNER-RAN BS/RRM Server Access Layer Self Configuration

SRRML Layer reconfiguration

Layer reconfiguration

Access Layer Self Configuration

Figure 5.7 Example of functional architecture for the inter-RAT handover process

A preliminary selection of HO targets is performed before the vertical handover procedure. Targets with signal level and load above thresholds are filtered and then the target with the best signal level is chosen. This target preselection reduces the FLC complexity and saves the processing time. The considered handover criteria are signal strength measurements, load information and UT velocity. As they are presented in the Figure 5.8, the triggers for the handover are based on the WWI triggering framework, which considers the mobility triggers from all the projects.

SPICE Triggers

E2R Triggers

AN Triggers

MobiLife Triggers

WINNER Triggers

Handover Failure

WWI Triggers

Handover rules

WWI HO Control FLC

Measurement reports

Handover procedure

Quality indicators

NN based learning

execution

MT velocity Optimized FLC parameters

Figure 5.8 Fuzzy multicriteria vertical handover scheme

180

Technologies for the Wireless Future – Volume 3

Neural networks (NN) have been applied successfully to solve complex problems by automatically learning the behaviour of a system and generalizing it to situations not experienced before. A neural network is used in order to determine the optimal FLC parameters whenever the system conditions change. It learns the behaviour of the multisystem network from a database of simulation results. NN training is very sensitive to learning rate (LR). If it is too high, the training algorithm may be unstable. If it is too low, the convergence will be too slow. Therefore, we use an adaptive LR. The LR adapts to the error evolution during the training process: if the error increases, the LR decreases, and vice versa. Moreover, a momentum is added to back-propagation learning in order to take into account the last weight in the computation of the new weight. The addition of a momentum prevents the back-propagation algorithm from being trapped in a local minimum. The weight correction 1wij applied to the weight of the connection between neuron i and neuron j at iteration n is defined by the following ‘generalized delta rule’: 1w ji .n/ D Þ1w ji .n
1/ C
Ž j .n/yi .n/ n: iteration number Þ: momentum constant
: learning rate parameter Žj: local gradient for neuron j (depends on the nature of neuron j) yi (n): output of neuron i at iteration n

5.4 Context Provisioning 5.4.1 Introduction and Motivation Context awareness is a crucial feature for the services, applications and networking functions of the future wireless communications. According to Dey and Abowd [13], context is any information that can be used to characterize the situation of an entity, where an entity can be a person, place or object that is considered relevant to the interaction between a user and an application, including the user and application themselves. The frameworks for context awareness must provide the context-aware applications with relevant information (where relevancy depends on the user’s task) which can be used to adapt the behaviour of these applications to the user’s situation. The personalization aspect of context awareness adjusts services to the specific circumstances and needs of a certain person. For example, one probably needs different services while working at the office by the desk and while walking in the street on holiday. A system that knows about the person’s situation and activities can be very helpful, not only by providing useful information and services but especially by filtering out those that are unnecessary. Typical context data include spatial information (location, speed, orientation), temporal information (time, duration since an event), environmental information (temperature, light, noise level), user characterization (activity, social surroundings) and resource availability. Expanding context information a little beyond the user’s space makes us consider the network level aspects of context awareness, used for two main purposes. The first is the automatic adaptation of the behaviour of network services to suit the network’s situation. The other is the adaptation of end user applications to account not only for user context but also for the relevant information about status and capabilities of surrounding networks within a user’s reach.

The WWI System Architecture for B3G Networks

181

Reconfiguration of routing media overlays or virtual private networks serving a number of users, dynamic instantiation of functionality to remove bottlenecks or address the need of a community of users, and traffic redirection for load balancing across heterogeneous networks are examples where the dynamic adaptation of networks services is enabled. More intelligent and informed handovers for mobile users and optimized delivery of telecommunication services are other examples showing what network context awareness can enable. 5.4.2 The Usage and Ontologies of Context Information To manage complexity and account for different requirements at different layers, the context provisioning-related architectures have been studied separately for different communication layers. The scope of the ‘context provisioning’ aspect of the WWI system architecture is to bring these solutions together, highlighting the interfacing and interworking challenges. 5.4.2.1 Usage of Context Information Context information can be classified as follows: ž Human user/group context – characterizes the situation and circumstances of a user or user group, including their moods, activities/presence, preferences, available devices, environmental information, temporal information, spatial information, etc. ž Device context – characterizes the features of the device, including for example screen size and resolution, computational power, battery power and SW capabilities. ž Network context – characterizes the resources of the networks and their status, including access types, coverage, bandwidth, supported QoS, supported security functions, etc. ž Flow context – characterizes the features of the flow, including the state of the links and nodes that transport the flow, such as latency, jitter, loss, error rate, etc. ž Service context – characterizes the service features and current status of running service: required/recommended UIs and multimodality, required execution environment characteristics, service profiles, current mode of activity, current constellation of service components (setup), etc. The characteristics of the context are: ž ž ž ž ž ž ž

Context can grow and become very diverse. Context sources may provide huge amounts of data. Context sources may be very diverse. Context sources have a distributed nature. Context information may change frequently. Context data may be incomplete, inconsistent and erroneous. Relevance of context information can depend on the application and situation at hand.

On the application layer (represented by the MobiLife project), the context information is related to two kinds of entity: individual users and groups of users. Determining the situation of users and groups is necessary to enable situation-aware applications and proactive service provisioning. Context information used in the applications can be related to the location, weather, wellness, presence, preferences, etc. of users and groups.

182

Technologies for the Wireless Future – Volume 3

On the service platform layer (represented by the SPICE project), the context-aware service platform functions – which we refer to as intelligent service enablers – such as context sensing and exchanging, context interpreting, brokering and semantic matching, assist in supporting seamless service and application provisioning for ubiquitous mobile systems. Benefits of the ‘intelligence’ (the context awareness part) for end users include getting access to personalized and tailored applications in a mobile setting. Within the SPICE project the intelligent service enablers encompass service platform solutions for user profile and contextual information management and anticipatory middleware functionality. Heterogeneous information stemming from context, user profile and service profile sources is processed with advanced reasoning methods targeting at plausible and usable results. On the network layer (represented by the Ambient Networks project), the networking enablers are needed to achieve a user-centric vision of ambient intelligence. Within these settings the rationale for designing an infrastructure for network context awareness is twofold. On one hand, within very dynamic environments such as ambient networks, there is the need to enhance the operations of other networking functions of the ambient control space to become more capable of reacting to changes in network context in a similar way to how end user applications can be made aware of end user context. On the other hand, the availability of more detailed knowledge about networking context, such as traffic load, QoS, cost, coverage, current performance with regards to specific services (e.g. real-time, file-sharing, etc.), security, etc., is essential in realizing a full contextualization of end user data-communication services and applications. On the reconfigurability layer (represented by the E2 R project), global knowledge, as well as context information, is needed on connectivity and service reconfiguration. The knowledge about connectivity involves, for example, facts regarding radio propagation characteristics, relationships and theorems defined by the graph theory, and design rules from communication theory. Knowledge related to service provisioning embraces service classes and their mapping to network resources and standards, QoS, trust and reputation rules, strengths of crypto schemes and so on. Consequently, a number of context data, coming from different sources, including not only sensors but also network repositories and devices themselves, are taken into account.

5.4.2.2 Ontologies Ontologies are formalisms whose purpose is to support humans or machines to share some common knowledge in a structured way. They allow the concepts and terms relevant to a given domain to be identified and defined in an unambiguous way. The ontologies are used to define basic contextual categories and the (logical) relations among them to ensure interoperability in the communication with and between different context providers. The axiomatic descriptions of context elements, such as personal situations (e.g. Working, AtHome, etc.), can directly be used by logical inference engines to realize high-level context reasoning. It is important to note that we do not propose the ontologies described hereafter as the main representation format for all aspects of context modeling, as ontologies are limited to the formulation of qualitative aspects and the available inference engines are generally weak in handling large amounts of data efficiently. Instead, the elements of the XML-based context meta model can be linked to the elements of ontologies to represent qualitative aspects of context information.

The WWI System Architecture for B3G Networks

183

Context providers may link context elements representing qualitative data to corresponding elements of the context ontologies. For example, the location provider may allow users to associate location categories to important place (i.e. location clusters) using vocabulary from the ontologies (such as ‘home’ and ‘office’). If a set of qualitative context elements characterizing a user’s context, collected from different context providers, is linked to the same set of decidable ontologies, an automatic reasoner can be used to classify the situation. The goal of the ontologies for supporting context-aware computing applications is that they are simple and easy to understand and use. Currently such ontologies have not been defined which fulfil these basic requirements. 5.4.3 Architecture and Interfaces for Context Awareness In the pervasive environments, the end-to-end context management (CM) system is needed in order to make context information available in a structured and organized manner on all the layers of the future wireless communication system. Specifically, the context management system needs to collect raw data from available context sources (sensing), process and represent these data into relevant context information (reasoning) and disseminate this information to be used by applications and services (acting). At the same time, such a system has to support the storage and maintenance of context information, as well as their sharing between different system components and applications. The context provisioning functions have to be distributed to different communication layers in a proper way and they have to interact via the commonly defined interfaces as illustrated in Figure 5.9. The system functionality can be divided into the following functional groups: ž Context source (CS) – sensors, information services (web). ž Context provider (CP) – context data gathering and aggregation, reasoning, distribution, push, subscription, event mechanism, internal data storage for e.g. previously reasoned data,

CIB CS CO Standards

CR

Store context info/ Get context info

Wraps

CP

CC

CB

May use

Advertise CP

Standard CO Standards

Subscribe to context info/ Get context info

Find context info

Services & Platform Layer

Advertise CP

CR CS CIB

Advertise CP

May use

Wraps

CP

Store context info/ Get context info

CB

Find context info

Subscribe to context info/ Get context info

CC

Subscribe to context info/ Get context info

Figure 5.9 System-level architecture for context awareness

Network Layer Reconfigurabilt Layer

184

ž ž ž ž

Technologies for the Wireless Future – Volume 3

context data transformation, security and privacy handling, legal situation handling, context data management. Context client (CC) – context data query, context consuming. Context broker (CB) – context provider registration and maintaining updated list of context providers, mapping context requests to context provider addresses, interdomain brokering. Context representation (CR) – context data model, context reasoning support, ontology. Context information base (CIB) – storage for inferred context information, storage for basic context information (resources of a network), context history (e.g. sequence of context profiles in time).

5.4.4 Use Cases The appropriateness of the system-level architecture for the context awareness in services and applications, as illustrated in Figure 5.9, can be demonstrated by some of the use cases. For simplicity, only the following use cases have been selected for presentation in this chapter: ž Register Context Provider. ž Deregister Context Provider. ž Get a Context Provider for an Entity and Parameter. A key thing for reaching the interlayer interoperability in the presented use cases is that the registration data in the context brokers at the different layers can be maintained synchronized. In this approach the context consumers are able to get the addresses of all context providers from the context broker of their own layer. An alternative for this approach would be that the context brokers know what type of context provider each of them is registering, and when receiving a request from a context consumer they are able to forward that request to the right context broker. In this approach the periodical updates between the context brokers are needed for sharing information on what context items they are authoritative for. 5.4.4.1 Use Case: Register Context Provider The use case Register Context Provider has been illustrated in Figure 5.10. This use case includes the following main transactions: ž A CP registers to the CB(Network) by sending its advertisement message. ž The CB(Network) makes the registration and forwards the advertisement message to the CB(Service&Platform). ž The acknowledgement is sent to the CP after the successful registration. The Identification Code (ID) that a CP needs e.g. for deregistering could be ID D URI D CB’s URL/ID in CB. Example of ID: http://127.0.0.1.8080/broker/Context Broker/advertisement/3.

The WWI System Architecture for B3G Networks

Network Layer Context Provider

185

Services & Platform Layer Context Broker(N)

Context Consumer

Context Provider

Context Broken(S&P)

Context Consumer

Context Provider Advertisement Context Provider Advertisement Return CP's ID in CB(S&P) Return CP's ID in CB(N)

Figure 5.10 Use case: Register Context Provider

Network Layer Context Provider

Services & Platform Layer Context Broker(N)

Context Consumer

Context Provider

Context Broken(S&P)

Context Consumer

Context Provider unAdvertisement (ID in CB(N) Context Provider unAdvertisement (ID in CB(S&P)) Return Acknowledgement Return acknowledgement

Figure 5.11 Use case: Deregister Context Provider

5.4.4.2 Use Case: Deregister Context Provider The use case Deregister Context Provider has been illustrated in Figure 5.11. This use case includes the following main transactions: ž A CP deregisters from the CB(Network) by sending its unAdvertisement message with (ID in CB(Network)). ž The CB(Network) makes the deregistration and forwards the unAdvertisement message to the CB(Service&Platform) with (ID in CB(Service&Platform)). ž The acknowledgement is sent to the CP after the successful de-registration. 5.4.4.3 Use Case: Get a Context Provider for an Entity and Parameter The Use Case Get a Context Provider for an Entity and Parameter has been illustrated in Figure 5.12. This use case includes the following main transactions: ž A CC requests a CP from a CB for wellness information. ž The CB returns the address of a CP (e.g. URL) to the CC.

186

Technologies for the Wireless Future – Volume 3

Network Layer Context Provider

Services & Platform Layer Context Broker(N)

Context Consumer

Context Provider

Context Broken(S&P)

Context Consumer

getAdvertisement (parameter) Return CP's advertisement (URI) Context Query Context Information

Figure 5.12 Use case: Get a Context Provider for an Entity and Parameter

ž The CC requests wellness information directly from the CP. ž The CP returns wellness information to the CC. 5.4.4.4 Challenges When defining the roles of the communication layers in the end-to-end communication management system and allocating different functions to them, the following challenges have to be paid attention to: ž Scalability – the context management mechanisms must be scalable for larger numbers of entities, such as users, operators, devices, different RATs, etc. ž Data modeling – the joint representation, storage and exchange of context data has to be standardized. This depends very much on the particular context data, and the respective mechanisms should be obtained from the specific domain; especially in the telecommunication environment, legacy systems already work with their own formats and it is not probable that they change them to conform with one single format. However, a uniform representation of context information must be achieved, and therefore the context management system should take into account mechanisms of mapping to legacy context sources. Since context information is distributed by its own nature, and involves cross-domain interactions, the data model should provide corresponding linking mechanisms to face these issues. Finally, the data model should be easy to work with, understand and represent, as well as allow for reasoning and inferring knowledge. ž Extensibility – since the context information can be any information, the mechanisms must be kept extensible, so that the different levels of the system architecture can cooperate and exchange context information. ž Common ontology – a common ontology for the different projects must be developed in order to share semantics of contextual information; alternatively, mechanisms for building such a common ontology must be provided. Note that context data model and context ontology are not the same. The former includes the latter, but should also keep such information as the location of context data, its source, its availability, etc.

The WWI System Architecture for B3G Networks

187

ž Information base – the context information base (CIB) is needed to store context and profile information according to a shared ontology. CIB may need to be distributed for performance reasons (to allow both efficient retrieval from context clients and fast updates from context sources). ž Event mechanism – the description, creation and distribution of events in the system must be uniform within the system architecture. ž Privacy and security – there must be mechanisms in place to protect context information from other operators, users and other entities. These privacy right management solutions and security mechanisms must work across all different architecture levels and involved entities. ž Legal issues – different countries have different laws regarding context data and its storage, exchange, etc. It is necessary to define the legal situation for each piece of information (e.g. user name, cell ID, preference items, battery level of the phone, URL bookmark, call logs, etc.) and its use with regard to the availability (to the operator, the service developer, the user), the further use and modification, the format (plain, obfuscated, encrypted, etc.), the duration of access, etc. at each stage (sensing, collecting, storing, processing, etc).

5.5 Network Management in the WWI System Architecture The introduction of the management topic in the context of the WWI system architecture follows the approach to organize, merge and interpret multiple system aspects developed in the involved projects. Applying the management issues in the study of interrelations between the WWI projects, extraction of the ubiquitous conclusions and identification of the differences in approaches, offers specific perspectives on observing and locating the multitude of functionalities involved across the whole end-to-end chain of elements in the emerging telecommunication systems in WWI. Relevant to this study are the projects AN and E2 R, mainly as their scope implies resolving and developing solutions for the entire end-to-end communications chain and/or impacting all layers of the WWI reference models, either by having explicitly defined management functionalities or by specifying the interfaces for interactions with the layers. In specific terms, management in the WWI system Architecture refers to the holistic approach in analyzing and converging on the specifics and commonalities of the involved projects and their management/control structures, as well as resolving and understanding the relevant architectural approaches. 5.5.1 Analysis of Main Assessment Criteria The first observation on the approaches to solving management issues in the overall system architecture of the AN and E2 R projects is based on the identification and analysis of the key assessment criteria typically encountered in general management considerations and development. This is summarized in Table 5.1, with listings of the key management assessment criteria selected and, where applicable, the status of the solutions.

188

Technologies for the Wireless Future – Volume 3

Table 5.1 Status of management assessment criteria in WWI projects No.

Management Evaluation Criteria

Ambient Networks

E2 R

1

Self-managing networks (self-healing, self-optimizing, self-protecting and self-configuring) Self-adaptive applications

Monitoring and controlling self-configuring algorithms

Overall E2 R system architecture and constituent support mechanisms

Monitoring algorithms, adaptive transcoding of media streams Yes

Multimedia application adaptation

2

3

4 5

Autonomic management of networks, systems and services Self-repairing distributed systems Integrated control and management

No Common context and policy planes

6

Distributed, decentralized and scalable management

All management functions are distributed

7

P2P approaches for scalable network management Policy and role-based management Programmable, active and adaptive management

Overlay management network, P2P resource discovery Yes

8 9

10 11

Resilience, dependability and survivability ‘Plug-n-Play’ component-based management

12

Customer controlled and managed networks

13

Proactive and reactive management

14

Biologically-inspired management systems and techniques

The monitoring algorithm, policies

Boot-strap function of AN node

Control of self-x algorithms, scalable monitoring framework. No

No

Overall E2 R system architecture and constituent support mechanisms and building blocks Self-repairing operations of BS Self-ware reconfiguration management plane (S-RMP) concept Mapping of E2 R SA to scalable physical configurations: a) enhancement of 3 GPP SAE networks; b) distribution of FEs to all IP architecture UE self-governance: P2P policy information exchange between UE and network policy servers Dynamic policy generation, policy hierarchies OMA DM enablers for: a) setting self-reconfiguration schedule and b) creating DiagMon TRAP.Mo at the device Disaster recovery scenario Component-based framework for protocol reconfiguration, evolution through the introduction of autonomic protocol component Optional user interaction during autonomic reconfiguration operations Operation around autonomic control loop for proactive and reactive actions Bio-inspired algorithms for spectrum and radio resource management

The WWI System Architecture for B3G Networks

189

In the following, the selected key management criteria are discussed in more detail regarding the essential explanations of the solutions and the fitting approaches to the involved project(s). 1. Self-managing networks (self-healing, self-optimizing, self-protecting and selfconfiguring) – AN: monitoring and controling self-configuring algorithms; E2 R: the E2 R II SA incorporates functionalities for a) self-healing UEs and base stations (through automatic recovery techniques), b) self-optimizing protocols and protocol stacks (through autonomic component replacement and optimization heuristics), c) self-protecting UEs (through secure sandbox mechanisms and secure component replacement), d) self-configuring UEs (through autonomic RAT-switching at the UE and base stations). 2. Self-adaptive applications – AN: self-configuring FE provides a point of control for localized optimization algorithms. This FE captures the set commands issued by external algorithms and provides a set of filtering actions to limit or shape the new configuration values. In this value, stability of configuration can be assured; E2 R: E2 R considers the case where real time multimedia application sessions cannot be maintained during the reconfiguration process (or during part of this process) and proposes an automatic set up of new sessions (based on new QoS characteristics) once the new configuration is enforced, for continuing the previous multimedia sessions which have been interrupted. 3. Autonomic management of networks, systems and services – E2 R: E2 R II SA incorporates dedicated modules and functional entities for autonomic management of networks/systems (cf. knowledge and context management modules, decision-making and reconfiguration management, self-configuration and self-management building blocks) and services (cf. cognitive service provision building blocks). 4. Self-repairing distributed systems – AN: registries management FE provides a highly resilient general distributed registry service for functional entities. An important self-repairing system component is the efficient and self-stabilizing overlay maintenance algorithm of the underlying DHT in registries management FE; E2 R: E2 R introduces mechanisms to assure the connectivity of the base station to the rest of the cellular network, opting for auto-administration of the reconfigured BS through the incorporation of self-repairing actions. This requires discovery of the candidate network point to attach to (BSGW) and the association between BS and the BSGW, as well as transport capacity negotiation between the BS and the controller/gateway. 5. Integrated control and management – AN: management in AN is not working in a completely separate management plane but is integrated into the ambient control space (ACS); E2 R: the overall E2 R system architecture is based on the concept of a self-ware reconfiguration management plane (S-RMP), viewed as a unified control and management framework for the coordination of end-to-end interactions between the involved entities, and for enabling the decision-making and enforcement of mechanisms supporting reconfiguration in a dynamic fashion. 6. Distributed, decentralized and scalable management – AN: distributed and decentralized management for scalability is one of the key concepts of each component of the integrated AN management layer; E2 R: the E2 R II SA has been mapped to the 3GPP SAE network, which provides scalability through the hierarchical structure of control elements. In addition, mapping to an All-IP network configuration facilitates distribution of functional capabilities via flatter architectures. Mass upgrade mechanisms bear inherent scalability properties.

190

Technologies for the Wireless Future – Volume 3

7. P2P approaches for scalable network management – AN: the use of DHTs for AN registries, AN monitoring; E2 R: E2 R foresees the migration from the traditional policy control model between the network policy decision function and the UE policy enforcement function, towards a model whereby the UE and the network operate in P2P fashion. 8. Policy and role-based management – AN: the policy management FE provides node- and AN-level policy management services for all functional entities in the ACS. Policies are stored in nodes (node-level) and in AN-wide registries provided by the registries management FE (AN-level); E2 R: E2 R proposes dynamic policy generation, taking into account defined policies and contextual information with the use of ontologies. E2 R foresees prioritization of policies with the use of policy hierarchies, defining certain reconfiguration actions in an autonomic way. 9. Programmable, active and adaptive management – AN: the monitoring algorithms adapt to the dynamics of the monitored variables and the network environment; E2 R: E2 R exploits OMA DM enablers for such management features, and specifically the possibility to set a self-reconfiguration schedule in the device to reconfigure itself regularly or according to certain situations, or even create DiagMon TRAP/MO at the device. 10. Resilience, dependability and survivability – E2 R: E2 R exploits resilience, dependability and survivability considering disaster recovery scenario. For example, in the case of a natural disaster, one may anticipate that whole areas become inaccessible and some deployed access points may become completely devastated. E2 R proposes mechanisms that will enable remaining access points to be reconfigured in an autonomous and intelligent manner that will also improve the system integrity and maintainability. This way the system can, even at short notice, become once again operational and enhance its availability and reliability by offering vital services to users in a safe manner. 11. ‘Plug-n-Play’ component-based management – AN: a collaborative mechanism is defined for wireless base stations. A wireless network can be deployed through a plug-n-play mechanism: base stations sense and discover each other and negotiate the optimal configuration, like channel assignment; E2 R: E2 R proposes a component-based protocol stack approach. The introduced framework aims to cope with the dynamic binding of component services into a fully fledged protocol service and the runtime replacement of protocol functionality. An enhancement of this solution is also proposed with the introduction of autonomic protocol components, reducing the functionality of centralized management and introducing intelligence to the protocol components. 12. Customer controlled and managed networks – E2 R: the introduction of E2 R reconfigurability classmarks allows the user to interact with the terminal equipment during different stages of the reconfiguration process. 13. Proactive and reactive management – E2 R: the E2 R II system architecture is built around an autonomic control loop that facilitates real-time decision-making dependent on both past observations and online measurements. 14. Biologically-inspired management systems and techniques – E2 R: E2 R exploits general heuristics (meta-heuristics) inspired by a model of division of labour used by social insects such as ants and wasps, and incorporates this model in an innovative and efficient distributed spectrum allocation algorithm, called Distributed Agent-Based Variable Threshold OFDMA (DABVT-OFDMA), whose objective is to maximize the system throughput by exploiting multiuser.

The WWI System Architecture for B3G Networks

191

5.5.2 Management Planes Following the above identification of the major management aspects in the involved WWI projects, E2 R and AN, we highlight the specific issue of management plane(s) and management plane overlay(s) as functional placeholders for the collective sets of features and functionalities. 5.5.2.1 E2 R Management Plane Overlay The E2 R project envisages cognitive networks as the paradigm of heterogeneous next-generation mobile systems that exploit SDR/CR and consist of terminal equipment and network elements with autonomic decision-making and self-management behaviour, while being capable of enriching their knowledge and generating dynamic policy rules based on contextual information. The overall system architecture is based on the concept of a self-ware reconfiguration management plane (S-RMP), viewed as a unified control and management framework for the coordination of end-to-end interactions between the involved entities, and for enabling the decision-making and enforcement of mechanisms supporting reconfiguration in a dynamic fashion. E2 R specified the E2 R II system architecture for autonomously reconfigurable user equipment and network elements, integrating functional entities for software and cognitive radios while capturing autonomic communication aspects. Such work resulted in the identification of four S-RMP modules (Figure 5.13): the knowledge and context management (KCM) module for knowledge and context retrieval and interpretation; the decision-making and reconfiguration management (DM&RM) module for decision making and orchestration of reconfiguration operations; and the self-configuration and management (SC-M) module for applying self-ware procedures, altogether yielding cognitive service provision (CSP) offering to end users. 5.5.2.2 AN Management Plane Overlay Amongst the design principles of ambient networks there is a new approach to providing the functionality that has been traditionally created in the management plane in the past. Such

Cognitive Terminal Domain

Cognitive Network Domain

Knowledge & Context Management

S-RMP

Context & Knowledge Management

Decision-Making and Reconfiguration Management

Self-ware Reconfiguration Management & Control Plane

Decision-Making and Reconfiguration Management

Self-Configuration & Self-Management

Self-Configuration & Self-Management

Cognitive Service Provision

Cognitive Service Provision

CPC

Figure 5.13 E2 R II high-level system architecture for cognitive reconfigurable wireless networks

192

Technologies for the Wireless Future – Volume 3

a novel approach advocates the existence of integrated management solutions that support cooperating networks in highly dynamic scenarios where the need for autonomic behaviour is paramount to reduce complexity and operating costs. These solutions are to allow some level of network functions self-organization and reconfiguration, based on a tight interaction between a network context management system, a scalable policy framework and an underlying network monitoring system. Through continuous awareness of underlying network features, resources, faults and performance while delivering various types of applications, ambient networks become more service aware and eventually capable of dynamically reconfiguring the allocation of resources and enabling their efficient usage. Moreover, distributed monitoring of network performance and the subsequent mapping of network context and network management information to higher-level policies enable the creation of a set of constraints that limit the freedom of operation for the functional entities (FE) composing the ambient control space. As we anticipated, in order to achieve such objectives, ambient networks depart from the traditional centralized management approach to account for more distributed ways of collecting monitoring information, aggregating it and reacting to it. This is realized through the concept of the ambient network integrated management node (see Figure 5.14). As the picture shows, the role of management in ambient networks is carried out collectively through a set of functional components as follows. The monitoring FE provides network-wide metrics, representing the state of the network considered as

ASI ContextWare Primitives

Trigger Repository

Context Clients / Othe FE's Adaptation

FE 2 Adaptive Real-time Monitoring Patterns for Distributed Management Echo

Policies Triggering FE

Local State

A-GAP

Overlay Topology

Execution Environments

FE 1 ContextWare

ConCoods

Context Managers

Context Collection

Co SoContext Source

ASI Management Primitives

ASI

Context Context Aggregation Dissemination

É

FE 3 Policy Manag FE 3 Policy Manag FE 3 Policy Manag. Systems Policy Authoring Point Policy Mang. Point

Neighbor State Policy Repository

SATO Manag. FE

F

Policy Exec Policy Po Execution Point

Management Initialisation & Reconfiguration of FE s

FE 4 Composition Management

FE 5 Distributed Manag. System Primitives (AN Globally Accessible DHTs) CIB/MIB (Context Information Base)

Co Ne SoContext enNetwork Source Sensor

Figure 5.14 AN management node model

FEs

ANI

ACS Registration (Framework)

The WWI System Architecture for B3G Networks

193

a whole. ContextWare FEs confer to ambient networks to become context aware, or to be more precise, aware not only of the services that are being delivered but also of the network’s continuous ability to support their requirements. TRG FE and related ANISI (ambient networks information service infrastructure) empower ambient networks with a subscription/notification mechanism enabling clients to quickly react to triggers that cross given thresholds. The self-configuration FE provides general services to other FEs to correctly accomplish their configuration tasks. Its main purpose, currently investigated, is to guarantee stability of a certain configuration over time and to avoid oscillations. The policy FE facilitates the role of authoring and injecting policies which then provide additional constraints to the behaviours of other management components, as well as other FEs in the ambient control space. The composition management FE manages the logical structure of the ACS, as well as FE instances in this structure. Finally, the goal of a common AN distributed registry (registry management FE) is to exploit synergies and provide a unified distributed registry functionality for all functional entities, as a number of functional entities need registries to store and look up various entities. In order to allow for a high degree of flexibility, it is envisaged that such nodes will be dynamically deployed in the most appropriate locations in a network topology through a ‘virtual management backbone’ overlay. Each individual management node part of this overlay may not necessarily implement the complete set of management functionality, but only the required subset according to what particular management task needs to be carried out. 5.5.3 Analysis of Management Plane Overlays The conducted considerations of the management solutions and specifics in AN and E2 R projects reveal the extent and types of difference in approach of the two projects, but also point out the commonalities in addressing the similar management issues, as captured in the management assessment criteria, either by providing specific or, in fact, offering similar solutions. The starting point in the analysis of the two projects is their guiding general concepts for advancing the operations and orchestrations of telecom systems, AN introducing the concept of ambient connectivity based on the notion of network compositions, E2 R introducing various types of adaptation actions based on the notion of end-to-end reconfigurability. Regarding the implementations of the management planes, AN introduces a distributed approached by integrating the management aspects of cooperating networks and proposes a model based on the AN management nodes which contain the relevant collections of functional entities that govern the management operations and are located and distributed in the network topology via a ‘virtual management backbone’ overlay. E2 R introduces the dedicated management plane called self-ware reconfiguration management plane (S-RMP), which orchestrates the whole set of end-to-end reconfiguration actions and contains the functional blocks for resolving the decision-making and invocation of functions supporting reconfigurations. The models in which the management tools are realized in the two projects can be associated with the scope and general guiding concepts applied in the projects. As advocated in the projects, these models provide the most efficient way of placing and organizing those functionalities in the system. Hence, if an observation is made on the possible common and integrated WWI management plane structure, the specifics and tools realized in the two projects would have to be initially included in the form they are developed in the projects, particularly as they are the required models for targeting the relevant solutions. Then, in such a coexisting environment, a

194

Technologies for the Wireless Future – Volume 3

further study could be conducted on resolving the commonalities and solving the operational conflicts (if applicable). It seems that in such scenarios the current features of the management solutions in the projects could individually resolve or at least accommodate the integration of the two sets of solutions. Finally, related to the details and options for resolving the management objectives, the identification of the solutions for the management assessment criteria presents the manner in which both projects tackle them. However, it can be highlighted that both projects apply some common management principles related to the awareness and implementations of degrees of autonomics and self-management in the management of system operations. In fact, by observing the specific functionalities in both projects, autonomics and self-management are becoming an inherent aspect of their management functionality.

5.6 Conclusions This chapter presents a system architecture for B3G networks, which are legacy-respecting networks building on current 3G cellular networks and integrating telecommunication and data networks in the fixed and mobile domain more closely than 3G systems already do. The architecture presented here follows a horizontally-layered design distinguishing basic connectivity, comprising access and core/backbone connectivity, a network control layer, a service layer on which the end-user applications reside, as well as a vertical reconfiguration plane spanning all of the horizontal layers mentioned before. Apart from the presentations of an overall reference model explaining the different layers and the reference points residing between them, four functional areas found to be of key importance to B3G networks are discussed in more detail. These functional areas are: heterogeneous radio resource management, mobility management, context awareness and management. The architecture presented here is the result of a joint activity of five large-scale integrated projects executed in the 6th Framework Program of the European Commission. These projects have gathered under the umbrella of the Wireless World Initiative (WWI) to jointly develop a vision for B3G.

5.7 Acknowledgements The editors would like to especially thank Imran Ashraf (Alcatel Lucent, UK), Giovanni Bartolomeo (University of Rome ‘Tor Vergata’, Italy), Zachos Boufidis (University of Athens, Greece), Khadija Daoud (Orange Labs, France), Alex Galis (University College London, UK), Raffaele Giaffreda (BT, UK), David Grandblaise (Motorola, France), Oliver Holland (King’s College London, UK), Alexandros Kaloxylos (University of Athens, Greece), Herma van Kranenburg (Telematica Instituut, Netherlands), Jijun Luo (Siemens, Germany), Juha Mikola (Nokia, Finland), Markus Muck (Motorola, France), Eleni Patouni (University of Athens, Greece), Mikael Prytz (Ericsson, Sweden), Abed Samhat (Orange Labs, France), Ove Strandberg (Nokia Siemens Networks, Finland), Anthony Tarlano (DOCOMO Eurolabs, Germany) and Dorota Witaszek (Fraunhofer Institute FOKUS, Germany) for the contributions to this work.

The WWI System Architecture for B3G Networks

195

References [1] http://www.ist-winner.org/. [2] http://www.ambient-networks.org/. [3] [4] [5] [6]

http://www.ist-mobilife.org/. http://www.ist-spice.org/. http://e2r2.motlabs.com/. http://www.wireless-world-initiative.org/.

[7] WWI Cross Issue System Architecture Mobility Management Final Report. [8] D. Bourse, M. Muck, O. Simon, N. Alonistoti, K. Moessner, E. Nicollet et al., “End-to-End Reconfigurability (E2 R II): Management and Control of Adaptive Communication Systems”, IST Mobile Wireless and Communication Summit Conference, Greece, Mykonos, June 2006. [9] J. Zoric, E. Postmann, J. van der Meer, J. Hendriks, W. Kellerer, T. Srisakul and J. Rovira, “Service and Information Roaming – Architecture and Design Aspects”, WWRF 17, Heidelberg, November 2006. [10] A. Schieder, E. Tragos, A. Mihailovic, A. Kaloxylos, M. Prytz, O. Strandberg et al., “Mobility Management and Radio Resource Management in the WWI System Architecture”, WWRF 17, Heidelberg, November 2006. [11] J. Lundsj¨o, R. Aguero, E. Alexndri, F. Berggren, C. Cedervall, K. Dimou et al., “A Multi-radio Access Architecture for Ambient Networking”, 14th IST Mobile and Wireless Communications Summit, Dresden, Germany, 19–23 June 2005, available at http://www.ambient-networks.org/phase1web/main/publications_incl_Y2.html. [12] G.P. Koudouridis, R. Aguero, K. Daoud, J. Gebert, M. Prytz, T. Rinta-aho et al., “Access Flow Based Multi-radio Access Connectivity”, PIMRC 07, Athens, Greece, September 2007. [13] A.K. Dey, G.D. Abowd and D. Salber, “A Conceptual Framework and a Toolkit for Supporting the Rapid Prototyping of Context-aware Applications”, Human-Computer Interaction, 16, pp. 97–166, 2001.

6 New Air Interface Technologies Edited by Dr Angeliki Alexiou (Bell Labs, Alcatel-Lucent UK) and Dr Gerhard Bauch (DOCOMO Euro-Labs Germany)

6.1 Introduction In this chapter, air interface enabling technologies, performance evaluation and modeling issues for next-generation wireless systems are addressed. As a continuation of the two previous volumes of the WWRF Book of Visions, the objective is to report the latest technological and research developments and, to this end, this chapter focuses on three major subjects expected to play a significant role in the design of future air interfaces. These are: ž error control coding ž multi-dimensional channel modeling ž multiuser MIMO systems. Over recent years, research developments in the areas of error control coding (ECC) and soft decision decoding (SDD) have pushed the limits of a single communication link performance to closely approach the Shannon capacity bound. A classic lower bound on the ‘block error rate’ (BLER) for finite block length n and given code rate R is the sphere-packing bound of Shannon, assuming additive white Gaussian noise (AWGN) channel and not taking into account modulation. Comparison of different decoders can usually be based on this sphere-packing Shannon bound as benchmark. It has been shown that both turbo product codes (TPC) and low-density parity check (LDPC) codes can closely approach the Shannon limit in the case of very long codewords. For intermediate codeword lengths, LDPC codes provide the most promising ECC option to most next-generation wireless applications. For short block lengths, iterative processing based on turbo codes or LDPC deviate more substantially from the optimality benchmark. Designing a ‘perfect’ code in this range remains an open research question.

Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

198

Technologies for the Wireless Future – Volume 3

Design requirements for ECC options for next-generation wireless systems will need to focus on a number of performance targets: ž Sufficient performance in terms of approaching the optimality benchmark for a large range of realistic channel propagation scenarios (and not only for the ideal AWGN case). ž Addressing the open question of ECC performance for short packet lengths. ž Adaptivity of EEC design to channel conditions, such as noise level and complexity constraints. ž Introduction of a generalized EEC design that allows for parameterization and reconfiguration. In the first part of this chapter, an overview of existing families of codes, together with their properties, design and optimization guidelines is presented, and comparison approaches are discussed. A wide range of decoding algorithms is then described. Particular attention is given to LDPC decoder implementation, introducing architecture and hardware requirements. Moreover, related standardization activities are presented. Finally, the general turbo principle is explained and a new highly performing turbo-interleaving scheme is analysed. In the second part of this chapter, the latest developments in the area of multi-dimensional channel modeling are presented. The use of multiple antennas at both ends of a wireless communication link introduces a number of channel modeling challenges for the simulation and testing requirements of future wireless systems. Adequate channel modeling would need to accurately describe the characteristics of new frequency bands and propagation scenarios, large bandwidths, advanced multi-antenna structures and new network topologies. At the same time, accurate representation of a large number of parameters and propagation scenarios results in modeling complexity often unrealistic in terms of simulation and computational burden, and also in limited efficiency when the objective is to, preferably, be able to isolate and qualitatively assess the impact of certain parameters. Striking the right balance between efficiency and accuracy has in fact been the major challenge in multi-dimensional channel modeling over the last years. As multiple antenna processing has been adopted by all evolving wireless standards, multi-dimensional channel modeling has recently received a great amount of attention in the standardization community, where accurate and realistic channel models can play a decisive role in the shaping of requirements for future wireless system concepts. In this part of the chapter, we first describe the most commonly used channel modeling methodologies, namely deterministic, stochastic and geometry-based modeling, and briefly explain their attributes, advantages and limitations. Then an overview of the most important multi-dimensional channel modeling activities is presented, covering both research and standardization, and a comparison of their main characteristics is attempted. Finally, remaining open issues and potential future research questions are discussed, mostly related to complexity reduction, inter-cell interference modeling, and stationarity and scale effects. In the last part of this chapter we discuss novel approaches and performance enhancements and limitations associated with multiple antenna systems. multiple input multiple output (MIMO) wireless systems offer the potential of significantly improving capacity by exploiting beamforming gains when focusing the transmitted/received power to desirable spatial direction and diversity, and spatial multiplexing gains when taking advantage of a number of independent spatial signatures or modes resulting from the induced multidimensional channel. However, in multiuser scenarios, several co-channel users attempt to obtain

New Air Interface Technologies

199

access of the radio resources simultaneously and therefore create co-channel interference (CCI). The efficient management and mitigation of the interference effects becomes in this case critically important in order to preserve the potential gains offered by the spatial dimension. A wireless system equipped with multiple antennas at both ends of the communication link allows for spatially multiplexing signals intended for different users on the downlink and spatially separating signals transmitted from a number of different users on the uplink, provided that the induced interference can be successfully mitigated. This type of multiple access in the space domain, called space division multiple access (SDMA), promises considerable gains in capacity – upper bounded by a gain factor equal to the number of antennas – but relies heavily on the availability of channel state information at both ends of the communication link. To optimally utilize the available, short-term (or ‘instantaneous’) or long-term channel state information for the downlink transmission in a multiuser MIMO system, joint precoding techniques of the signal intended for different users, multiuser precoding, have been proposed. Linear multiuser precoding and multiuser beamforming can result in considerable gains, even when only long-term channel state information is available, with relatively low complexity. Nonlinear precoding techniques have a higher computational complexity and require additional signaling but can potentially provide substantially better performance compared to linear techniques. ‘Dirty-paper’ coding (DPC) techniques can achieve the maximum sum rate of the system and provide the maximum diversity order but require much more advanced decoding approaches. Tomlinson Harasima precoding techniques are nonlinear precoding approaches that offer a good compromise between performance and complexity. In the last part of this chapter, a comparison of linear and nonlinear multiuser MIMO precoding techniques is presented. The influence of channel estimation errors is studied and the impact of scheduling assumptions analysed.

6.2 Error Control Coding Options for Next-generation Wireless Systems 6.2.1 Introduction Recently, much progress has been made in designing error control coding (ECC) and soft decision decoding (SDD) to closely approach the Shannon limits. A classic lower bound on the block error rate (BLER) for finite block length n and given code rate R is the sphere-packing bound of Shannon [1], hereafter denoted SP59. This bound applies only for the additive white Gaussian noise (AWGN) channel and does not take into account the modulation scheme that might lower the capacity of the AWGN channel. The starting point for comparing different decoders over various channels (e.g. 3GPP, 802.11, 802.15.3a- UWB, 802.16d/e channel models) is to assume binary phase-shift keying (BPSK) modulation over AWGN, and to target a prescribed BLER, of say Pt D 10
4 . We then investigate the minimum E b =N0 to attain the target BLER. The difference between the required E b =N0 and the minimum possible E b =N0 implied by SP59 to attain the same Pt is the measure of the ‘imperfectness’ of a given code. The sphere-packing argument considers the transmitted codewords as ‘code points’ located p on a sphere of radius n E s in R n , where E s is the average energy of the transmitted signals. Then a code is ‘perfect’ (i.e. reaches the SP59 with equality) only if equal-sized nonintersecting cones could be drawn from the origin around every code point to completely fill the

200

Technologies for the Wireless Future – Volume 3

n-dimensional Euclidean space R n . Note that for moderate block length (less than 1000 bits), a computationally efficient formula to derive the SP59 bound was introduced in [2]. It is well known that iterative decoders based on turbo product codes (TPC) can approach Shannon capacity by 0.27 dB based on [1023, 1013] Hamming code as a component code and a target BER of 10
5 (see [3]). It is also reported that the SP59 can be approached within 0.1 dB based on turbo codes, and within 0.0045 dB for irregular LDPC, typically for very long codewords of length greater than 100 000 bits. For lengths 1000 to 100 000, the NASA CCSDS turbo codes of rates 1=2; 1=3; 1=4; 1=6 [4] have imperfectness of 0.7 dB. For block sizes ranging from 56 to 11 970 information bits, a classical serial concatenation scheme of block codes with convolutional codes CRC-SCCC, BCH-SCCC is shown [5] to have ‘imperfectness’ of 1 dB and less using iterative the Viterbi algorithm (IVA). This approach is slightly better than the CCSDS codes at short block length and about 0.7 dB worse than long LDPC codes. In this range, LDPC codes provide a promising ECC option to most next-generation wireless applications, as will be elaborated in this section. For block length of 50 to 1000 bits, iterative processing based on turbo codes or LDPC deviate more substantially from perfectness. In this range, it is an open question how to approach ‘nearly perfect’, say 1 dB from SP59, with practically decodable ECC algorithms. It is argued in [6] that LDPC codes can operate within 0.5 dB of the SP59 bound for block length less than 1000 bits. Simulation results show that LDPC codes of length 1020 and k D 510 are about 1.5 dB away from the sphere-packing bound at BLER D 10
4: . We summarize the desired properties we wish to find in ECC for the next-generation wireless channels: ž Bound meeting performance over real-world wireless channels – we target ECC schemes that are optimal (in the sense of maximum likelihood (ML) or maximum a posteriori (MAP)) or near optimal and should approach channel capacity. Note that for real-world wireless channels the soft metrics used for the optimal ECC should be carefully explored, and in general the capacity expression for the AWGN channel is just the ‘first approximation’ model for the real-world channel. ž Performance for finite block length – approaching the sphere packing bound SP59 (or other relevant bounds) for any finite block length from short packet length (8 to 100 bits) to medium packet length (100 to 1000 bits) to large block length (1000 and more bits). ž Adaptive complexity to channel conditions – it is highly desirable to introduce optimal and near-optimal ML or MAP decoders whose complexity adapts to the noise level of the channel. Ideally, we would like to perform algebraic hard decoding at high SNR and thus drastically reduce the inherently exponential decoding complexity. ž Ease of implementation and reduced (average and worst-case) complexities – complexity near capacity (worst-case from a complexity point of view) tends inherently to be exponential for optimal SDD. Complexity at 1 and more dB from capacity could be greatly reduced if the SDD will adapt its procedure automatically to the channel conditions. ž Generic decoding procedures – it is desirable to implement ECC procedures in which we can easily replace the code used with another (by changing one of the code parameters such as the length, dimension, minimum Hamming distance, etc.) to better suit our needs. The decoding procedures should be general enough to decode ‘any’ linear block code from the same family. The structure of this section is as follows: Section 6.2.2 provides full details about existing families of codes, together with their properties, design and optimization guidelines, whilst

New Air Interface Technologies

201

giving comparison elements. Particular attention is then paid to the wide range of decoding algorithms in Section 6.2.3. Section 6.2.4 focuses on LDPC decoder implementation, introducing architecture and hardware requirements. An exhaustive overview of existing standards activities from the pure ECC point of view is given in Section 6.2.6. The general turbo principle is introduced in Section 6.2.5, by describing an innovative and promising turbo-interleaving scheme. 6.2.2 Coding 6.2.2.1 General Code Types and Taxonomy The most important coding schemes that can be decoded using an iterative (turbo) algorithm can be classified as parallel concatenated codes, serial concatenated codes and low-density parity check codes (LDPC codes), as indicated in Figure 6.1. In parallel concatenated codes, the data sequence is encoded by the first constituent encoder. The second constituent encoder encodes an interleaved version of the data sequence. The data bits are sent only once as systematic bits of the concatenated code, whereas only the parity bits of the constituent encoders are transmitted. Usually, recursive systematic convolutional codes are used as constituent codes. However, other code types, e.g. block codes, can also be used and more than two constituent codes can be concatenated with different interleavers. Parallel concatenated convolutional codes (PCCC) are what are usually referred to as ‘turbo codes.’ In serial concatenated codes, the second (inner) constituent code encodes the interleaved code bits of the first (outer) constituent code. Convolutional codes are the most common constituent codes for serial concatenated coding schemes. However, this scheme can be generalized if we consider other components in the transmission chain as an inner encoder, e.g. the mapper of a QAM modulation scheme, the ISI/MIMO channel, a rate 1 precoder and the like. Low-density parity check (LDPC) codes are block codes, where the codeword is generated by multiplying the data sequence d D [d1 ; d2 ; : : : ; d N ]T with a generator matrix G. The code is defined by a sparse parity check matrix H which satisfies HG D 0. LDPC codes are often represented by their Tanner graph, as indicated on the right hand side of Figure 6.1. The nodes on the left hand side are called variable nodes. Each of them represents a code bit. The nodes on the right hand side are called check nodes and represent the parity check equations. A connection between variable node i and check node j exists in the graph if the element h ji in the parity check matrix H is 1. The modulo 2 check sum of all variable nodes which are connected to the same check node is 0.

Parallel Concatenation (Turbo Codes) info bits

code bits encoder 1

interleaver encoder 2

Serial Concatenation encoder 1

Π

LDPC Codes

encoder 2

• convolutional code • rate 1 precoder • QAM mapper

Figure 6.1 Coding schemes with iterative decoding

interleaver

202

Technologies for the Wireless Future – Volume 3

LDPC codes were invented in 1962 [27]. They have regained attention in the context of iterative decoding since the so-called message passing decoding algorithm can be viewed as iterative decoding, with the check nodes and the variable nodes as constituent decoders. A main reason why LDPC codes have become so popular is that they allow highly parallel implementation. While the trellis decoder for a convolutional code needs a backward and a forward recursion through the trellis, all decoding operations in the variable nodes and the check nodes, respectively, can in principle be done in parallel. This allows a decoder implementation with high throughput as required in future wireless systems. A disadvantage of LDPC codes is that in general the encoding complexity grows quadraticaly with the block size, while the encoding complexity of convolutional codes grows only linearly with the block size. However, with structured LDPC codes, the encoding complexity can be greatly reduced. Many block codes can be regarded as special cases of LDPC codes. We will explain repeat accumulate codes as one example later in this section. Further variants are discussed in Section 6.2.2.2. The three classes of coding schemes in Figure 6.1 have in common that they can be decoded by an iterative (turbo) decoding scheme as indicated in Figure 6.2. The received data is input to the decoder of the first constituent code, which produces a soft a posteriori information, i.e. decisions plus reliability information. The output of decoder 1 is used by the second constituent decoder as a priori information on top of the received channel information. After both constituent codes have been decoded once, the output of the second constituent decoder is fed back to decoder 1 and used as additional a priori information in a further decoding step. Several decoding iterations can be performed in this way in order to lower the error rate. It is essential that only extrinsic information of the other constituent decoder is used in order to avoid multiple use of the same information. Extrinsic information on a bit is the information which can be obtained from all the other bits in the block based on the code constraints. It is the part of the a posteriori information that is newly generated in the current decoding step. Using soft information in the form of log-likelihood ratios L.d/ D log

P.d D 0/ P.d D 1/

(6.1)

extrinsic information is obtained by bit-wise subtraction of the input log-likelihood ratios from the output log-likelihood ratio. Usually, LDPC codes require a significantly larger number of iterations than PCCC or SCCC. In the following, we elaborate on serial concatenated convolutional codes (SCCC) and compare them to parallel concatenated convolutional codes (PCCC) in terms of performance and complexity.

decoder 1

Lce(cˆk ′) Lca(ck ′)

Π−1

Ld(ck) d L e(cˆk)

Π

Figure 6.2 Iterative decoding

decoder 2

New Air Interface Technologies

203

BER

parallel serial earlier convergence

lower error floor SNR

Figure 6.3 BER of PCCC and SCCC

Simply put, the bit error rate (BER) performance of iterative decoders is characterized by three regions, as indicated in Figure 6.3. At very low SNR, iterative decoding cannot achieve a reasonable error rate. At an SNR where iterative decoding starts to become effective, the BER curve decreases with a very steep slope. We call this area in the BER plot the waterfall region. At higher SNR, we observe an error floor which is determined by codewords with small Hamming distance. The error floor can be reduced by proper interleaver design. Simply put, parallel concatenated convolutional codes tend to converge at lower SNR, i.e. the waterfall region starts at lower SNR. On the other hand, serial concatenated convolutional codes tend to show lower error floors. Consequently, serial concatenated coding schemes are more suited for applications which require a very low BER, whereas parallel concatenated codes are more suitable for applications which can handle a certain error rate, e.g. by means of ARQ or error concealment. 6.2.2.2 Designing Codes Based on Graphs The design of codes based on graphs can be understood as a multi-variable multi-constraint optimization problem. The constraints of this problem are the performance requirements, flexibility (i.e. block lengths, rates, etc.) and encoding/decoding complexity. The last one also includes the complexity of hardware realizations of the encoder and decoder, as well as latency issues. Figure 6.4 shows an illustration of the constraints and variables involved in the design of codes defined on graphs. Some of the typical variables that can be optimized to meet the specified constraints are represented as circles in Figure 6.4. One of the first decisions that should be taken when designing codes defined on graphs is whether their graphs should have a pseudo-random or an algebraic or a combinatorial underlying structure. These different structures have their advantages as well as their downsides. For instance, pseudo-random structures provide the code designer with lots of freedom. The codes originating from these structures can have practically any rates and block lengths. However, these codes prove themselves to be difficult to implement because of their complete lack of regularity. On the other hand, algebraic and combinatorial designs, which we will call structured designs from now on, cannot exist for all rates and block lengths. This is because the algebraic or combinatorial constructs used are based on group and number theory, and, therefore, they are inherently of quantized nature, mainly based on prime numbers. Another characteristic of structured designs is that it is normally possible to obtain good codes for small to medium block lengths. Otherwise, pseudo-random designs have better performance for long block lengths. From an implementation point of view, structured designs have a lot

204

Technologies for the Wireless Future – Volume 3

Encoding/Decoding Complexity

Block Codes Constraint Length

Block Length

Interconnection Minimization Matching for Turbo Eq.

Performance

Convolutional Codes Rate Compatibility

Distances Optimization

Channel Matching

Algebraic Structures

Pseudo-Radom Structures Girth Maximization

Flexibility

Figure 6.4 Design of codes defined on graphs as an optimization problem

of advantages. For instance, the regular connections in their graphs facilitate tremendously the hardware implementation of the decoders for these codes. Additionally, the algebraic or combinatorial structure can be exploited to simplify the encoding algorithm. 6.2.2.3 Code Optimization Constructing a good short to medium length code that will decode well under iterative decoding is an art, with many heuristic rules and constraints to fulfil, some to do with decoder performance, some with easy encoder implementation, etc. Speaking of ‘code optimization’ in this context is an overstatement, and the final stages of practical code design are usually performed by tweaking and analyzing the simulated error performance curve in its three stages: the ‘error floor’, the ‘waterfall’ and the ‘error ceiling’. The next sections will elaborate on the design of practical short to medium length codes. In the asymptotical regime, when the code length goes to infinity, the performance of the coding system can be predicted very accurately using techniques such as density evolution or EXIT charts. Based on these techniques, it is possible to run numerical algorithms to optimize the code design. The error performance for infinite code length is binary: above a threshold in terms of SNR, the probability of error tends to zero, and below the threshold it tends to one. Therefore, analysis techniques concentrate on predicting and optimizing the threshold. The most accurate method for predicting the threshold of a concatenated code is density evolution [38]. This method consists in tracking the probability density function of the messages exchanged in the decoder graph throughout the iterations. This can be done analytically for some channels and decoders, or approximated numerically using sampled densities. The analysis is done for isolated components in the graph, assuming their input messages to be independent, and deductions are made for the convergence of the global algorithm. This independence assumption is verified asymptotically when the code length goes to infinity and interleavers are random, which is why density evolution is only accurate in the asymptotical regime.

New Air Interface Technologies

205

The threshold for LDPC codes is a function of the ‘degree polynomials’, which specify the densities of 1s per column and per row in the parity-check matrix used by the iterative decoder. These can be optimized to maximize the threshold for a given code rate. A popular web-based source of optimized binary LDPC designs is available on R¨udiger Urbanke’s web site (http://lthcwww.epfl.ch/research/ldpcopt/). Density evolution can be simplified by tracking a reduced set of parameters of the message densities. EXtrinsic Information Transfer (EXIT) charts [49] plot the mutual information between the messages and the corresponding code digits. EXIT charts provide a visual design tool that can be adapted and applied to a wide variety of scenarios, like serial or parallel (turbo) concatenation of convolutional codes, or iterative detection C MIMO multiuser detection [10]–[12]. They are less accurate than density evolution, but allow an understanding of effects that are difficult to pinpoint otherwise, such as decoder-induced error floors, the effect of scheduling on the convergence of doubly-iterative processes, or code design to minimize the number of iterations, to name just a few. Pseudo-random Designs The history of pseudo-random designs coincides with the history of codes on graphs. Already in his seminal work Low-density Parity-check Codes [14], which introduced the LDPC codes, Gallager considered pseudo-random designs. Throughout the years other researchers have also considered some important pseudo-random designs. This is mainly because such codes are very flexible, as mentioned earlier, and also because they enable the use of some powerful techniques to study their asymptotic behaviour. In this section, some of these designs will be discussed for turbo codes and LDPC codes. Pseudo-random Designs for Turbo Codes When we mention the different designs of turbo codes, we are currently referring to their interleavers. The performance of a turbo code depends on how effectively the data sequences that produce low-weight codewords at the output of one encoder are matched with permutations of the same data sequence that yield higher encoded weights at the outputs of the others. Random interleavers do a very good job of combining low weights with high weights for the vast majority of possible information sequences. In this section, the most known interleavers of this class will be presented. S-random Interleavers S-random interleavers were introduced by Divsalar and Pollara in [13]. The design of an S-random interleaver guarantees that, if two input bits to an interleaver 5 are within distance S1 , they cannot be mapped to positions less than S2 apart at the interleaver output, and usually S1 D S2 D S is chosen. So, considering two indices i; j such that: 0 < ji
j j < S

(6.2)

j5.i/
5. j/j > S:

(6.3)

the design imposes that:

p When designing these interleavers, it was observed that S < N =2 usually produces a solution in reasonable time, where N is the length of the interleaver to be designed.

206

Technologies for the Wireless Future – Volume 3

Pseudo-random Designs for LDPC Codes It is well known that the message-passing algorithm used to decode LDPC codes converges to the optimum a posteriori probability (APP) solution, if the Tanner graph representing the parity-check matrix of the code has a tree structure. In light of this, Gallager, in his work of 1963, considered some pseudo-random designs that avoid short cycle lengths. In appendix C of his thesis [14], the algorithms for generation of codes that avoid a certain minimum cycle length, called girth of the graph, are presented. Below, we present an algorithm that has become the basis of pseudo-random LDPC code generation with large girth properties. Progressive Edge-growth Tanner Graphs The main idea behind this graph construction method, which was presented in [15], is to progressively establish the edges or connections between variable and check nodes in an edge-by-edge manner so that the resulting graph shows the desired girth properties. In summary, the progressive edge-growth (PEG) algorithm works as follows. Given the number of variable nodes n, the number of check nodes m, and the symbol-node-degree sequence of the graph [16], an edge-selection procedure is started such that the placement of a new edge on the graph has as small an impact on the girth as possible. After a best-effort edge has been determined, the graph with this new edge is updated, and the procedure continues with the placement of a next edge. As we see, the PEG algorithm is a general, nonalgebraic method for constructing graphs with large girths. It is worth mentioning at this point that it is possible to obtain codes with linear-time encoding complexity using this algorithm. In this case, the edges associated with the m variable nodes of the codeword should be placed to form the so-called zigzag pattern [15]. After these edges are placed, the conventional PEG algorithm can be used to place the remaining edges. The code obtained this way can be encoded using the back-substitution procedure. 6.2.2.4 Binary vs. Nonbinary Binary LDPC codes can be generalized to nonbinary LDPC codes (NB-LDPC). The parity-check equations are written using symbols in the Galois field of order q, denoted GF(q), where q D 2 is the particular binary case. The parity-check matrix defining the code has only a few nonzero coordinates which belong to GF(q), and a single parity equation involving the codeword symbols has then the form: dc X

h ji Ð ci D 0

(6.4)

iD1

where {h ji } are the nonzero values of the j th row of H. In terms of algebraic properties and error correcting capabilities, there is not much difference between nonbinary and binary codes, and the question whether it is useful to consider NB-LDPC codes is a valid question. If we leave aside the better behaviour of nonbinary codes for correcting bursts of errors, the principal reason for using NB-LDPC codes lies in the fact that the practical decoder is suboptimal, which is the case of the belief propagation (BP) decoder, or its reduced complexity derivatives. In particular, it is useful to consider nonbinary LDPC codes when the nonbinary decoder is much closer to optimal maximum likelihood decoding (MLD) than its binary counterpart.

New Air Interface Technologies

207

Let us discuss some general issues that help the understanding of NB-LDPC codes advantages. It is well known nowadays that the drawbacks of belief propagation decoding of binary LDPC codes come from the dependence of the messages in the Tanner graph representation of the code. The dependence comes from very specific topological structures in the Tanner graph of the code, e.g. cycles, stopping or trapping sets. The bad behaviour of the BP decoder on these topological structures is even enhanced if the log-likelihood ratio (LLR) messages that initialize the decoder are already correlated by the channel. In the following two examples, the use of NB-LDPC codes helps to bypass correlation effects of the messages: Short/moderate-length Codes The Tanner graph of the NB-LDPC code is much sparser than the one of a binary code with same parameters. This has been pointed out by several authors [17]–[20]. As a consequence, the higher girth of NB-LDPC graphs helps to avoid the short cycles and also mitigates the effect of stopping or trapping sets, making the BP decoder closer to MLD. Actually, when q is larger than q
26, best error rate results on binary input channels are obtained with the lowest possible variable node degree, that is dv D 2. Those codes have been named cycle-codes in the literature, or ultra-sparse LDPC codes [17], [18]. For example, the girth of a binary irregular LDPC code with length N D 848 bits and rate R D 0:5 is at most gb D 6 for the good degree distributions, while the girth of an NB-LDPC code with same parameters is gnb D 14 when a good graph construction is used [19]. High-order Modulation (M-QAM) For binary LDPC coded modulations, the output of the Bayesian maximum a posteriori demapper gives correlated probability weights, which means that the initialization of the BP decoder will experience correlated messages even without any short cycles. Of course, there are several ways of fighting this effect, by using an interleaver (BICM-LDPC), or using multilevel coding. However, if the LDPC code is built in a field with order equal or higher than the modulation order, the non-binary LDPC decoder is initialized with uncorrelated vector messages, which helps the BP decoder to be closer to MLD. This way, the code operates in the modulation signal set, like in the Trellis-coded modulations. The application of NB-LDPC codes to high-order modulations has been proved very efficient both with analytical approaches and in simulations [21]–[23]. Therefore, if one agrees to increase the decoding complexity of the receiver, it is possible to expect a significant performance gain in the above described cases. We give hereafter some evidence of the advantages of NB-LDPC codes. Before describing in detail a few interesting simulation results, we discuss briefly the most recent results regarding the optimization of NB-LDPC codes. Because of the very low density of NB-LDPC graphs, there is not much room left to optimize irregularity profiles, as is done for the binary irregular LDPC codes. Some authors have generalized the methods based on density evolution used in the optimization of binary LDPC codes. All convenient optimization methods are based on a Gaussian approximation of the densities, also referred to as EXIT charts for LDPC codes [23], [25], [26]. However, the obtained irregularity profiles only apply to very long codeword lengths. For short-length codes, better results are obtained with quasi-cyclic nonbinary LDPC codes [24] or ultra-sparse LDPC codes with large girths whose coefficients are chosen appropriately [17], [20], [29], [30].

208

Technologies for the Wireless Future – Volume 3

6.2.2.5 Performance Results of Nonbinary LDPC Codes Small Codeword Lengths Figures 6.5 and 6.6 show two examples of the interest of NB-LDPC codes at small codeword length on the BI-AWGN channel. In each figure, NB-LDPC codes optimized using the method proposed in [29] are simulated together with an irregular binary LDPC code with the same parameters (size and rate). The binary code irregularity is taken from [33] and the parity matrix is built with the progressive edge growth (PEG) algorithm. One can see that most of the performance gain is obtained by going from GF(2) to GF(64), and that GF(256) codes are only interesting if one wants to lower the error floor. Note that regardless of the decoding complexity, these results are the best presented in the literature, obtained with iteratively decoded codes. The gap to the theoretical limit is quite low, especially if we consider that the sphere-packing bound has not been corrected with the shaping loss in the drawn curves.

Performance Comparison, K = 188 bytes, Rate = 1/2

100

binary irregular LDPC regular LDPC GF(64) regular LDPC GF(256) SP59 lower bound

10−1

Frame Error Rate

10−2

10−3

10−4

10−5

10−6

10−7

10−8

0

0.5

1

1.5

2 Eb/N0 (in dB)

2.5

3

3.5

Figure 6.5 Performance comparison binary vs. NB-LDPC codes. N D 3008 coded bits and rate R D 1=2

New Air Interface Technologies

209

Performance Comparison, K = 53 bytes, Rate = 2/3 100 binary irregular LDPC regular LDPC GF(64) regular LDPC GF(256) SP59 lower bound

10−1

Frame Error Rate

10−2

10−3

10−4

10−5

10−6

10−7

10−8

0

1

1.5

2

2.5

3 3.5 Eb/N0 (in dB)

4

4.5

5

5.5

Figure 6.6 Performance comparison binary vs. NB-LDPC codes. N D 564 coded bits and rate R =2/3

6.2.3 Decoding 6.2.3.1 Soft Decision Algebraic Decoders We first establish the notation that will be used in this section. Let GF(q) be the finite field with q elements, and D D {x1 ; : : : ; x n } stand for a set of points over GF(q). An RS code Cq .n; k/ of code length n and information length k is defined by def Cq .n; k/ D {. f .x1 /; : : : ; f .x n // : x1 ; : : : ; x n 2 D; f .X / 2 G F.q/[X ]; deg f .X / < k}. Given the vector r D [r1 ;
: : : ; rn ] observed Ð at the channel output, we compute 5 D [³i j ], where ³i j D Pr f .x j / D i
1/=r for i D 1; : : : ; q and j D 1; : : : ; n. Furthermore, let M D [m i j ] denote the multiplicity matrix with m i j corresponding to ³i j ; 1  i  q; 1  j  n. With respect to algebraic soft decision decoding of RS codes [32]–[34], Koetter and Vardy’s (KV) decoding algorithm consists of three key steps: multiplicity calculation, bivariate interpolation and factorization.

210

Technologies for the Wireless Future – Volume 3

Multiplicity Calculation Pq P Define m.C 0 / D {M : iD1 njD1 m i j Ð .m i j C 1/=2 D C0 }. This step solves the following problem: for a given cost C0 (the number of constraints or unknowns in the interpolation def step) Pq and Pn reliability matrix 5, choose M 2 m.C 0 / to maximize the score, < M; 5 > D jD1 m i j ³i j , of the transmitted codeword [34]. It is known that the complexity of the iD1 original multiplicity calculation algorithm in [34] is O(n3 / for a full reliability matrix and O(n2 / for a sparse one. Observing that the object value can be expressed as an increasing function with respect to an independent variable ½, we use the following bisection method for fast multiplicity calculation, with complexity reducing to O(n2 / for full reliability matrix and O(n) for a sparse one: Input: Reliability matrix 5, the cost C0 and ". " is a predetermined small positive number, say, 0.001. Pq P Output: Multiplicity matrix M such that C0
C M D iD1 njD1 m i j Ð .m i j C 1/=2. r P P ½  h P P i q q n n 2 2 Step 1: Set ½ L :D
n C n 2 C 8C0 Ð ³ ³ 2 Ð jD1 i j jD1 i j iD1 iD1 .h P P i q .C/ 1 .C/ .C/ .C/ n 2 and ½ H :D ½1 D ½0 C [2½0 Ð n C n Ð q] n C 2½0 Ð ³ . Š Pq Pn ł iD1 ŠjD1ł i j Step 2: Set ½ :D ½ L C ² Ð .½ H
½ L / and C M :D iD1 jD1 ½ Ð ³i j Ð . ½ Ð ³i j C 1/=2. If C M < C0 , set ½ L :D ½; else set ½ H :D ½. Step 3: If ½ H
½ L  " return M :D b½ L Ð 5c; else go to step 2. 1 ½.C/ 0 D



Bivariate Interpolation With the input points P D {.x 1 ; y1 /; : : : ; .x S ; yS /} and their corresponding multiplicities M P D {m 1 ; : : : ; m S }, this step constructs a nonzero polynomial Q.X; Y / of minimal .1; k
1/-weighted degree that passes through points in P with multiplicities M P . A fast algorithm to compute the bivariate polynomial is available at [35]. We refer to the algorithm as iterative interpolation algorithm (IIA). Input: {.xi ; yi ; m i / : .xi ; yi / 2 P} and S P Initialization: Q v .X; Y / D ltD0 qv;t .X /Y t 0  v  l, with the largest Y-degree l: Iteration: for .s D 1I s  SI s C C/ Ov D deg1;k
1 Q v .X; Y /; for 0  v  l: for .a D 0I a < m s I a C C/ for .b D 0I b < m s
aI b C C/ for .v D 0I v  lI v C C/ dv.a;b/ D coe f .Q v .X C xi ; Y C yi /; X a Y b / end If it exists
D arg min0vl;d .a;b/ 6D0 {Ov } v

for .v D 0I v  lI v C C/ if v 6D
and dv.a;b/ 6D 0 Q v .X; Y / :D Q v .X; Y / C

.a;b/

dv

.a;b/

d


Q
.X; Y /

end end Q
.X; Y / :D Q
.X; Y /.X
xi /and O
:D O
C 1 end end end end Output: Q.X; Y / D {Q
.X; Y /}, where
D arg min0vl {Ov }.

New Air Interface Technologies

211

Factorization This step solves the following problem: given the bivariate polynomial Q.X; Y /, list all the factors of Q.X; Y / of type Y
f .X / with degree deg f .X / < k. Efficient algorithms to accomplish the factorization can be found in the references of [34]. Here we give a procedure by Roth [36]: // A global array [0; : : : ; k
1] is assumed. The initial call needs to be with Q.X; Y / 6D 0; k > 0 and i D 0. Procedure: Reconstruct (bivariate polynomial Q.X; Y /, integer k, integer i/. Find the largest integer r such that Q.X; Y /= X r is a bivariate polynomial; Set M.X; Y / :D Q.X; Y /= X r ; Find all the roots over GF(q) of the univariate polynomial M.0; Y /; For each of the distinct roots  of M.0; Y / do { [i] :D  I If .i DD k
1/ output [0]; ; Ð Ð Ð ; [k
1]; _ _ Q Q else {{M .X; Y / :D M.X; Y C /I M.X; Y / :D M .X; X Y /I Reconstruct. M.X; Y /; ~k; i/;}}

6.2.3.2 Graph vs. Trellis Decoding Algorithms BP-based Algorithms Decoding algorithms for LDPC codes operate on a factor graph, i.e. a graphical representation of the partity-check matrix made up of variable nodes, check nodes and edges (see Figure 6.1). A decoding algorithm passes messages along the edges, first from variable to check nodes, then vice versa, etc. The principal decoding algorithm for decoding LDPC codes is known as ‘belief propagation’ (BP) or Gallager’s decoding algorithm [7]. For this method, the messages sent back and forth are probability distributions of the code digits, and the mappings at the nodes are a posteriori probability calculations. The ‘log-BP’ or ‘sum product algorithm’ implements BP in the logarithmic domain. Messages are log-likelihood ratios (LLRs) L D log (p0 = p1 / for binary codes. If L ch is the channel LLR and L 1i : : : L di are the incoming messages of a node, then the outgoing message L o1 for edge 1 is computed as: X L (6.5) L o1 D L ch C kD2:::d ki for variable nodes, and: L o1 D 2 arctanh [5kD2:::d tanh .L ki =2/]

(6.6)

for check nodes. An alternative way to compute Equation 6.6 is by recursive use of the following formula: arctanh [tanh.a=2/tanh.b=2/] D min.a; b/ C log [.1 C e
.aCb/ /=.1 C e
ja
bj /]

(6.7)

For nonbinary codes over GF(q/, decoding in the logarithmic domain has been introduced in [37], by using so-called pseudo-log-likelihood ratios. The algorithms described so far are optimal for infinite code length. Simplified algorithms offer a range of choices on the complexity vs. performance trade-off scale.

212

Technologies for the Wireless Future – Volume 3

Operations in the check nodes can be replaced by a simple minimization, giving the min-sum algorithm (MSA). The MSA looses typically between 0.5 and 1 dB for long regular binary LDPC codes, and more for irregular codes. For nonbinary codes, an approximation similar to the MSA for decoding in the logarithmic domain has been presented in [38]. It is possible to regain part of the losses incurred by the MSA by post-processing the minimum. Heuristic post-processing approaches have been proposed in the literature (linear and affine post-processing), and the optimal post-processing mapping was derived analytically in [39]–[41]. A further approximation of the sum–product algorithm is the ½-min algorithm [42]. For the binary symmetric channel, Gallager’s algorithms A, B, and others [27] use binary messages only and provide a very low-complexity alternative to BP-based methods. Finally, recent developments may yield new classes of practical decoding algorithms for LDPC codes. One class of algorithms [43] is based on methods from statistical physics and equates the problem of decoding an LDPC code to the problem of minimizing the Bethe free energy of a system. Another class of algorithms [44] broadens the range of allowable values for the code digits from binary to an interval of real numbers, and thus replaces the integer programming problem of decoding by a problem that can be solved via linear programming. Under certain conditions, the solution found can be mapped back to a solution for the decoding problem. BCJR-based Algorithms Iterative decoding of PCCC, or convolutional turbo codes (CTC), relies on SISO algorithms (soft-in/soft-out), which exploit a priori information and output a posteriori information. The optimal decoding algorithm of PCCC, or convolutional turbo codes (CTC), is based on the minimization of the probability of bit error given independent inputs. This algorithm, called BCJR [27], is also known as MAP algorithm (maximum a posteriori), or forward–backward algorithm. It produces a posteriori probabilities (APP): P.u k D ijreceived sequence y/

(6.8)

Because of the complexity involved in the numerical representation of the probabilities and the computation processing, decoding is performed in the logarithm domain, with log-APPs. The decoder operates on log-likelihood ratios (LLRs), defined as: L.uO k / D L.ujy/ D ln

P.u k D C1jy/ P.u k D
1jy/

(6.9)

For a systematic code, the soft output information for bit u is given: L.uO k / D L c Ð y C L.u k / C L ext .uO k / L(uk): a priori values for information bits

Soft-input Soft-output Decoder

Lc*y: channel values for all coded bits

Figure 6.7

ˆ k): extrinsic values for Lext(u all information bits L(u ˆ k): a posteriori values for all information bits

Soft-in/soft-out decoder

(6.10)

New Air Interface Technologies

213

The log-MAP [28] is the equivalent of the MAP algorithm in the logarithm domain. It enables equivalent performances without the complexity limitations. Several sub-optimal algorithms have been derived, based on simplifications: the max-log-MAP algorithm [31], the enhanced max-log-MAP [45], the constant-log-MAP [46] and the linear-log-MAP [47]. The soft-output Viterbi algorithm (SOVA) [48] represents another possible decoding algorithm. Several performance and complexity comparisons of these algorithms have been performed [49], [50]. Typically, max-log-MAP with appropriate scaling factors seems to be the best trade-off between performance and complexity. 6.2.4 Architecture and HW Requirements This section gives a short overview of the architecture and hardware requirements of LDPC and turbo decoders based on the IEEE 802.16e standard, considered by the WiMAX Forum. This standard defines duo-binary turbo codes and LDPC codes as their optional channel coding schemes. Architecture and synthesis results are presented for both decoder types. However, the implementation results are very difficult to compare since both decoders have a different degree of flexibility concerning the block length, supported code rates and throughput. 6.2.4.1 LDPC Decoders One straightforward LDPC decoder implementation is a fully parallel implementation where the Tanner graph is directly mapped to hardware. Each variable node (VN) and check node (CN) is instantiated and the connectivity is hardwired. The resulting decoder has a very high throughput, which was shown in [51]. However, two major problems can be seen immediately. For larger block lengths, the routing congestions make this approach infeasible, and the support of different block lengths and code rates is very difficult. Upcoming standards require a flexibility regarding block lengths and code rates, while supporting a high decoding throughput. Thus a partly parallel architecture becomes mandatory to process LDPC codes as specified in DVB-S2, WiFi or WiMAX standard. For partly parallel decoder architectures, only a subset of VNs/CNs are instantiated, while a permutation network has to provide the connectivity. Figure 6.8 shows an LDPC decoder architecture which can process all codes specified by the WiMAX standard. P D 96 functional units are instantiated in a serial manner which corresponds to a processing of P D 96 edges of the Tanner graph per clock cycle. The decoder utilizes a layered decoding scheduling, i.e. updated CN messages are directly passed back to VNs and participate immediately in the iterative decoding process. The major advantages of the layered architecture are the fast convergence speed and a smaller area compared to a classical two-phase architecture. However, not all codes can be processed in a straightforward manner. Scheduling methods and sorting of the code vectors are mandatory if the code is not explicitly designed for layered decoding. The parallelization P is determined by the maximum submatrix size z D 96 of the structured code, which is composed of permuted identity matrices. The mapping of the VNs and CNs to the functional units is determined by the given code structure. Efficient mapping of the nodes to the functional units can be guaranteed in all cases by designing LDPC codes using permuted identity matrices. z VNs/CNs with the connection described by a z-by-z identity matrix are always allocated to z distinct VFU and z distinct CFU respectively. This mapping was shown by many LDPC decoder realizations [52]–[55].

Channel RAM

Channel RAM

Controller

Channel RAM

Technologies for the Wireless Future – Volume 3

Address RAM

214

FIFO

+

−

CFU

Msg RAM

Permutation RAM

Permutation Network Π

CNB

CNB

+ CNB

Figure 6.8 LDPC decoder architecture for layered decoding

A permutation network has to realize the permutation of the identity matrix which results in a simple barrel shifter. For the WiMAX code the size of permuted identity matrices varies from z D 24 to z D 96 in steps of 4. Thus the network has to mirror this kind of flexibility, too. A logarithmic barrel shifter is utilized, which is composed of modular cells which provide wraparounds at 19 positions, one for each supported block size. Each check node block (CNB) is composed of a check node functional unit (CFU), a FIFO which bypasses the CN input information, and a memory (Msg RAM) which stores the extrinsic information. As already mentioned, the CFUs are instantiated in serial manner, which allows for high code rate flexibility. To reach reasonable communications performance and minimize the area consumption, the CFUs are implemented with a hardware-optimized version of the min-sum algorithm with correction term. For the layered processing we do not explicitly instantiate variable nodes. This is achieved by storing the input information of the CFU in a FIFO, and adds the new extrinsic information to the corresponding input message bypassed by the FIFO. Thus the correct a posteriori information is passed back to the channel memory, which always holds the updated VN information. This basic concept was first presented by [56]. The disadvantage when passing already updated VN information to the channel RAMs is the increased bit width. The input channel values are represented by 6 bit; an a posteriori message is represented by 9 bit. These 9 bit messages have to be passed through the barrel shifter as well. However, only one shifter

New Air Interface Technologies

215

Table 6.1 Synthesis results for the WiMAX 802.16e LDPC decoder [57] turbo codes WiMAX 802.16e LDPC decoder Synthesis results: 65 nm @ 400 MHz Area [mm2] Logic CNB

Shuffling Network

RAMs

Overall AREA

Infobits/Cycle

Max. Efficiency [Mbps/mm2]

Throughput [Mbps]

0.35

0.18

0.46

0.99

0.18–1.6

646

72–640

is instantiated accepting messages of 9 bits. A second shift network, as in the case of the two-phase processing, is not required for the layered processing since we can ensure the correct message retrieval by a modified offset value which takes the old shift value into account. Implementation Results Table 6.1 presents the synthesis results of an LDPC decoder architecture using an ST microelectronics 65 nm technology. The decoder is capable of processing all LDPC codes specified by the WiMAX 802.16e standard. Six different macro parity check matrices are specified with code rates between R D 1/2 and R D 5/6, with block lengths ranging from 576 to 2304 bits. Only two code rates are specified directly for layered processing; all other code rates require a dedicated stalling and scheduling process to ensure the correct APP calculation. A quantization of 6 bits is utilized for the incoming channel LLRs, and 9 bit messages for the exchanged APP messages. The overall area is about 0.99 mm2 , with ³ 50 % determined by the logic part. This high portion of the area costs results mostly from the required flexibility of the shuffling network and the CFU. All memories in this architecture are based on single port RAMs The throughput numbers are calculated based on a clock frequency of f D 400 MHz. The throughput mainly depends on the number of edges in the graph, and of course the number of iterations. The throughput numbers in Table 6.1 show the throughput range for R D 1/2 to R D 5/6. For the R D 1/2 code, 15 iterations are assumed, and for the highest code rate, 10 iterations. The latency can be roughly calculated by simply dividing the codeword size N by the associated throughput number. Note that it is always possible to decrease the throughput by decreasing the clock frequency, which will in turn lead to an increased latency. The overall area can be further reduced by instantiating only a smaller portion of the functional units, e.g. with P D 24. Then the area of the logic part is roughly 1/4 of the given logic area, while the RAM area remains nearly constant, since the same number of edges has to be stored in all cases. However, the throughput will scale down by a factor of 4. In Figure 6.9 a parallel concatenated turbo code decoder is shown with two component decoders as utilized in the UMTS, the CDMA2000 and the WiMAX standard. The decoding of turbo codes is based on the exchange of extrinsic information, which is denoted as LLR values in the following independent if this value reflects exchanged information of a binary TC or duo-binary TC. For a TC decoder architecture, two major problems have to be solved: the realization of the utilized SISO algorithm and the exchange of the extrinsic information.

216

Technologies for the Wireless Future – Volume 3

SISO Decoder 1

λs p0

λ

INT

MEM 2 Λ

SISO Decoder 2 λ

Figure 6.9

DEINT

MEM 2 Λ

p1

Basic parallel concatenated turbo decoder

Interleaving The exchanged information has to be interleaved and stored before passing the next SISO decoder. This is indicated by 31 and 32 memories and the INT and DEINT boxes in Figure 6.9. The required throughput determines the parallelization of the SISO and thus the number of LLRs which have to be interleaved or deinterleaved per clock cycle. To build high–throughput turbo decoders it is mandatory to produce 2, 4 or even 8 LLR values per clock cycle. Then the interleaving can be a major bottleneck for a hardware realization. If the interleaver is not carefully designed, dedicated interleaver networks become mandatory to resolve occurring memory access conflicts. Different techniques for solving this interleaving problems are presented in [58]. For the WiMAX codes, the interleavers are designed with respect to a parallelization of P D 4, which enables parallel interleaving without any memory access conflicts occurring. SISO Architecture For the SISO decoder, the utilized algorithm could be the SOVA or the log-MAP algorithm. In [59] it was shown that the log-MAP algorithm is better in terms of communications performance and has a better scalability for VLSI realization. A SISO decoder has to produce several LLRs per clock cycle to achieve higher throughputs. Windowing is a well-known technique that allows the dividing of the entire block into individual sections [60]. Each section can be processed independently, while the boundary states of the corresponding trellis sections are initialized by an acquisition phase. Each section can be allocated to a separate SISO instance, which is denoted as producer in the following. How many producers are instantiated depends on the required throughput. Within a producer which processes a section of a codeword, a further windowing scheme can be applied. One possible windowing scheme for a one-producer architecture which produces one LLR value per clock cycle is shown in Figure 6.10. The block is divided into four windows. At each time step, three different computations can take place: one acquisition calculation, one forward recursion calculation (Þ/ and one backward recursion (þ/ with the final LLR calculation [58]. Figure 6.11 shows the basic concept of a corresponding SISO architecture with three recursion units (RU). One RU calculates the forward recursion, one the backward recursion and one the acquisition calculation. Further components are the branch metric unit (BMU), state metric memory (SMM), LLR unit (LLRU) and buffer control, which has to ensure the correct input value to each BMU. The major challenge for the SISO decoder realization is the correct synchronization of these recursion units and the data exchange between them. A further challenge is to provide the corresponding branch metrics to each of these RU units. To enable this, the incoming channel LLRs and the a priori information have to be stored in the indicated buffers. These buffers are mandatory to ensure the correct sequence, which is determined by

New Air Interface Technologies

217

Position index within a block

K = 4WL Window 4 3WL Window 3 2WL Window 2

Acquisiton

WL+ACQ WL b Recursion + LLR calc.

Window 1 a Recursion 1

Processing time t

Figure 6.10 Windowing scheme for one-producer SISO

l

BMU

buf 1 buf 2 buf 3

Buffer Control

BMU

BMU

g

acq RU binit

g

bwd RU

g

fwd RU

b+g LLRU

Λ

SMM a

Figure 6.11 SISO architecture with three recursion units

the chosen windowing scheme. Since one recursion, alpha or beta, has to be done prior to the other to obtain the final LLR values, the intermediate results have to be stored in the state metric memory (SMM). The required memory depends on the window size and, of course, the number of states. Many different possibilities exist to optimize the state metric memory, which is pretty often a trade off between required throughput, latency and processing power. The final LLR calculation is done in the LLR unit (LLRU), which also calculates the hard decision outputs. Implementation Results Table 6.2 shows the synthesis results of a WiMAX 802.16e duo-binary TC decoder based on a 65 nm technology of ST microelectronics. The supported codeword length is 128–4800 information bits, with a maximum number of 9600 parity bits. Thus code rates from R D 1/3 to R D 5/6 are supported. The quantization of input data and exchanged extrinsic information is 6 bits and 8 bits, respectively. The SISO decoder features a two-producer architecture and

218

Technologies for the Wireless Future – Volume 3

Table 6.2 Synthesis results for the WiMAX 802.16e duo-binary turbo decoder [61] WiMAX 802.16e Duo-binary Turbo decoder Synthesis results: 65 nm @ 400 MHz Area [mm2] ACS/LLR Units

2-Prod. SISO

RAMs

Overall AREA

Infobits/Cycle

Max. Efficiency [Mbps/mm2]

Throughput [Mbps]

0.15

0.2

1.58

1.8

0.06–0.23

51

24–92

is thus capable of processing two LLRs per clock cycle. The resulting net throughput ranges from 24 to 92 MBit/s, which depends on the block length. The throughput is independent of the code rate, which is in contrast to a typical LDPC decoder implementation. The size of the two-producer SISO architecture is 0.2 mm2 , which is mainly determined by the ACS units and the LLR units. The ACS units are implemented to support a max-log-MAP algorithm with scaling factor [45]. The decoder is highly memory dominated and can be extended to a four-producer architecture, which would double both the logic part and the throughput. 6.2.5 Turbo Principle 6.2.5.1 Introduction The WWRF-WG4 white paper of the WWRF#17 meeting, dedicated to error control codes, provides a complete tutorial on turbo decoding principles and their implementations through various forward error control types. Turbo-FEC applications are exposed in considering standardization issues. The turbo concept shall be extended to other digital signal processing of wireless communication to improve receiver performance. Joachim Hagenauer draws an overview of the ‘turbo principle’ applied to wireless communications [62] and its extension to coded MIMO and precoded QAM symbols with irregular channel codes. The turbo principle is a general principle in decoding and detection where a digital processing is integrated in a local loop characterized with two inputs and one output. One of the inputs uses the receiver information and the second one exploits the digital processing of the local loop structure issued from the previous iteration associated with the feedback of the extrinsic information of the digital signal processing. This structure shall be described as a parallel or a serially concatenated system. The turbo processing can be used in many applications to improve performance through iterative processing. The wireless communications, such as concatenated codes, equalization, coded modulation, multiple-input/multiple-output (MIMO) detection, joint source and channel decoding, may exploit turbo structures to significantly improve performance. The turbo principle is also efficiently exploited in multicarrier channel estimation where the system uses an iterative processing including the extrinsic information to refine and reduce noise level on estimated complex gains of the multipath channel [63], [64]. The common processing related to the turbo principle is the interleaving processing included within the local loop which provides diversity to the turbo processing. In the next section, we describe a novel application of the turbo principle at the transmitter side, which utilizes this

New Air Interface Technologies

219

principle to generate in a flexible way interleaving patterns for either inner FEC interleaving, outer binary interleaving or subcarrier mapping allocation in a multicarrier system. 6.2.5.2 Turbo-based Interleaving The turbo-like interleaving algorithm is clearly detailed in the WWRF-WG4 white paper of the WWRF#17 meeting and implemented for short-range radio transmissions at 60 GHz in [65], [66]. In this section, we give the main properties of the algorithm. General Principles of the Algorithm The proposed algorithm arises from internal studies on optimized and scalable interleaving designs where we attempt to define an interleaving processing which maximizes the interleaving spreading (distance between input sample position index after the interleaving processing) while ensuring a compatibility with special data multiplexing issued from independent digital processing modules resulting in baseband processing optimization. The interleaving spreading corresponds to the minimum distance between input sample indexes after the interleaving processing. To control interleaving spreading, Crozier et al. [67], [68] prove that interleaving systems utilizing modulo operations are well suited to high distance interleavers, even though they suffer from periodic structures in the case of large interleaving spreading. The proposed algorithm combines a second property detailed below. The second desired property consists in preserving a special data partitioning with a size p upon a single stream interleaver. This property may be translated in considering that the single stream interleaving is equivalent to a parallel multistream interleaver considering p virtual streams (Figure 6.12). The number p of virtual streams corresponds to the interleaving partitioning size of the proposed algorithm that is composed of p samples, each issued from a virtual stream. In that way, the interleaving processing does not break dedicated data multiplexing resulting from pre and post processing. This constraint has been initially motivated by inner frame oriented turbo code interleaving design requirements to avoid tail bits resulting in self-terminated trellis [69]. This property is translated in a mathematical rule imposing that the difference between I(k) and k is a multiple of p: I .k/
k D Q.k/ ð p

(6.11)

These properties are assessed, for example, to optimize the binary interleaving spreading in considering both the binary interleaving spreading within every data symbol and between adjacent data symbols composed each of m bits with 2m constellation points. For that purpose, we select interleaving parameters that maximize the interleaving spreading between bits separated by s-1 bits where s is ranged from 1 to m and s is a multiple of m. Furthermore, this property is promising for future MIMO interleaving and OFDMA subcarrier mapping allocation, as proposed in [70]. The turbo-based structure results from scalability requirements in connection nonstationary multipath transmissions. Turbo-like Interleaving Structure The proposed algorithm is a block interleaving with a size K which defines the permutation unit thanks to two turbo-based algebraic functions combining modulo operations, and two integer parameters p and q. This basic structure I(k) is integrated in a transmitter turbo structure composed of two parallel inputs and a single output. One of the inputs is fed with the

220

Technologies for the Wireless Future – Volume 3

Virtual stream #1

Stream #1 interleaving

X(k)

Stream #2 interleaving

Interleaving S/P

Ip,q(j )(k) interleaving

Y(k) S/P

Y(k) = X(I(n)(k)) interleaving Stream #p Virtual stream #p

X0 X1 X2.. Xpo−1 Xpo Xp−1

Xp Xp+1 X2p−1

Xpq X1+pq X2+pq X3+pq

X′0 X′1 X′2 X′3 …X′p−1 X′p….X′2p−1 X′2p

..Xp(q+1)−1

X′pq X′1+pq X′2+pq X′3+pq

..X′p(q+1)−1

Figure 6.12 The equivalent multistream interleaving of the proposed algorithm

initial subcarrier index position of the input sequence S, and the second one is fed with the interleaving pattern output of the previous iteration (Figure 6.13). The interleaving iteration (j), the parameters p and q modify the interleaving spreading of the algorithm using the same mathematical description I(k) that led to a scalable transmitter . j/ turbo-based interleaver design I p;q .k/. This algorithm provides large gains when the propagation channel is time and frequency-selective and when we consider high modulation levels. Results are illustrated through several publications and many IEEE standardization groups (IEEE802.15.3c, IEEE802.22, B42 project). (j )

Ip, q (k)

Ip, q (k)

Ip, q(2)(k)

k I

I

I

I

I(k) = f(p, q, K, k) k Io

(j−1)

Ip, q (k)

(j )

I1

Ip, q (k)

(j )

Ip, q (k) = I1 (k, Io(j )(k))

Figure 6.13 Transmitter turbo structure of the proposed algorithm

New Air Interface Technologies

221

6.2.6 Standardization Overview 6.2.6.1 Cellular (UMTS, 3GPP LTE) Advanced error-correcting schemes have been introduced into standards defining third generation (3G) wireless systems. Turbo codes have been introduced in the late 90s in the UMTS specification, edited by 3GPP, and in the ‘cdma 2000’ standard, supported by 3GPP2. In both cases, binary turbo codes, based on 8-state constituent codes, have been specified. The UMTS turbo code has a natural coding rate of 1/3, with generator polynomials (13,15)8 . Trellis termination is performed with the addition of tail bits. The interleaver is defined for any information blocksize in the range 40–5114 bits. The 3GPP2 has a lower natural coding rate (1/5) and can cover information blocksizes up to 8192 bits. Discussions occurred in the working group named ‘Long Term Evolution’ (LTE) under 3GPP, in charge of defining an improved UMTS standard, and it was decided to modify the internal interleaver of the UMTS binary turbo code, in order to improve parallelism and meet LTE throughput requirements (100 Mbps).

6.2.6.2 IEEE 802.16/WiMAX The IEEE 802.16 standard define mobile and fixed broadband wireless access [71], and include three different advanced FEC ž Block turbo code (also known as turbo product codes) – these codes are obtained from the 2D product of two block codes, among a list of possible constituent block codes such as extended Hamming codes and parity codes. Different block sizes can be matched through deletion of some of the rows and/or columns, and through combinations of different constituent codes. ž Convolutional turbo codes – duo-binary turbo codes [72] are included in the specifications [71], similar to the turbo codes standardized into DVB-RCS [73] and DVB-RCT [74]. This turbo code is based on 8-state constituent codes. The constituent encoders operate on couples of data bits instead of single bits. Circular encoding is applied, avoiding the requirements for tail bits at the end of each codeword. Adjustment of the turbo codes to different blocksizes is performed through modifications of four parameters, which are optimized for each blocksize. The turbo code can therefore cover a wide range of blocksizes with limited memory requirements. Different coding rates are enabled through puncturing. High throughputs can be achieved as the internal interleaver of the DB TC can be designed to allow different degrees of parallelism. ž LDPC codes – The LDPC codes used are block type LDPC [75], relying on base model matrices (nb D 24 columns) for each coding rate, and expansion process for ensuring codeword length scalability (n D nb * Z f /. There are four different coding rates supported, Rc D 1/2, 2/3, 3/4 and 5/6, and the codeword length scalability is supported from 576 to 2304 bits, or equivalently from an expansion factor Z f D 24 to 96 by step of 4, which means 96 bit granularity. It’s worth noticing that, contrary to [76], only six different parity-check matrices (Rc D 2/3 and 3/4 propose two options each) are defined, then used for any codeword length.

222

Technologies for the Wireless Future – Volume 3

6.2.6.3 IEEE 802.11n The IEEE 802.11 family of standards defines wireless local area networks (WLAN), and the next-generation IEEE 802.11n enables higher throughputs, includes LDPC codes as an optional coding scheme. The structure of such codes still relies on block type LDPC, with expansion process from base model matrices. As only three different codeword lengths are supported (648, 1296 and 1944 coded bits), such base model matrices have been optimized for each CW case. Besides, four different coding rates, 1/2, 2/3, 3/4 and 5/6 are supported. Thus the standard includes 12 optimized parity check matrices for each CW and coding rate [76], ensuring a certain scalability in information length from 324 to 1620 bits (11 different sizes). 6.2.6.4 Satellite (DVB-RCS, DVB-S2) Satellite communication is a particular field of interest for the application of advanced error-correcting codes, as very low packet error rates are often required and the restrictions on complexity are usually less stringent than for other communications systems. Turbo codes were first introduced into satellite communication standards: the CCSDS (Consultative Committee for Space Data Systems) standard includes a binary turbo code, while the broadcasting standard DVB-RCS (return channel for satellite distribution systems) specifies a duo-binary turbo code. The DVB-S2 standard [77] proposes a brand new coding scheme constituted with serial concatenation of an outer BCH code with an inner LDPC code. Only two different frame formats are proposed (64 800 and 16 200 bits). While the BCH code is still targeting correction of minor remaining errors, and thus lowering error floors, the LDPC code in use is an irregular repeat accumulate (IRA) code. Full details of the Tanner graph parameters can be found in [53], following instructions given in [77]. The DVB-S2 standard has attracted a lot of industry attention, leading to some concrete solution proposals for its LDPC coded implementation. We’d like to particularly point out the following two references of high interest for further reading: [53] (IST 4MORE project) and [77]. It is claimed that such new proposal BCH/LDPC leads to 30 % capacity increase compared with the former DVB-S Reed–Solomon/convolution code. 6.2.6.5 Others Other standards have included advanced error-correcting schemes in their specification. The DECT (digital cordless telecommunication standard), an ETSI standard for digital portable phones, and IEEE 802.20 (mobile broadband wireless access), still under discussion at the IEEE, included a binary turbo code in their specifications. Duo-binary turbo codes are also present in the standard DVB-RCT (return terrestrial channel for digital television). Finally, a European FP6-funded project, called WINNER, has been investigating duo-binary turbo codes and block-LDPC codes for the concept design on B3G/4G cellular system. In-depth investigations and fair comparisons are available [80], pointing out the pros and cons of each scheme. Some domains of suitability [79] are highlighted, depending on the packet length, coding rates, etc., thus emphasizing the increasing interest for a generic/reconfigurable decoder.

New Air Interface Technologies

223

6.2.7 Conclusions and Challenges Through this entire section our first intention was to highlight and make accessible to the greatest number the theoretical breakthroughs, as well as the technological achievements and challenges we still need to address in the near future whilst designing the ECC schemes for next-generation wireless systems. Whilst considering theoretical aspects of capacity approaching ECC options, a global trend towards investigating a unifying framework can be noticed lately, especially by the use of Tanner graphs and their generalization. Following this concept, turbo codes and LDPC codes can be seen as the offspring of the same family of codes. Such directions fulfil the generic property requested within reconfigurable systems. From a ‘real-world’ point of view, many implementation aspects have been covered in this section, introducing brand new results concerning the latest standardized codes. We can see that the LDPC codes technology roadmap is moving fast, and they should soon be as mature as their predecessors, turbo codes, thus gaining more weight in next-generation standards since they provide cost-efficient solutions. Certainly, LDPC offers many advantages in the encoding and decoding of large blocks of data, of more than 1000 bits. Some interesting aspects have also been covered considering short block length transmission, for which the capacity approaching schemes based on LDPC or turbo code deviate from ‘perfectness’. We provided directions towards suboptimal decoders for block length below, say, 1000 bits. Finally, the soft decision decoding is once again promoted through the turbo principle, which is likely to become mandatory when designing new systems, due to its wide range of applications (iterative channel estimation, equalization, synchronization, etc.) and its impressive performance gain. The desired features of modern ECC schemes for next-generation wireless systems are usually: ž ‘Close to capacity’ performance over real-world wireless channels. ž Low complexity and ease of implementation. ž Generic decoding procedures in the sense that the same family of codes is reconfigured to different cases (e.g. adjusting the parameters of the code to long block length in the downlink and to short block length for the uplink, increasing the error correction capabilities for users with low SNR conditions, etc.). ž Adaptive complexity to channel conditions (i.e. reduced complexity when channel conditions are improved) to improve power-efficiency of mobile systems. This section has demonstrated that these are inherent key features of future coding components, and proposed some guidelines for future R&D. The big challenge of error control coding for next-generation wireless systems has not yet been completely resolved. Some topics for further work and open issues: ž It is still necessary to adapt the ECC scheme to the real-world conditions and characteristics of the channel. Performance results demonstrated by bound-meeting ECC for the AWGN channel give rise to the optimistic belief that a range of practically decodable, ‘near-optimal’ codes will be found to answer the real-world channel impairment.

224

Technologies for the Wireless Future – Volume 3

ž Bound meeting ECC for short packet length of, say, less than 1000 bits that reach the SP59 within 0.5 dB and less and have the desired features mentioned above. ž Efficient implementations for nonbinary codes.

6.3 Multi-dimensional Channel Modeling Advanced multi-dimensional radio channel models are needed to meet the simulation and testing requirements of future wireless systems. The requirements include new frequency bands and propagation scenarios, large bandwidth, advanced multi-antenna structures and new network topologies, for example. This section reviews the state of the art of multi-dimensional radio channel modeling, as well as the recent developments in channel model standardization activities. Future wireless communication systems are expected to employ a wider range of carrier frequencies than before due to the limited spectrum resources and because of novel high data rate services that require wide bandwidths. It is obvious that a key enabler to improve spectral efficiency significantly is the adaptive multi-antenna technology, which constantly takes advantage of the characteristics of wideband multi-dimensional radio channels. In addition to the basic radio channel research, the development and standardization of realistic multi-dimensional radio channel models is therefore critically important for the design, optimization and evaluation of future wideband wireless communication systems. The generic development process of a standardized channel model is illustrated in Figure 6.14. Most standardized radio channel models are limited to relatively narrow bandwidths (5 MHz) and to frequency bands below 2.5 GHz. They are based on extensive radio channel measurement campaigns [81]–[91]. The specification of multi-dimensional radio channel models for multiple input multiple output (MIMO) systems is an ongoing task for the novel radio access systems. Indeed, the development of future wireless communication systems requires realistic, measurement-based radio channel models, which are specified in standardization bodies such as ITU, IEEE and ETSI. Since the standardization of radio systems is an ongoing task, investigation and definition of new radio channel models must continue as an indispensable accompanying activity. The existing knowledge about propagation and channel modeling is definitely not adequate to support the future mobile radio system development. The importance of channel modeling activities for system design and standardization follows from information theory, which clearly states that it is the channel that sets the fundamental performance limits in any transmission system. Advanced system design must be aimed at finding an economic balance between system performance and implementation effort. A precise understanding of the propagation mechanisms and the resulting channel statistics is needed in order to assess the performance of a system design. It may also help to overcome performance limitations by changing the system paradigm.

Basic Radio Channel Research

Applied RC Research

PreStandardization

Standardization

Standard model

Figure 6.14 Generic process for developing a standardized radio channel model

New Air Interface Technologies

225

The idea of MIMO, for example, was unleashed from the understanding of multichannel propagation in rich scattering environment. The expected high spectral efficiency of future wireless communication systems relies on the assumption that the complicated wideband phenomena of multi-antenna radio channels can be fully exploited. However, there are still many open issues which remain unclear from the overall system performance point of view. For example, network planning, antenna deployment, near-field effects, feedback requirements for radio channel state information and the actual implementation of transmitter/receiver adaptivity may have a significant impact on the achievable spectral efficiency. Understanding the critical parameters associated with radio channel modeling is therefore crucial for next-generation wireless system engineering [92]. 6.3.1 Multi-dimensional Modeling Methods Radio channel models for MIMO systems are typically applied in spatial matrix modulation transceiver architectures, in which a number of parallel data streams are transmitted simultaneously after undergoing spatio-temporal coding. Figure 6.15 illustrates a system model for the MIMO communication scheme. The task of the transmitter is to ‘match’ the whole transmit scheme to the characteristics of the MIMO radio channel, and the task of the receiver is to extract MIMO channel parameters for efficient reception of the transmitted information. If the MIMO radio channel model has to reflect various propagation aspects, the realistic representation of the physical propagation environment is necessary. The modeling can be classified as deterministic or stochastic. Deterministic modeling involves a detailed reproduction of the actual physical wave propagation process. Thus, the geometry of the specific environment and the characteristics of the reflecting/scattering surfaces are used as input to compute channel properties. In stochastic modeling, a large number of measurements are conducted in order to define a statistical representation of the channel. For example, in cellular systems models can be created for macro, micro and pico cells, which may further include several subcategories such as urban, suburban or hilly terrain. Stochastic channel modeling is often based on the assumption of space-time stationarity. It may hence not be adequate to describe rapidly time-varying dynamics or the performance of adaptive reconfigurable techniques that attempt to adjust to abrupt statistical changes of the environment or the interference. In such cases, a more accurate deterministic description of the channel may be required, which is usually the case with nearly all the investigations

M antennas

RTX

scatterer

MIMO radio channel

N antennas

HMxN RRX

Access point

Mobile terminal

Figure 6.15 Communication system with M transmit and N receive antennas and an MxN MIMO radio channel H. Spatial correlation matrices RT X and R R X define the inter-antenna correlation properties

226

Technologies for the Wireless Future – Volume 3

related to handset terminals. On the other hand, the deterministic models are site-specific and they are not capable of fully characterizing the scenario in general. In addition, they tend to have high numerical complexity. Geometry-based stochastic channel models enable accurate modeling of the physical phenomena of the radio channel while extending generality with stochastic features. In such an approach the probability density function (PDF) of the geometrical location of the scatterers is prescribed. For each channel realization, the scatterers’ locations are taken randomly from this PDF, and the multipath delays and directions are calculated from the TX-RX-scatterer geometry. The PDF must be selected in such a way that the resulting power-delay profiles (PDP) and power-azimuth spectra (PAS) agree reasonably well with the measured values. The scatterers are usually grouped into clusters as suggested by the measurement results. 6.3.2 Recent Results Currently, several ongoing collaborative research projects carry out studies on channel modeling. In Europe, COST has finalized the COST273 project and reported the results in a book [97], and it continues with the COST2100 project [98]. The IST-WINNER project (Wireless World Initiative New Radio [99]) is working towards specifying next-generation air interface and the enabling technologies, including channel models. The IST-PULSERS project studies ultra wideband (UWB) systems. In the IST-MAGNET project, extensive wideband and UWB MIMO radio propagation investigations are carried out for personal and body area network scenarios. Multi-dimensional channel models were also developed in the IST-NEWCOM, which is a network-of-excellence type of cooperation under the European Commission’s 6th Framework Programme. 6.3.2.1 WINNER Project The IST-WINNER project is focusing on beyond-3G (B3G) radio system design, using frequency bandwidths of up to 100 MHz [99]. Within the channel modeling activities of the WINNER project, double directional channel models have been developed for the WINNER investigation scenarios. At the beginning of the project, initial channel models were selected from the literature for early link level and system level simulations. Then the IST-METRA project-based IEEE 802.11n channel model [105] was selected for indoor simulations and the 3GPP SCM [91] was selected for outdoor simulations. The SCM model has a bandwidth of 5 MHz with a centre frequency of 2 GHz. However, the required bandwidth and centre frequency in the WINNER project are up to 100 MHz and at 2–6 GHz respectively. For that reason, a new channel model was implemented, called SCME (SCM extension) [104]. Even with these modifications, the initial channel models were proved to be insufficient for the WINNER system simulations, due to the different requirements (see Table 6.3). This led to the development of new spatio-temporal models within WINNER. The WINNER I models are described in [116]. Channel model parameters have been specified for seven different scenarios. The developed models are based on both literature and extensive measurement campaigns of the WINNER project (Figures 6.16 and 6.18). There are two types of channel model for each scenario: a generic model and a reduced-variability model. The latter is denoted as ‘clustered delay line’ (CDL) model, which can, for example, be used for calibration and comparison simulations. Matlab implementation of WINNER I channel models is available online [118].

New Air Interface Technologies

227

During the second phase of the WINNER project (IST-WINNER II), the channel modeling work has been continued in order to refine and complete the models proposed during the first phase. The updated models cover a wide scope of propagation scenarios and environments, including the indoor-to-outdoor and outdoor-to-indoor models (Figure 6.17), a bad urban micro-cell, a bad urban macro-cell, a feeder link from a base station to a fixed relay station, and moving networks. The WINNER modeling approach is based on a generic propagation characterization that is antenna independent. This enables simple variations of the number of antenna elements, the antenna array configurations (geometry) and the antenna beam patterns, without changing the basic propagation model. Within WINNER II a realistic 3D antenna model [135] was supported. This approach enabled the use of the same channel data in different link level and system level simulations and it was well suited to evaluation of adaptive radio links, equalization techniques, coding, modulation and other transceiver techniques. The results have been published in [117]. WINNER phase I models incorporate clusters with azimuth spread, but with no delay spread. The refined channel models of WINNER phase II model intra-cluster delay spread with three mid-paths which have relative delay intervals of 0 ns, 5 ns and 10 ns. However, if all the clusters are spread in the delay domain, the simulation complexity is increased. A trade-off solution is therefore to use intra-cluster delay spread in only the one or two strongest clusters. This already has the positive effect that the frequency correlation becomes more realistic. Another new feature in WINNER II models is a smooth transition between drops modeled by a birth–death process of clusters. SCM, SCME, WINNER I and WINNER II models are compared in Tables 6.3 and 6.4 [119] . 6.3.2.2 COST 273 & COST 2100 The COST 273 channel modeling activities are a result of a long traditional research work, whereby the proposed models are stepwise generalized from initial path loss and delay dispersion models (COST 207) to more special environment models (COST 231) and to directional channel modeling (COST 259).

Figure 6.16 One measurement setup in a rural environment

228

Technologies for the Wireless Future – Volume 3

Figure 6.17 One measurement setup in an outdoor-to-indoor environment

Figure 6.18 One measurement setup in a suburban environment

The recently published double directional model in [97] is a geometry-based stochastic approach, which is based on the twin-cluster concept to represent multiplied reflected or diffracted multipath components (MPC). The locations of the two twin clusters can be chosen independently, in order to correctly reflect angle of arrival (AoAs) and angle of departure (AoDs). The transferring between the two clusters adds a delay that also matches the measured delay of the MPCs. Furthermore, the model parameters are based on statistics derived from measurement data. Hence, different environments can be modeled based on the same channel model core. A public version of the channel model implementation is available in [100]. A new COST action, COST 2100, was launched in December 2006 with a topic ‘Pervasive Mobile & Ambient Wireless Communications’. COST 2100 is organized in three working groups, which deal with propagation and antennas, physical layer and radio network aspects. In Working Group 2 (radio channel), the basic research work is continued in order to improve the understanding and the modeling of radio propagation in single and multiple antenna systems. The goal is to define simplified but accurate models intended for radio

New Air Interface Technologies

229

Table 6.3 Channel model features Feature

SCM

SCME

WINNER I

WINNER II

Bandwidth > 100 MHz Indoor scenarios Outdoor-to-indoor and indoor-to-outdoor scenarios AoA/AoD elevation Intra-cluster delay spread TDL model based on the generic model Cross-correlation between large scale parameters Time evolution of model parameters

No No No No No No No No

Yes No No No Yes Yes No YesŁŁ

Yes Yes No Yes No YesŁ Yes No

Yes Yes Yes Yes Yes Yes Yes Yes

Ł TDL ŁŁ

model is based on the same measurements as generic model, but analyzed separately. Continuous time evolution.

Table 6.4 Channel model parameters Parameter

Unit

SCM

SCME

WINNER I

WINNER II

Max. bandwidth Frequency range No. of scenarios No. of clusters No. of mid-paths per cluster No. of sub-paths per cluster No. of taps BS angle spread MS angle spread Delay spread Shadow fading standard deviation

MHz GHz

5 2 3 6 1

100Ł 2–6 3 6 3–4

100ŁŁ 2–6 7 4 – 24 1

100** 2–6 14 8 – 20 1/3

20

20

10

20

6 5 – 19 68 170 – 650 4 – 10

18 – 24 4.7 – 18.2 62.2 – 67.8 231 – 841 4 – 10

4 – 24 3.0 – 38.0 9.5 – 53.0 1.6 – 313.0 1.4 – 8.0

12 – 24 3 – 58 16 – 53 16 – 240 3–8

Ž Ž

Ns dB

Ł extrapolation ŁŁ based

from 5 MHz bandwidth on 100 MHz measurements

systems simulation and standardization. An important objective of COST 2100 is also to address the new requirements imposed on the definition of radio channel properties for the successful development of novel wireless communication systems. Specific scenarios, parameters and system configurations will be identified, which may impact the radio channel. The main topics include physical propagation mechanisms at specific bands of the radio spectrum, radio propagation modeling techniques, measurements and advanced analysis of the double directional radio channels, MIMO channel sounding and modeling, and UWB channel sounding and model extraction. Research topics also include the propagation of multihop, relay and sensor networks, the frequency dependence of the propagation, and consequences on radio channels for multi-standard and cognitive radio systems [115].

230

Technologies for the Wireless Future – Volume 3

6.3.2.3 Wing TV The emerging broadcasting standards, like DVB-H (digital video broadcasting – handhelds) will use SFN (single frequency network). There was therefore a need for models specifically targeting the broadcasting radio channel and, to this end, a work item was identified in the pan-European Celtic project Wing-TV to investigate the measurement-based radio channel models for DVB-H [112]. According to the SFN principle, several transmitters are transmitting the same data on the same frequency band. Thus, the impulse response seen by the DVB-H receiver corresponds to a channel with very long echoes, since the terminal receives signals from two (or more) transmitters at the same frequency. It is worth noticing that there is often a clear line-of-sight component. It becomes apparent that the number of taps for the radio channel simulations must be larger than in typical cellular test cases due to the SFN nature (i.e. several transmitters are visible). It can also be seen from the measurements that all the delay values are larger than in normal cellular test cases, reflecting the relatively large size of a DVB-H cell. Thus, the final radio channel models have to take into account this issue. The final, simplified 12-tap models (with affordable complexity for computer and hardware simulations) have been proposed in [113]. The model has been verified with laboratory experiments. The experiments revealed that the radio channel model has a clear effect on the radio system performance under test. The difference between a six-tap typical urban model (which is generally accepted as the worst-case scenario in DVB-H testing) and the Rayleigh model is around 7 dB, thus giving a lot of uncertainty to the design. The new model, Pedestrian Indoor, results in performance levels somewhere in between, giving a more realistic view on the system design. The work started in Wing-TV on realistic radio channel models is currently progressing within the pan-European Celtic project B21C [114]. The key objective of the B21C project radio channel modeling is to characterize the MIMO utilization in broadcasting and define satellite propagation models for DVB-SH. Special attention will be given to the case of dual polarized antennas in MIMO.

6.3.3 MIMO Radio Channel Models in Wireless Standards Standardized channel models provide a full description of the parameterization and stochastic properties along with a number of test cases for specific values of the model parameters. These test cases offer a number of scenarios to be considered for simulations and aim at establishing comparability in terms of channel models for calibration and performance evaluation purposes. The projects COST, CODIT and METRA (funded by the European Commission) provided channel modeling methodology without proposing specific test cases. The newly developed SCM and IEEE 802.11n models include specific model parameters for outdoor and indoor scenarios, respectively. The COST259 model [101] is a model framework enabling tuning of the three-level parameter set of a specific simulation environment. 6.3.3.1 3GPP SCM The scope of the 3GPP-3GPP2 Spatial Channel Model (SCM) Ad Hoc Group was to develop and specify parameters and methods associated with the spatial channel modeling that are

New Air Interface Technologies

231

common to the needs of the 3GPP and 3GPP2 organizations [91]. The model includes two sub-models: calibration (link-level) model simulation (system-level) model. A. Calibration Model The SCM link-level channel model considers a MIMO link, where a single transmitter sends a signal through a radio channel to a single receiver. It is a spatial extension of the ITU-R Rec. M.1225 IMT-2000 model [103]. The spatial extension includes Laplacian distributed rms angle spread (AS) values for the base station and the mobile terminal, and propagation path directions in three environments: urban macro cell, urban micro cell and suburban macro cell. B. Simulation Model The SCM system-level model follows COST 259 [101], which is a geometry-based stochastic model with a three-level top-to-bottom structure. Also, the SCM model is based on geometry, but a subset of the parameters is stochastic. It is designed for the bandwidth of 5 MHz and the centre frequency of 2 GHz. System-level simulations often include multiple base stations and multiple mobile terminals. A drop-based model is assumed for the generation of the clusters of scatterers. During a drop, the channel undergoes fast fading according to the mobile movement. However, delays and AoA/AoD are kept constant. Two consecutive drops are independent and include randomly located clusters, which make the channel model discontinuous. The channel modeling methodology of SCM can be described in three steps: 1. Specify an environment: suburban macro, urban macro or urban micro. 2. Obtain the parameters to be used in simulations associated with that environment. 3. Generate the channel coefficients based on the parameters. The channel model includes several parameters, such as number of paths, number of subpaths, mean angular spread (AS) at BS and MS, AS per path at BS and MS, AoA/AoD distributions, variable delay spread, path loss and shadowing. These parameters are fixed in the specification of a number of test cases. The number of paths in all test cases is six. 6.3.3.2 3GPP LTE Extended spatial channel models based on [104] were proposed for 3GPP long-term evolution (LTE) and adopted for initial system evaluation [106]–[109]. The multi-antenna channel model is a clustered delay line model with per-cluster covariance matrices describing the fast fading correlation and power distribution over transmit and receive antennas. The model has three mid-paths (paths separated by delay) in each cluster. Thus, the number of simulation taps is three times higher than in SCM. The benefit of the model is that it describes more accurately the frequency correlation in frequency selective 20 MHz channels than the original SCM. The LTE channel models cover four scenarios: suburban macro, urban macro (low spread), urban macro (high spread) and urban micro. Mobile station arrangement can be ‘handset – talk position’, ‘handset – data position’ and ‘laptop’. Channel models for LTE conformance tests are under discussion. A simplified approach where ITU-R models [103] are extended by a few taps and with correlation matrix for MIMO was proposed.

232

Technologies for the Wireless Future – Volume 3

6.3.3.3 IEEE 802.11n The IEEE 802.11n model has been developed for indoor WLAN high-throughput applications. Different environments exist, from small office to large open space (models A–F) [110]. Both line-of-sight (LOS) and non line-of-sight (NLOS) cases are included in the models. RMS delay spreads range from 15 to 150 ns. It is assumed that the propagation environment can be modeled using the cluster principle. The number of clusters and multi-path components vary between models. The maximum number of delay positions considered is 18. The 802.11n model uses tap-specific AoA and AoD characteristics and antenna geometry. Transmitter and receiver correlation matrices are calculated analytically from the geometry. The MIMO correlation matrix is obtained via the Kronecker product of the spatial correlation matrices at transmit and receive antenna arrays. The actual model includes tables of parameters for seven indoor environments. The following parameters and features are defined: cluster structure and excess delay, power, AoA, AoD, angular spread of arrival and departure, incidence angles for each multipath component, path loss, shadow fading, deterministic LOS component, Doppler components due to fluorescent lights, and cross-polarization discrimination. 6.3.3.4 IEEE 802.16 In the IEEE 802.16e standard, MIMO channel models are not defined. Therefore, channel model and test cases are specified in the WiMAX Forum [120]. The proposed models are based on ITU IMT-2000 tap delay line models [103]. Complex correlation matrices are defined for each tap. IEEE 802.16m was initiated recently, and several channel model contributions have already been submitted. The discussed models cover more advanced double-directional geometry-based stochastic channel models. 6.3.3.5 ITU-R IMT-Advanced International Telecommunication Union (ITU) is the leading United Nations agency for information and communication technologies [121]. In addition to a General Secretariat, ITU comprises three specialized sectors dealing with Radiocommunication (ITU-R), Telecommunication Standardization (ITU-T) and Telecommunication Development (ITU-D). ITU-R focuses on rational, equitable, efficient and economical use of the radio-frequency spectrum. ITU-R Working Party WP8F has started the process leading to standardization of IMT-Advanced (4G) [122], [123]. The work on IMT-Advanced was moved to Working Party WP5D and the first WP5D meeting was held on January 2008. Several contributions on channel models were presented in WP8F/WP5D meetings in Yaoundé, Cameroon, Kyoto, Japan, Geneva, Switzerland, and in Dubai, United Arab Emirates, in January 2007–June 2008. Most of the channel model contributions are based on the geometry-based stochastic approach. The same approach can be applied for all propagation scenarios from short-range to wide-area. In the ITU-organized World Radiocommunication Conference (WRC-07) in Geneva, Switzerland, 22 October–16 November 2007, the globally harmonized spectrum identified for use by International Mobile Telecommunications (IMT) represents an important step in the worldwide development of IMT systems. The bands are 450–470 MHz, 698–862 MHz, 2.3–2.4 GHz and 3.4–3.6 GHz. Some of them do not have global allocation, but were accepted by many countries [124].

New Air Interface Technologies

233

Table 6.5 Comparison of the basic features of two standardized MIMO channel models Parameter

Unit

SCM

.11n

Max. bandwidth Frequency range Number of scenarios Wide-area scenarios Short-range scenarios Suitability for link-level simulations Suitability for system-level simulations Probability of LOS Kronecker approximation Ray based / correlation based Antenna Independency Antenna Pattern Spatial Correlation Doppler obtained from geometry

MHz GHz

5 2 3 Y N Y Y Y N R Y Y Implicit Y

<100Ł 2&5 6 N Y Y N N Y CŁŁ N N Explicit N

Ł Tap

Y/N Y/N Y/N Y/N Y/N Y/N R/C Y/N Y/N Y/N

spacing is 10 ns. Thus frequency correlation is 1.0 in 100 MHz offset. matrix calculated from the geometry-based model (AoA/AoD).

ŁŁ Correlation

6.3.3.6 Comparison Table 6.5 compares briefly two well-known standardized MIMO radio channel models. The comparison is very interesting because the compared models represent two major approaches in MIMO channel modeling – geometric- and correlation matrix-based approaches [126]. From the comparison, it can be observed that the correlation-based model offers further insight on the impact of certain parameters, such as spatial correlation, whereas the geometric model offers explicit representation of the scattering environment, independence from the antenna array characteristics and suitability for system level analysis. For up to 4 ð 4 MIMO (16 antenna element pairs), the complexity of the correlation- and geometry-based methods is similar. For larger MIMO constellations the correlation approach is more complex. The computational complexity is not in any case a clear argument to prefer the correlation matrix-based method to the geometric method. Regardless of the choice between these two methods, simulation complexity (convolution) dominates the overall complexity [125]. 6.3.4 Future Research Topics and Open Issues There are still numerous open issues in developing realistic multi-dimensional channel models for future wireless communication systems. These include – but are not limited to – complexity, inter-cell interference modeling, and stationarity and scale effects. There is an enduring discrepancy between realistic modeling and computational complexity requirements in the standardization of MIMO channel models. When standardized models are used for link or system-level performance evaluation with software or hardware simulations: ž Reduction of simulation time is a strong argument to keep the models simple. ž Backward compatibility is often required to keep the new results comparable to the old ones. In some cases this has led to very artificial and unrealistic models, when the

234

Technologies for the Wireless Future – Volume 3

old narrowband and nondirectional channel models are modified to MIMO-compatible models. ž Increased number of features and improved accuracy leads to more specialized and complex models. Furthermore, system-level performance evaluation of future radio access technologies requires accurate and realistic modeling of the intra- and inter-cell interference. This is especially true for multiple link and cooperative systems, such as distributed MIMO, relay and multihop systems, and sensor networks. Thus modeling of intra- and inter-cell radio correlations of large-scale channel parameters will become important for evaluation of the future cooperative systems, in which link-level transmission schemes should be optimized at system (network) level. The current research results are providing only partial insight and therefore further characterizations of these phenomena are necessary [130]. For stochastic modeling, stationarity during simulation runs is generally assumed. At present, the length of these stationarity intervals is often just an ‘educated guess’. Transition between different scenarios including LOS< >NLOS is still not adequately represented in existing channel models. Additional issues are that different visions exist on whether changes in stochastic regime are large-scale effects, and that small-scale processes are not necessarily stochastically uniform. Up to now it is not clear where the dividing line is between large- and small-scale effects. Recent measurements have shown that simple specular path modeling is not enough to reproduce the measured channel capacity. In addition, modeling of diffuse scattering is necessary [92], [133]. However, identification and modeling of the spatial structure of diffuse scattering is still an open task which needs further research [131]. Existing channel models and measurement procedures do not adequately consider the three-dimensional (3D) nature of the radio wave propagation. This includes realistic 3D antenna models [135] and their polarimetric response [127]. It has been shown that neglecting a dimension (e.g. elevation) and/or the polarimetric response may severely deteriorate measured channel parameters [134] and can lead to completely wrong results.

6.4 Multiuser MIMO Systems 6.4.1 Introduction Space division multiple access (SDMA) promises high gains in the system throughput of wireless multiple antenna systems. If SDMA is used on the downlink of a multiuser MIMO system, either long-term or short-term channel state information has to be available at the base station (BS) to facilitate the joint precoding of the signals intended for the different users. Precoding is used to efficiently eliminate or suppress multiuser interference (MUI) via beamforming or by using ‘dirty-paper’ codes. It also allows us to perform most of the complex processing at the BS, leading to a simplification of the mobile terminals. Linear precoding techniques have an advantage in terms of computational complexity. Significant gains can be expected by beamforming to the averaged channel if long-term channel state information (CSI) is available at the BS. Even higher gains are obtained by beamforming to the instantaneous channel if short-term CSI is available at the BS, but some of the benefits

New Air Interface Technologies

235

might be wiped out by channel estimation errors or delays between measuring the channel and forming the beams. Nonlinear techniques have a higher computational complexity and require some signaling overhead but might provide a better performance than linear techniques. There has been considerable interest in MIMO wireless communications in view of their potential for a significant improvement in channel capacity [136]–[140]. In multiuser scenarios, several co-channel users with multiple antennas aim to communicate with a base station in the same frequency and time slots. In this case, it becomes necessary to design transmission schemes that are able to suppress the co-channel interference (CCI) at the end users. Multiple antennas at the base station and the user terminals of future wireless communication systems allow spatial multiplexing of signals intended for different users on the downlink and spatial separation of signals simultaneously transmitted by a number of users on the uplink. The information theoretic results on the sum capacity of the multiuser MIMO downlink system have shown that some kind of Costa or Tomlinson-Harashima precoding (THP) is necessary to attain it [138], [141]–[144]. ‘Dirty-paper’ codes (DPC) can achieve the maximum sum rate of the system and provide the maximum diversity order. However, these techniques require the use of a complex sphere decoder [143] or an approximate closest-point solution [144], which makes them hard to implement in practice. A simple but practical technique that eliminates a part of the MUI and improves the system diversity is THP. THP is a nonlinear precoding technique, originally developed for single input single output (SISO) multipath channels, which has been proposed for the pre-equalization of the spatial interference in [145]. In this section we provide a review and comparison of linear and nonlinear multiuser MIMO precoding techniques. Linear techniques are covered in Section 6.4.2 and nonlinear techniques are covered in Section 6.4.3. In Section 6.4.4 we compare these techniques and show the influence of channel estimation errors and scheduling on their performance. In Section 6.4.5 we give a short summary. 6.4.2 Linear Precoding Techniques We consider a multiuser (MU) MIMO downlink channel, where MT transmit antennas are located at the base station and M Ri receive antennas are located at the ith mobile station (MS), i D 1; 2; : : : ; K . There are K users (user terminals (UTs)) in the system. The total number of receive antennas is: MR D

K X

M Ri

(6.12)

iD1

A block diagram of such a system is depicted in Figure 6.19. Block diagonalization (BD) is a linear precoding technique for the downlink of MU MIMO systems [146], [147]. It decomposes an MU MIMO downlink channel into multiple parallel independent single-user MIMO downlink channels. The signal of each user is preprocessed at the transmitter using a modulation matrix that lies in the null space of all other users’ channel matrices. The MUI in the system is thereby efficiently set to zero. BD can be used with any other previously defined single-user MIMO technique [148], as the different users do not

236

Technologies for the Wireless Future – Volume 3

interfere with each other. BD is attractive if the users are equipped with more than one antenna. However, the zero MUI constraint can lead to a significant capacity loss when the users’ subspaces significantly overlap. It also imposes a dimensionality constraint, i.e. the number of transmit antennas at the base station has to be greater than or equal to the total number of receive antennas at the user terminals. Another technique also proposed in [146], named successive optimization (SO), addresses the power minimization and the near–far problem, and it can yield better results in some situations but its performance depends on the power allocation and the order in which the users’ signals are preprocessed. The zero MUI constraint is relaxed and a certain amount of interference is allowed. Minimum mean-square-error (MMSE) precoding improves the system performance by allowing a certain amount of interference, especially for users equipped with a single antenna. However, it suffers a performance loss when it attempts to mitigate the interference between two closely spaced antennas, a situation that occurs when the user terminal is equipped with more than one receive antenna. In [149] the authors proposed a new algorithm called successive MMSE (SMMSE) which deals with this problem by successively calculating the columns of the precoding matrix for each of the receive antennas separately. The complexity of this algorithm is only slightly higher than the one of BD but it can provide a higher diversity gain and a larger array gain than BD. In [150] it was shown that a zero-forcing beamforming (ZFBF) strategy, while generally suboptimal, can achieve the same asymptotic sum capacity as that of DPC, as the number of users goes to infinity. Regularized block diagonalization (RBD) precoding improves the system performance even further, since it is able to extract the full array and diversity gain. The precoding matrices are designed in two steps. In the first step, the overlap of the row spaces of different users is minimized. Here, instead of forcing the multiuser interference to zero, we introduce a new cost function in order to balance multiuser interference and noise enhancement. In the second step, the system performance can be optimized for each user separately, since the channel is decomposed into a set of parallel single-user (SU) MIMO channels. Hence, in the second step SU-MIMO optimization techniques can be used, which allow optimizing for low BER, high SNR or maximum throughput. The technique is called regularized BD since at high signal-to-noise ratios, and if the number of transmit antennas is not smaller than the total number of receive antennas, the effective combined channel matrix becomes block diagonal, as in BD. RBD is discussed in detail in [161], along with a generalized framework for the design of MU-MIMO precoding matrices.

1

s1

F1

s2

F2

sK

FK

Base Station (BS)

D1

MR1

1

1

MT

MR2

r1 D2

r2

1

DK rK MRK User terminals (UT)

Figure 6.19 Block diagram of multiuser MIMO downlink system

New Air Interface Technologies

237

6.4.3 Nonlinear Precoding Techniques It is well known that linear equalization suffers from noise enhancement and hence has a poor power efficiency in some cases. The same drawback is experienced by linear precoding, which combats noise by boosting the transmit power. This disadvantage of linear equalization at the receiver side can be avoided by decision-feedback equalization (DFE), which is for instance used in V-BLAST [151]. Tomlinson-Harashima precoding (THP) is a nonlinear precoding technique developed for single input single output (SISO) multipath channels. THP can be interpreted as moving the feedback part of the DFE to the transmitter. Recently it has also been applied for the pre-equalization of MUI in MIMO systems [145], where it performs spatial pre-equalization instead of temporal pre-equalization for ISI channels. Thereby, no error propagation occurs. Hence, the precoding can be performed for the interference-free channel. MMSE precoding in combination with THP is proposed in [152]. MMSE balances the MUI in order to reduce the performance loss that occurs with zero-interference techniques, while THP is used to reduce the MUI and to improve the diversity. SMMSE in combination with THP [153] reaches the sum-rate capacity of the broadcast channel at low SNRs. However, at high SNRs it fails to reach the maximum sum-rate capacity of broadcast channels. Another advantage of SMMSE THP is that it is not restricted by the number of antennas at the user terminals and at the base station. The total number of antennas at the user terminals can be larger than the number of antennas at the base station as long as the number of users is less than or equal to the rank of the combined channel matrix. SO THP, proposed in [154], combines SO and THP in order to reduce the capacity loss due to the cancellation of overlapping subspaces of different users and to eliminate the MUI. After the precoding, the resulting equivalent combined channel matrix of all users is again block diagonal. This also facilitates the definition of a new ordering algorithm. Unlike in [141] and [152], this technique allows more than one antenna at the mobile terminals and has no performance loss due to the cancellation of interference between the signals transmitted to two closely spaced antennas at the same terminal. The advantage of this technique is that it is a zero-forcing technique which achieves the maximum sum rate capacity, but only when the total number of antennas at the users’ terminals is less than or equal to the number of antennas at the base station. The modulo operators at the receivers allow for a more general choice of the additive perturbation signal than obtained by THP, as was highlighted by Hochwald, Peel and Swindlehurst in [143]. They proposed to use a linear transformation at the transmitter, whose input is the data signal superimposed with a perturbation signal with properties known from THP, i.e. its entries are integer multiples of the modulo constant. Since the algorithm to find the perturbation signal is closely related to the sphere decoder, the algorithm was named ‘sphere encoder’ [143]. In [144], Windpassinger et al. replaced the sphere decoder necessary for finding the perturbation signal with the respective lattice reduction-aided detector. In [155] the authors propose a minimum mean square error (MMSE) solution to vector precoding for frequency flat multiuser scenarios with a centralized multi-antenna transmitter. The receivers employ a modulo operation, giving the transmitter the additional degree of freedom to choose a perturbation vector. Similar to existing vector precoding techniques, the optimum perturbation vector is found with a closest point search in a lattice. The MMSE vector precoder does not, however, search for the perturbation vector resulting in the lowest transmit energy, as

238

Technologies for the Wireless Future – Volume 3

proposed in all previous contributions on vector precoding, but finds a compromise between noise enhancement and residual interference. In order to reach the maximum sum rate capacity of broadcast channels, a technique proposed in [156] combines zero forcing with successive encoding (ZF-SE), proposed for the downlink of multiuser systems with single receive antennas, and the singular value decomposition (SVD) technique, which decomposes the single-user channel while preserving capacity. The number of spatial dimensions allocated to each user and the encoding order of the allocated subchannels are two degrees of freedom of this technique which can be exploited for system design. The technique results in a set of virtually decoupled scalar subchannels, upon which power allocation can be performed according to any criterion of interest. 6.4.4 Simulation Results In this section we will compare the performance of BD, SO THP and SMMSE precoding. To do so, we take into account a purely stochastic spatially-uncorrelated flat-fading channel Hw and a frequency-selective MIMO channel with the power delay profile as defined by IEEE 802.11n -D with non line-of-sight conditions [157]. We assume data transmission using an OFDM system with a DFT of length N D 64, a subcarrier spacing of 150 kHz and a cyclic prefix that is Npre D 4 samples long. Data is encoded using the convolutional code rate 1/2 (561 753)oct . After coding, the data is mapped using QAM and 16 QAM modulation. Coded and modulated symbols are transmitted using Nc D 48 subcarriers and Nsymb D 2 OFDM symbols. The real advantage of SMMSE can be seen from the BER performance shown in Figure 6.20. By allowing MUI, SMMSE provides a higher diversity and a higher array gain than BD. SO THP does not have the same diversity gain as BD that can be explained with the influence of the modulo operator used in THP. IEEE 802.11D, {2, 2} × 4, {N, Npre, Nc = 48, Nsimb = 2} = {64, 4, 48, 2}, QAM CC1/2 100 SMMSE ch. est. error BD ch. est. error SO THP ch. est. error

BER

10−1

10−2

10−3 12

14

16

18

20

22

PT/s2n [dB]

Figure 6.20 BER performance comparison of BD, SO THP and SMMSE in configuration

New Air Interface Technologies

239

Next, we will investigate the influence of scheduling on the performance of the MU MIMO precoding techniques using the IlmProp channel model [158]. All precoding techniques suffer from two major drawbacks: they can only spatially multiplex a limited number of users in any given resource slot and their performance largely degrades when serving spatially-correlated users. In a real system it is reasonable to assume that there exists a larger number of users than those a BS can simultaneously transmit to. Therefore, there is a need for a scheduling algorithm responsible for deciding which users are to be spatially multiplexed at any given time and frequency. The scheduler should take its decisions avoiding grouping spatially-correlated users, maximizing the system performance, while remaining fair [159], [162]. In Figure 6.21 we show the 10 % outage capacity as a function of the ratio of total transmit power PT and additive white Gaussian noise at the input of every antenna ¦ n2 . In Figure 6.22 we investigate the influence of scheduling on the BER performance of BD and SMMSE. As can be seen in these two figures, spatial scheduling has a great impact on the capacity of the system. The capacity gain is about 1 bps/Hz and the SNR gain is about 2 dB. We also see that SMMSE clearly outperforms BD. In Figure 6.23 we compare the capacity of SMMSE, SMMSE THP and BD to the capacity of a TDMA system and the DPC bound [160]. SMMSE outperforms BD at low SNRs, while BD has higher capacity at the high SNRs. By combining SMMSE and THP we manage to reach the sum rate capacity of the broadcast channels at the low SNRs. SMMSE THP outperforms both SMMSE and BD. In Figure 6.24 we compare the BER performance of SMMSE and SMMSE THP. SMMSE THP provides higher diversity than SMMSE at high SNRs. However, the point where SMMSE THP becomes better than SMMSE depends on the antenna configuration and channel statistics. Thus, it can happen that SMMSE provides a better performance than SMMSE THP at the SNRs of interest. {2, 2} × 4, N = 64, f0 = 150 kHz 11 SMMSE, scheduling SMMSE, no scheduling BD, scheduling BD, no scheduling

10 9

C [bps/Hz]

8 7 6 5 4 3 2 1

16

18

20

24 22 PT/s2n [dB]

26

28

30

Figure 6.21 10 % outage capacity performance of BD and SMMSE, with and without scheduling

240

Technologies for the Wireless Future – Volume 3

{2, 2} × 4, N = 64, f0 = 150 kHz

100

BER

10−1

10−2

10−3

10−4

SMMSE, scheduling SMMSE, no scheduling BD, scheduling BD, no scheduling 10

12

14

16

18

20

PT/s2n [dB]

Figure 6.22 BER performance comparison of BD and SMMSE in configuration {2, 2} ð 4 with and without scheduling

Hw, {2, 2} × 4 20 SMMSE BD SMMSE THP DPC bound TDMA

10 % Outage capacity [bps/Hz]

18 16 14 12 10 8 6 4 2 0

0

5

10 PT/s2n

15

20

[dB]

Figure 6.23 10 % outage capacity performance of BD, SMMSE and SMMSE THP

New Air Interface Technologies

241

{N, Nc, Npre, Nsymb} = {64, 48, 4, 2} 100 SMMSE SMMSE THP

10−1 10−2

BER

10−3 10−4 10−5 10−6 10−7 10−8 10

Hcorrelated {2, 2, 2} × 8 IEEE802.11n D f0 = 150 kHz QAM CC 1/2 12

14

16

18

20

22

24

26

PT/s2n [dB]

Figure 6.24 BER performance comparison of SMMSE and SMMSE THP

{2, 2, 2} × 6, HW, flat fading channel 100 BD SMMSE RBD

Uncoded BER

10−1

10−2

10−3

10−4

10−5 −2

0

2

4

6 PT/s2n

8

10

12

[dB]

Figure 6.25 BER performance comparison of BD, SMMSE and RBD

14

242

Technologies for the Wireless Future – Volume 3

Finally, Figure 6.25 shows the comparison of BD and SMMSE with RBD in a Rayleigh fading environment for a fully loaded system. We consider six antennas at the base station and two antennas at each of the three user terminals. The graphic displays the uncoded bit error rate versus the signal-to-noise ratio. We can clearly see that the BER for SMMSE is superior to that of BD. Also, RBD outperforms both methods. Moreover it can be seen that the diversity order extracted by the RBD algorithm is higher than that of all the other techniques. 6.4.5 Conclusion In this section we have given an overview of different transmit precoding techniques in a downlink multiuser scenario. Depending on the set of constraints, like the size of the overhead or the amount of the MUI allowed, different techniques can be optimal. Linear techniques are computationally less expensive and generally require no signalling overhead. Moreover, they can use either instantaneous or long-term channel state information (CSI) to perform precoding. On the other hand, the nonlinear techniques can provide a higher sum-rate capacity and a higher diversity gain, but can use only instantaneous channel state information for precoding. DPC achieves the maximum sum rate of the system and provides the maximum diversity order. However, these techniques require the use of a complex sphere decoder or an approximate closest-point solution, which makes them hard to implement in practice. THP is strongly related to DPC, and it represents a suboptimal implementation of DPC. By combining linear precoding techniques with a nonlinear technique called Tomlinson-Harashima precoding, we are able to reach the achievable sum-rate capacity of the system when there is perfect CSI available at the transmitter. All precoding techniques suffer from two major drawbacks: they can only spatially multiplex a limited number of users in any given resource slot and their performance largely degrades when serving spatially-correlated users. In a real system it is reasonable to assume that there exists a larger number of users than those a BS can simultaneously transmit to. Therefore, there is a need for a scheduling algorithm [159], [162] responsible for deciding which users are to be spatially multiplexed at any given time and frequency. The scheduler should take its decisions avoiding grouping spatially-correlated users, maximizing the system performance, while remaining fair. In fact, for spatially-correlated users, the MUI is very large, due to the overlapping signal spaces. This of course degrades the performance of non-zero MUI techniques. Forcing the interference to zero, as the zero MUI techniques do, does not help much since it would be inevitable to use inefficient modulation matrices. With appropriate scheduling, linear techniques can also reach the sum-rate capacity of broadcast channels. Which techniques should be used will depend on the price we have to pay to acquire CSI at the transmitter. Using the reciprocity principle, in a TDD system it is possible to use the estimates of the uplink channel to perform precoding on the downlink. In an FDD system the only way to acquire the exact CSI at the transmitter is feedback from the users’ terminals. Low to medium user speeds in indoor, hotspot and microcellular scenarios allow the use of reciprocity in a TDD system to acquire CSI at the transmitter. Nonlinear precoding techniques provide a higher diversity than their linear counterparts at high SNRs. However, the point where nonlinear precoding techniques become better than linear techniques depends on the specific antenna configuration of the system, e.g. the number of antennas at the base station and the number of user terminals and antennas at the user terminals. Linear precoding techniques can achieve the sum-rate capacity bound of the broadcast channel when the

New Air Interface Technologies

243

number of users in the system is large and appropriate spatial scheduling of users is performed. Furthermore, linear precoding techniques allow the combination of perfect and long-term CSI, unlike nonlinear precoding techniques, which require the exact CSI in order to be able to presubtract the noncausal interference. Together with a lower computational complexity, this renders linear precoding techniques more favourable for practical implementation than nonlinear precoding techniques. Related research and development subjects include: ž Efficient pilot allocation and channel estimation schemes in multiuser MIMO systems that use SDMA. ž Efficient feedback signaling of the estimated CSI from the terminals to the base station in FDD systems. ž Signaling and channel estimation concepts for multiuser MIMO precoding schemes with multiple base stations. ž The investigation of the sensitivity of nonlinear precoding and equalization techniques to realistic channel estimation errors.

6.5 Acknowledgements The following individuals contributed to this chapter: Contributors and editors to Sections 6.2 and 6.3 Thierry Lestable (Samsung Electronics Research Institute, UK) and Moshe Ran (Holon Institute of Technology, Israel). Contributors and editors to Sections 6.3 Mr Tommi J¨ams¨a (Elektrobit, Finland), Juha Ylitalo (Elektrobit, Finland), Jukka-Pekka Nuutinen (Elektrobit, Finland), Pekka Ky¨osti (Elektrobit, Finland), Juha Meinil¨a (Elektrobit, Finland), Reiner Thom¨a (Ilmenau University of Technology, Germany), Christian Schneider (Ilmenau University of Technology, Germany), Milan Narandˇzi´c (Ilmenau University of Technology, Germany), Angeliki Alexiou (Bell Labs, Alcatel-Lucent, UK) Contributors and editors to Section 6.4 are Veljko Stankovic, Martin Haardt and Florian R¨omer (Ilmenau University of Technology, Germany), and Simon Gale and Andy Jeffries (Wireless Technology Laboratories, Nortel, UK).

References [1] C.E. Shannon, “Probability of error for optimal codes in a Gaussian channel”, Bell Syst. Tech J., 38, pp. 611–656, 1959. [2] A. Valembois and M.P.C Fossorier, “Sphere-packing bounds revisited for moderate block length”, IEEE Trans. Inform. Theory, IT-50, pp. 2998– 3014, Dec. 2004. [3] H. Nickl, J. Hagenauer and F. Bukert, “Approaching Shannon capacity limit by 0.2dB using simple Hamming codes”, IEEE communication lett., 1(5), pp. 130– 132, Sep. 1997. [4] Telemetry Channel Coding, JPL report from Consultative Committee for Space and data systems (CCSDS), JPL and CalTech, Pasadena, USA, Oct. 1998. [5] L. Wei, “High performance iterative Viterbi algorithm for conventional serial concatenation codes”, IEEE Trans. Inform. Theory, 46, pp. 1759– 1771, Jul. 2002. [6] L. Wei, “Several properties of short LDPC codes”, IEEE Trans. Comm., 52, pp. 721–727, May 2004. [7] R.G. Gallager, “Low-density parity-check codes”, IEEE Trans. on Information Theory, 8, pp. 21–28, Jan. 1962.

244

Technologies for the Wireless Future – Volume 3

[8] T. Richardson and R. Urbanke, “The renaissance of Gallager’s low-density parity-check codes”, IEEE Communications Magazine, 41(8), pp. 126–131, Aug. 2003. [9] A. Ashikhmin, G. Kramer and S. ten Brink, “Extrinsic information transfer functions: model and erasure channel properties”, IEEE Trans. Inform. Theory, 50(11), pp. 2657– 2673, Nov. 2004. [10] S. ten Brink and G. Kramer, “Design of repeat-accumulate codes for iterative detection and decoding”, IEEE Transactions on Signal Processing, 51(11), pp. 2764– 2772, Nov. 2003. [11] S. ten Brink, G. Kramer and A. Ashikhmin, “Design of low-density parity-check codes for modulation and detection”, IEEE Trans. Commun., 52(4), pp. 670–678, Apr. 2004. [12] G. Lechner, J. Sayir and I. Land, “Optimization of LDPC codes for receiver frontends”, 2006 IEEE International Symposium on Information Theory, Seattle, USA, 9–14 Jul. 2006 [13] D. Divsalar and F. Pollara, “Turbo codes for PCS applications”, Proc. Int. Conf. Communications, Seattle, USA, Jun. 1995. [14] R.G. Gallager, “ Low-density parity-check codes ”, MIT Press, Cambridge, MA, USA, 1963. [15] X. Hu, E. Eleftheriou and D. Arnold, “Progressive edge-growth Tanner graphs”, Proc. 2001 Global Telecommunications Conference (Globecom 01), San Antonio, TX, USA, 2001. [16] T.J. Richardson, M.A. Shokrollahi and R.L. Urbanke, “Design of capacity-approaching irregular low density parity-check codes”, IEEE Trans. on Information Theory, 47(2), pp. 619–637, Feb. 2001. [17] M. Davey and D.J.C. MacKay, “Low density parity check codes over GF(q)”, IEEE Commun. Lett., 2, pp. 165– 167, Jun. 1998. [18] D.J.C. MacKay and M. Davey, “Evaluation of Gallager codes for short block length and high rate applications,” Proc. IMA Workshop on Codes, Systems and Graphical Models, 1999. [19] X.-Y. Hu and E. Eleftheriou, “Binary representation of cycle Tanner-graph GF(2q) codes,” Proc. ICC04, pp. 528– 532, Paris, France, Jun. 2004. [20] C. Poulliat, M. Fossorier and D. Declercq, “Design of non binary LDPC codes using their binary images: algebraic properties”, Proc. ISIT06, Seattle, WA, USA, Jul. 2006. [21] D. Sridhara and T.E. Fuja, “Low density parity check codes over groups and rings,” Proc. ITW02, Bangladore, India, Oct. 2002. [22] D. Declercq, M. Colas and G. Gelle, “Regular GF(2q)-LDPC coded modulations for higher order QAM-AWGN channels”, Proc. ISITA04, Parma, Italy, Oct. 2004. [23] A. Bennatan and D. Burshtein, “Design and analysis of nonbinary LDPC codes for arbitrary discrete-memoryless channels,” IEEE Trans. on Inform. Theory, 52(2), pp. 549– 583, Feb. 2006. [24] S. Song, L. Zeng, S. Lin and K. Abdel-Ghaffar, “Algebraic constructions of nonbinary quasi-cyclic LDPC codes”, Proc. ISIT06, Seattle, WA, USA, Jul. 2006. [25] G. Li, I. Fair and W. Krzymien, “Analysis of nonbinary LDPC codes using Gaussian approximation,” Proc. ISIT03, Jul. 2003. [26] G. Byers and F. Takawira, “EXIT charts for non-binary LDPC codes,” Proc. ICC05, 1, pp. 652– 657, May 2005. [27] L. Bahl, J. Cocke, F. Jelinek and J. Raviv, “Optimal decoding of linear codes for minimizing symbol error rate”, IEEE Trans. on Inform. Theory, IT-20, pp. 284– 287, Mar. 1974. [28] P. Robertson, E. Villebrun and P. Hoeher, “A comparison of optimal and suboptimal MAP decoding algorithms in the Log domain”, Proc. ICC95, pp. 1009– 1013, Jun. 1995. [29] C. Poulliat, M. Fossorier and D. Declercq, “Using binary image of nonbinary LDPC codes to improve overall performance,” Proc. Turbo Codes, Munich, Apr. 2006. [30] D. Kimura, F. Guilloud and R. Pyndiah, “Construction of parity-check matrices for non-binary LDPC codes”, Proc. Turbo Codes, Munich, Apr. 2006. [31] P. Robertson, P. Hoeher and E. Villebrun, “Optimal and sub-optimal maximum a posteriori algorithms suitable for turbo decoding”, European Transactions on Telecommunications, pp. 119–125, Mar.–Apr. 1997. [32] M. Sudan, “Decoding of Reed-Solomon codes beyond the error correction bound”, J. Complexity, 12, pp. 180– 193, 1997. [33] V. Guruswami and M. Sudan, “Improved decoding of Reed-Solomon and algebraic-geometric codes”, IEEE Trans. Inform. Theory, 45, pp. 1757– 1767, Sep. 1999.

New Air Interface Technologies

245

[34] R. Koetter and A. Vardy, “Algebraic soft-decision decoding of Reed-Solomon codes”, IEEE Trans. Inform. Theory, 49(11), pp. 2809– 2825, Nov. 2003. [35] G.L. Feng and X. Giraud, “Fast algorithm in Sudan decoding procedure for Reed-Solomon codes”, preprint, Aug. 2002. [36] R.M. Roth, G. Ruckenstein, “Efficient decoding of Reed-Solomon codes beyond half the minimum distance”, IEEE Trans. on IT, 46(1), pp. 246– 257, Jan. 2000. [37] H. Song and J.R. Cruz, “Reduced-complexity decoding of Q-ary LDPC codes for magnetic recording”, IEEE Trans. Magn., 39, pp. 1081– 1087, Mar. 2003. [38] H. Wymeersch, H. Steendam and M. Moeneclaey, “Log-domain decoding of LDPC codes over GF(q)”, Proc. IEEE Intern. Conf. on Commun., Paris, France, pp. 772– 776, Jun. 2004. [39] G. Lechner and J. Sayir, “Improved sum-min decoding of LDPC codes”, Proc. ISITA04, Parma, Italy, pp. 997– 1000, 10–13 Oct. 2004. [40] G. Lechner and J. Sayir, “Improved sum-min decoding for irregular LDPC codes”, Proc. 4th International Symposium on Turbo Codes & Related Topics, Munich, Germany, 3–7 Apr. 2006. [41] G. Lechner and J. Sayir, “Improved sum-min decoding of irregular LDPC codes using nonlinear post-processing”, Proc. NEWCOM-ACoRN Joint Workshop, Vienna, Austria, 20–22 Sep. 2006. [42] F. Guilloud, E. Boutillon and J.L. Danger, “½-min decoding algorithm of regular and irregular LDPC codes”, Proc. 3nd International Symposium on Turbo Codes & Related Topics, Brest, France, pp. 451–454, Sep. 2003. [43] J.S. Yedidia, W.T. Freeman and Y. Weiss, “Bethe free energy, Kikuchi approximations and belief propagation algorithms”, Technical Report of Mitsubishi Electric Research Laboratories, TR 2001– 16, 2001. [44] J. Feldman, “Decoding error-correcting codes via linear programming”, PhD thesis, MIT, 2003. [45] J. Vogt and A. Finger, “Improving the max-log-MAP Turbo decoder”, IEEE Electronic Letters, 36, pp. 1937– 1939, Nov. 2000. [46] W. Gross and P. Gulak, “Simplified MAP algorithm suitable for implementation of Turbo decoders”, IEEE Electronic Letters, pp. 1577– 1578, 1998. [47] J. Cheng and T. Ottoson, “Linearly approximated log-MAP algorithms for Turbo decoding”, Proc. IEEE Vehicular Technology Conference (VTC Spring 2000), pp. 2252– 2256, May 2000. [48] J. Hagenauer and P. Hoeher, “A Viterbi algorithm with soft-decision outputs and its application”, Proc. GLOBECOM89, pp. 1680– 1686, Nov. 1989. [49] T.K. Blankenship and B. Classon, “Fixed-point performance of low-complexity Turbo decoding algorithms”, Proc. Vehicular Technology Conference (VTC Spring 2001), pp. 1483– 1487, May 2001. [50] P.H.-Y. Wu, “On the complexity of Turbo decoding algorithms”, Proc. Vehicular Technology Conference (VTC Spring 2001), pp. 1439– 1443, May 2001. [51] A. Blanksby and C.J. Howland, “A 690-mW 1-Gb/s, rate-1/2 low-density parity-check code decoder”, IEEE Journal of Solid-State Circuits, 37(3), pp. 404–412, Mar. 2002. [52] D.E. Hocevar, “A reduced complexity decoder architecture via layered decoding of LDPC codes”, Proc. IEEE Workshop on Signal Processing Systems (SiPS 04), pp. 107–112, Austin, USA, Oct. 2004. [53] F. Kienle, T. Brack and N. Wehn, “A synthesizable IP core for DVB-S2 LDPC code decoding”, Proc. 2005 Design, Automation and Test in Europe (DATE 05), Munich, Germany, Mar. 2005. [54] F. Kienle and N. Wehn, “Design methodology for IRA codes”, Proc. 2004 Asia South Pacific Design Automation Conference (ASP-DAC 04), pp. 459–462, Yokohama, Japan, Jan. 2004. [55] E. Boutillon, J. Castura and F.R. Kschischang, “Decoder-first code design”, Proc. 2nd International Symposium on Turbo Codes & Related Topics, pp. 459– 462, Brest, France, Sep. 2000. [56] M.M. Mansour and N.R. Shanbhag, “High-throughput LDPC decoders”, IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 11(6), pp. 976– 996, Dec. 2003. [57] T. Brack, M. Alles, F. Kienle and N. When, “A synthesizable IP core for WIMAX 802.16E LDPC code decoding”, 7th Annual IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Helsinki, Finland, 11– 14 Sep. 2006. [58] M.J. Thul, F. Gilbert, T. Vogt and N. Wehn, “A scalable system architecture for high-throughput Turbo-decoders”, Journal of VLSI Signal Proces.

246

Technologies for the Wireless Future – Volume 3

[59] J. Vogt, K. Koora, A. Finger and G. Fettweis, “Comparison of different Turbo decoder realizations for IMT-2000”, Proc. 1999 Global Telecommunications Conference (Globecom99), 5, pp. 2704– 2708, Rio de Janeiro, Brazil, Dec. 1999. [60] H. Dawid, “Algorithmen und schaltungsarchitekturen zur maximum a posteriori faltungsdecodierung”, PhD thesis, RWTH Aachen, Shaker Verlag, Aachen, Germany, 1996. In German. [61] F. Kienle, T. Lehnigk-Emden, T. Brack, M. Alles, T. Vogt and N. Wehn, “Turbo-codes vs. LDPC codes”, Workshop on VLSI-Architectures for LDPC Decoders, Pisa, Italy, Oct. 2006. [62] J. Hagenauer, “The turbo principle in wireless communications”, Nordic Radio Symposium, Oulu, Finland, Aug. 2004. [63] M. Siala and E. Jaffrot, “Semi-blind maximum a posteriori fast fading channel estimation for multicarrier systems”, GRETSI 17, Vannes, France, 13–17 Sep. 1999. [64] I. Siaud, O. Seller and D. Lacroix, “Turbo channel estimation based on Karhunen-Loeve decomposition for OFDM radio broadcasting systems over ionospheric channels”, International OFDM Workshop (InOFW03), Hamburg, Germany, 10–11 Sep. 2003. [65] I. Siaud and A.M. Ulmer-Moll, “A novel adaptive sub-carrier interleaving application to millimeter-wave WPAN OFDM systems (IST MAGNET project)”, submitted to IEEE Portable 2007 Conference and published in Mar. 2007. [66] I. Siaud and A.M. Ulmer-Moll, “Advanced interleaving algorithms for UWB-OFDM systems”, submitted to International Conference on Communications (ICC07) and published in Jun. 2007. [67] S. Crozier and P. Guinand, “High-performance low-memory interleaver banks for Turbo-codes”, Proc. 54th IEEE Vehicular Technology Conference (VTC 2001 Fall), pp. 2394– 2398, Atlantic City, New Jersey, USA, 7–11 Oct. 2001. [68] S. Crozier, “New high-spread high-distance interleavers for Turbo-codes”, 20th Biennial Symposium on Communications, pp. 3–7, Kingston, Ontario, Canada, 28–31 May 2000. [69] C. Berrou and M. Jézéquel, “Frame-oriented convolutional turbo codes”, Electronic Letters, 32(15), pp. 1362– 1364, Jul. 1996. [70] I. Siaud, A.M. Ulmer-Moll and P. Desmoulin, “A novel HiperMAN OFDM/OFDMA PHY dynamic sub-carrier mapping algorithm based on a dynamic turbo-structure interleaving (version V1)”, ETSI BRAN Contribution, (BRAN44d032r1), France, Feb. 2006. [71] IEEE 802.16e, “Air interface for fixed and mobile broadband wireless access systems”,IEEE P802.16e/D12 Draft Oct. 2005. [72] C. Douillard and C. Berrou, “Turbo codes with rate-m/(mC1) constituent convolutional codes”, IEEE Trans. on Communications, 53(10), Oct. 2005. [73] ETSI EN 301 790, “Interaction channel for satellite distribution systems”, DVB-RCS, 1(2.2), 2000. [74] ETSI EN 301 958, “Interation channel for digital terrestrial television”, DVB-RCT, 1(1.1), 2002. [75] T. Lestable and E. Zimmermann, “LDPC options for next generation wireless systems”, WWRF#14, San-Diego. [76] IEEE P802.11n/D1.04, “Draft amendment to standard for information technology – telecommunications and information exchange between systems – local and metropolitan area networks – Part 11”, WLAN MAC and PHY specifications, Sep. 2006. [77] ETSI EN 302 307, “Digital video broadcasting (DVB); second generation framing structure, channel coding and modulation systems for broadcasting, interactive services, news gathering and other broadband satellite applications”, 2004– 06. [78] P. Urard, E. Yeo, L. Paumier, P. Gerogelin, T. Michel, V. Lebars et al., “A 135 Mbps DVB-S2 compliant codec based on 64800 bits LDPC and BCH codes”, IEEE ISSCC05, Feb. 2005. [79] T. Lestable, E. Zimmermann, M-H. Hammon and S. Stiglmayr, “Block-LDPC codes vs duo-binary Turbo-codes for European next generation wireless systems”, IEEE VTC, Montreal, Canada, Fall 2006. [80] IST-2003-507581 WINNER, “D2.10 final report on identified RI key technologies, system concept and their assessment”, Dec. 2005. [81] H. Hashemi, “The indoor radio propagation channel”, Proc. IEEE, 81(7), pp. 943–968, Jul. 1993. [82] J. Bach Andersen, T.S. Rappaport and S. Yoshida, “Propagation measurements and models for wireless communication channels”, IEEE Communications Magazine, pp. 42–49, Jan. 1995.

New Air Interface Technologies

247

[83] B.H. Fleury and P.E. Leuthold, “Radiowave propagation in mobile communications: an overview of European research”, IEEE Communications Magazine, pp. 70– 81, Feb. 1996. [84] R.B Ertel, P. Cardieri, K.W. Sowerby, T.S. Rappaport and J.H. Reed, “Overview of spatial channel models for antenna array communication systems”, IEEE Personal Communications, Feb. 1998. [85] J. Bach Andersen, “A propagation overview”, 5th International Symposium on Wireless Personal Multimedia Communications (WPMC), 1, 27–30 Oct. 2002. [86] T.K. Sarkar, Z. Ji, K. Kim, A. Medouri and M. Salazar-Palma, “A survey of various propagation models for mobile communication”, IEEE Antennas and Propagation Magazine, 45(3), Jun. 2003. [87] A.F. Molisch, “A generic model for MIMO wireless propagation channels in macro- and microcells”, IEEE Transactions on Signal Processing, 52(1), Jan. 2004. [88] J.P. Kermoal, L. Schumacher, K.I. Pedersen, P.E. Mogensen and F. Fredriksen, “A stochastic MIMO radio channel model with experimental validation”, IEEE Journal on Selected Areas in Communications, 20(6), pp. 1211– 1226, Aug. 2002. [89] K. Yu and B. Ottersten, “Models for MIMO propagation channels, a review”, Wiley Journal on Wireless Communications and Mobile Computing, 2(7), pp. 653– 666, Nov. 2002. [90] M.A. Jensen and J.W. Wallace, “A review on antennas and propagation for MIMO wireless communications”, IEEE Trans. Antennas Propagat., 52(11), pp. 2810– 2824, Nov. 2004. [91] “Spatial channel model for multiple input multiple output (MIMO) simulations”, Technical Report, Release 6, 3GPP TR 25.996 V6.1.0 (2003– 09), http://www.3gpp.org/. ¨ [92] E. Bonek, M. Herdin, W. Weichselberger and H. Ozcelik, “MIMO – study propagation first!”, Proc. IEEE International Symposium on Signal Processing and Information Technology (ISSPIT03), invited paper (special session), Darmstadt, Deutschland, 14–17 Dec. 2003. [93] T. J¨ams¨a, J. Ylitalo, R. Thom¨a and A. Alexiou (eds), “Multi-dimensional radio channel measurement and modeling for wideband wireless access systems”, WWRF/WG4, White Paper, Toronto, Nov. 2004. [94] T. J¨ams¨a, J. Ylitalo, R.S. Thom¨a and A. Alexiou (eds), “Multi-dimensional radio channel measurement and modeling for future mobile and short-range wireless systems”, WWRF/WG4&5 Joint Session, White Paper, Jeju Island, Korea, Mar. 2005. [95] The Book of Visions, WWRF, Dec. 2001, http://www.wireless-world-research.org/. [96] http://www.ist-imetra.org/metra/presentations/METRA_pres_SMi-AAC.pdf. [97] L.M. Correia (ed.), “ Mobile Broadband Multimedia Networks – Techniques, Models and Tools for 4G”, (COST273), Elsevier Science Publishers B.V., 2006. [98] http://www.cost 2100 .org/. [99] https://www.ist-winner.org/. [100] http://www.ftw.at/cost273/. [101] L. Correia (ed.), “ Wireless Flexible Personalised Communications, COST259: European Co-operation in Mobile Radio Research ”, John Wiley & Sons, Ltd, 2001. [102] http://www.lx.it.pt/cost273/. [103] “Guidelines for evaluation of radio transmission technologies for IMT-2000 (Question ITU-R 39/8)”, Recommendation ITU-R M.1225, International Telecommunication Union, 1997. [104] D.S. Baum, J. Salo, G. Del Galdo, M. Milojevic, P. Ky¨osti and J. Hansen, “An interim channel model for beyond-3G systems”, Proc. IEEE Vehicular Technology Conference 2005, Stockholm, May 2005. [105] “IEEE P802.11 Wireless LANs, TGn Channel Models”, IEEE 802.11-03/940r2, 9 Jan. 2004. [106] 3GPP, “E-UTRA MIMO channel model”, text proposal, R1-061002 (agreed). [107] 3GPP, “System level channel models for E-UTRA evaluations”, R1-061594 (agreed). [108] 3GPP, “Update of polarization description in link level channel model”, R1-061371 (noted). [109] 3GPP TSG RAN, “Physical layer aspects for evolved Universal Terrestrial Radio Acces (UTRA) (Release 7)”, 3GPP TR25.814 V7.1.0 (2006-09), Annex A.1.3. [110] V. Erceg et al., “Indoor MIMO WLAN channel models”, IEEE P802.11 Wireless LANs, IEEE 802.11-03/161r2, Sep. 2003. [111] J. Kolu, J-P. Nuutinen, T. J¨ams¨a, J. Ylitalo and P. Ky¨osti, “Playback simulations of measured MIMO radio channels”, TD (04) 110, COST273, Gothenburg, Sweden, Jun. 2004.

248

Technologies for the Wireless Future – Volume 3

[112] http://projects.celtic-initiative.org/WING-TV/. [113] “Proposed additions to recommendation ITU-R BT.1368-5: new channel models for hand held reception and DVB-H receiver C/N-performance in pedestrian and mobile channels”, ITU-R/Document 6E. [114] http://www.celtic-initiative.org/Projects/B21C/abstract.asp. [115] http://www.cost 2100 .org/index.php?page D working-group-2-radio-channel. [116] D.S. Baum et al., “Final report on link level and system level channel models”, WINNER Deliverable D5.4, Oct. 2005, https://www.ist-winner.org/. [117] P. Ky¨osti et al., “WINNER II channel models”, WINNER II Deliverable D1.1.2 V1.1, Nov. 2007, https://www.ist-winner.org/. [118] http://www.tkk.fi/Units/Radio/scm/. [119] M. Narandvzi´c, C. Schneider, R. Thom¨a, T. J¨ams¨a, P. Ky¨osti and X. Zhao, “Comparison of SCM, SCME and WINNER channel models”, Proc. IEEE Vehicular Technology Conference (VTC), Dublin, Ireland, 22–25 Apr. 2007. [120] http://www.wimaxforum.org/home/. [121] http://www.itu.int/net/home/index.aspx. [122] http://www.itu.int/newsroom/press_releases/2007/12.html. [123] http://www.itu.int/newsroom/wrc/2007/itur_web_flash/20071019.html. [124] http://www.itu.int/newsroom/press_releases/2007/36.html. [125] P. Ky¨osti and T. J¨ams¨a, “Complexity comparison of MIMO channel modelling methods”, Proc. IEEE International Symposium on Wireless Communication Systems (ISWCS), Trondheim, Norway, Oct. 2007. [126] J. Meinil¨a et al., “A set of channel and propagation models for early link and system level simulations”, Deliverable D5.1, V1.01, IST-WINNER, Mar. 2004, https://www.ist-winner.org/. [127] M. Landmann, K. Sivasondhivat, J. Takada and R. Thom¨a, “Polarisation behaviour of discrete multipath and diffuse scattering in urban environments at 4.5 GHz”, EURASIP Journal on Wireless Communications and Networking, Special Issue on Space-Time Channel Modeling for Wireless Communications, Article ID 57980, 2007. [128] U. Trautwein, C. Schneider and R. Thom¨a, “Measurement based performance evaluation of advanced MIMO transceiver designs”, EURASIP Journal on Applied Signal Processing, 11, pp. 1712– 1724, 2005. [129] C. Schneider, A. Hong, G. Sommerkorn, M. Milojevic and R. Thom¨a, “Path loss and wideband channel model parameters for WINNER link and system level evaluation”, Third International Symposium on Wireless Communication Systems (ISWCS 2006), Valencia, Spain, 5–8 Sep. 2006. [130] A. Hong, M. Narandzic, C. Schneider and R.S. Thom¨a, “Estimation of the correlation properties of large scale parameters from measurement data”, 18th Annual IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC07), Athens, Greece, 3–7 Sep. 2007. [131] A. Richter. “On the estimation of radio channel parameters: models and algorithms (RIMAX)”, PhD thesis, TU Ilmenau, Germany, 2006. [132] N. Jalden, P. Zetterberg, B. Ottersten, A. Hong and R. Thom¨a, “Correlation properties of large scale fading based on indoor measurements”, IEEE Wireless Communication and Networking Conference, Hong Kong, Mar. 2007. [133] M. Landmann, M. Kaeske, R. Thom¨a, J. Takada and I. Ida, “Measurement based parametric channel modeling considering diffuse scattering and specular components”, ISAP 2007, Niigata, Japan, 20–24 Aug 2007. [134] M. Landmann, W. Kotterman and R.S. Thom¨a, “Estimated angular distributions in channel characterisation”, EUCAP 2007, Edinburgh, Scotland„ 11–16 Nov. 2007. [135] M. Narandˇzi´c, M. K¨aske, C. Schneider, M. Milojevi´c, M. Landmann, G. Sommerkorn et al., “3D-antenna array model for IST-WINNER channel simulations”, IEEE VTC2007, Dublin, Ireland, 23–25 Apr. 2007. [136] R.W. Heath, M. Airy and A.J. Paulraj, “Multiuser diversity for MIMO wireless systems with linear receivers”, Proc. 35th Asilomar Conf. on Signals, Systems and Computers, IEEE Computer Society Press, Paci?c Grove, CA, USA, Nov. 2001. [137] Q.H. Spencer, C.B. Peel, A.L. Swindlehurst and M. Haardt, “An introduction to the multi-user MIMO downlink”, IEEE Communications Magazine, 42(10), pp. 60–67, Oct. 2004.

New Air Interface Technologies

249

[138] S. Vishwanath, N. Jindal and A.J. Goldsmith, “On the capacity of multiple input multiple output broadcast channels”, Proc. IEEE International Conference on Communications (ICC), New York, NY, USA, Apr. 2002. [139] Z. Pan, K.K. Pan and T. Ng, “MIMO antenna system for multi-user multi-stream orthogonal space time division multiplexing”, Proc. IEEE International Conference on Communications, Anchorage, Alaska, May 2003. [140] K.K. Wong, “Adaptive space-division-multiplexing and bit and power allocation in multiuser MIMO ?at fading broadcast channel”, Proc. IEEE 58th Vehicular Technology Conference, Orlando, FL, USA, Oct. 2003. [141] G. Caire and S. Shamai, “On the achievable throughput of a multiantenna gaussian broadcast channel”, IEEE Trans. on Inf. Theory, 49(7), pp. 1691– 1706, Jul. 2003. [142] M.H.M. Costa, “Writing on dirty paper”, IEEE Trans. on Inf. Theory, 29(12), pp. 439– 441, May 1983. [143] C. Peel, B. Hochwald and L. Swindlehurst, “A vector perturbation technique for near-capacity multi-antenna multi-user communication”, Proc. 41st Allerton Conference on Communication, Control and Computing, Oct. 2003. [144] C. Windpassinger, R. F.H. Fischer and J. B. Huber, “Lattice reduction-aided broadcast precoding”, Proc. 5th International ITG Conference on Source and Channel Coding (SCC), Erlangen, Germany, pp. 403–408, Jan. 2004. [145] G. Cinis and J. Ciofi, “A multi-user precoding scheme achieving crosstalk cancellation with application to DSL systems”, Proc. Asilomar Conf. on Signals, Systems and Computers, 2, pp. 1627– 1637, Nov. 2000. [146] Q. Spencer and M. Haardt, “Capacity and downlink transmission algorithms for a multi-user MIMO channel”, Proc. 36th Asilomar Conf. on Signals, Systems and Computers„ IEEE Computer Society Press, Pacific Grove, CA, USA, Nov. 2002. [147] Q.H. Spencer, A.L. Swindlehurst and M. Haardt, “Zeroforcing methods for downlink spatial multiplexing in Multiuser MIMO channels”, IEEE Transactions on Signal Processing, 52(2), pp. 461–471, Feb. 2004. [148] L.U. Choi and R.D. Murch, “A transmit preprocessing technique for multiuser MIMO systems using a decomposition approach”, IEEE Transactions on Wireless Communications, 3(1), pp. 20–24, Jan. 2004. [149] V. Stankovic and M. Haardt, “Multi-user MIMO downlink precoding for users with multiple antennas”, Proc. 12th Meeting of the Wireless World Research Forum (WWRF), Toronto, ON, Canada, Nov. 2004. [150] T. Yoo and A. Goldsmith, “On the optimality of multi antenna broadcast scheduling using zero-forcing beamforming”, IEEE JSAC, 24(3), pp. 528– 541, Mar. 2006. [151] P.W. Wolniansky, G. J Foschini, G.D. Golden and R.A. Valenzuela, “V-BLAST: an architecture for realizing very high data rates over the rich-scattering wireless channel”, Proc. ISSSE 98, Sep. 1998. [152] M. Joham, J. Brehmer and W. Utschick, “MMSE approaches to multiuser spatio-temporal Tomlinson-Harashima precoding”, Proc. 5th International ITG Conference on Source and Channel Coding (ITG SCC04), pp. 387– 394, Jan. 2004. [153] V. Stankovic and M. Haardt, “Multi-user MIMO downlink precoding for users with multiple antennas”, Proc. of the 12th Meeting of the Wireless World Research Forum (WWRF), Toronto, ON, Canada, Nov. 2004. [154] V. Stankovic and M. Haardt, “Successive optimization Tomlinson-Harashima precoding (SO THP) for multi-user MIMO systems”, Proc. IEEE Int. Conf. Acoust., Speech and Signal Processing (ICASSP), Philadelphia, PA, USA, Mar. 2005. [155] D. Schmidt, M. Joham and W. Utschick, “Minimum mean square error vector precoding”, Proc. PIMRC05, Sep. 2005. [156] P. Tejera, W. Utschick, G. Bauch and J. Nossek, “A novel decomposition technique for multi user MIMO”, Proc. 6th International ITG Conference on Source and Channel Coding (ITG SCC05), Apr. 2005. [157] “IEEE P802.11 wireless LANs, TGn channel models”, Tech. Rep. IEEE 802.11-03/940r2, IEEE, Jan. 2004. [158] G. Del Galdo, M. Haardt and C. Schneider, “Geometry-based channel modeling in MIMO scenarios in comparison with channel sounder measurements”, Advances in Radio Science – Kleinheubacher Berichte, pp. 117– 126, 2004. [159] M. Fuchs, G. Del Galdo and M. Haardt, “A novel tree-based scheduling algorithm for the downlink of multi-user MIMO systems with ZF beamforming”, Proc. IEEE Int. Conf. Acoust., Speech and Signal Processing (ICASSP), Philadelphia, PA, USA, Mar. 2005. [160] N. Jindal, S. Jafar, S. Vishwanath and A.J. Goldsmith, “Sum power iterative water-filling for multi-antenna Gaussian broadcast channels”, Proc. Asilomar Conf. Signals, Systems, Computers, pp. 3–6, Paci?c Grove, CA, USA, Nov. 2002.

250

Technologies for the Wireless Future – Volume 3

[161] V. Stankovic and M. Haardt, “Generalized design of multi-user MIMO precoding matrices”, IEEE Trans. on Wireless Communications, 7, pp. 953– 961, March 2008. [162] M. Fuchs, G. Del Galdo and M. Haardt, “Low complexity space-time-frequency scheduling for MIMO systems with SDMA”, IEEE Transactions on Vehicular Technology, 56, pp. 2775– 2784, Sept. 2007.

7 Short-range Wireless Communications Edited by Prof. Rolf Kraemer (IHP Germany) and Marcos Katz (VTT Finland)

7.1 Introduction Short-range wireless communications is one of the most rapidly developing areas of wireless communication. The first wave of innovation in short-range wireless communications was the introduction of WLANs, followed by Bluetooth. Both were reflected in different IEEE standardization groups like IEEE802.11 for WLAN and IEEE802.15 for personal area networks (PANs). Both groups are continuously working on ever-new versions for WLANs and WPANs. The newest standard for WLAN, for instance, is IEEE802.11n, which increases the communication speed up to 600 Mb/s. The IEEE802.15.3c is working on a new standard for gigabit communications. Several other activities within IEEE are ongoing to cover all aspects of wireless communications, especially short-range communications. The vision of WWRF with respect to short-range communications states that we will face 7 trillion wireless devices serving 7 billion human beings in 2017. This would be approximately 1000 devices per human being. Most of these devices will be sensor nodes, RFID tags, etc. which service users in a nonvisible background manner. Several devices will directly serve users in the immediate vicinity for their personal needs. This growing group of PAN devices will take care of the users’ comfort in using services, handling devices, etc. Other devices will directly take care of communication and access to remote services. There will be an invisible hierarchy of wireless communication devices that adapt to the need of users in accessing services, using devices and communicating with other users. So, short-range communication will play an important role in the future of wireless communications. There is no clear definition of what ‘short range’ really means. As a working definition we consider short-range communications as technologies that are used to build service access in local areas. We do not address/cover the topic of full coverage, but do the topics Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

252

Technologies for the Wireless Future – Volume 3

of access speed, local performance and comfort, energy efficiency, etc. As you can see, even this definition will have several overlaps with cellular communication. In WG5 of the WWRF we concentrate on these short-range aspects. Within the following chapter you will find information about the completed work of two areas of short-range communications, ‘Integrative and Cooperative Aspects of Short-range Communications: Technologies, Designing Rules and Trends’ (Section 7.2) by Frank Fitzek from Aalborg University and Marcos Katz from VTT – the Technical Research Centre of Finland, and ‘Ultra Wideband Radio over Optical Fibre’ (Section 7.3) by Moshe Ran, Yossef Ben Ezra, Motti Haridim and Boris. I. Lembrikov from the Holon Institute of Technology (HIT). Moreover you will find two papers that show work in progress; ‘High Data Rate Wireless Communications in the Unlicensed 60 GHz Band’ (Section 7.4.1) by Eckhard Grass, Frank Herzel, Maxim Piz, Klaus Schmalz, Yaoming Sun, Srdjan Glisic, Milos Krstic, Klaus Tittelbach-Helmrich, Marcus Ehrig, Wolfgang Winkler, Christoph Scheytt and Rolf Kraemer from IHP GmbH, Frankfurt (Oder), Germany, and ‘Ultra Wideband Communication’ (Section 7.4.2), edited by Thomas Kaiser and Emil Dimitrov. The last two contributions give an overview of the work being conducted by members of WG5. We plan to publish an additional book on aspects of short-range communications in the second half of 2008 with an in-depth view of all mentioned aspects. Ultra high-speed communication in the short range is a prominent topic. Two basic possible solutions are being presented, ultra wideband solutions like DS-UWB or MB-UWB, and ultra high-frequency solutions, e.g. 60 GHz gigabit communications. Frank Fitzek and Marcos Katz, having discussed several short-range communication system aspects, address the overall integrative aspects in the interworking between short-range communication and cellular networks. Based on four different topologies, peer-to-peer networks, hybrid mobile device and sensor networks, wireless grids, and cellular controlled peer-to-peer networking, the different aspects of social networking will be discussed. Interesting aspects of the cooperation between the different network types are outlined. Furthermore, the contribution describes the results of a first short-range connectivity measurement campaign. The discussion of the overall aspects of network interworking leads to conclusions for the design of future mobile networks, including consideration for 4G networks. Moshe Ran et al. presents in his contribution aspects of coverage extension for UWB-based systems. Since UWB systems will only work in the short range due to regulation restrictions, considerations, scenarios, applications, and methods that address the distribution of the UWB signals within a larger area will be discussed. Several innovative applications will be outlined using radio-over-fibre technologies. The basic radio-over-fibre technologies will be discussed in an extensive state-of-the-art analysis. Several research aspects for electro/optical and optical/electrical signal conversion are being addressed and investigated. The UROOF (UWB radio-over-optical fibre) ideas can also be used for extending the range of 60 GHz systems. Therefore it is presented as a generic technology to extend the range of broadband and ultra high frequency systems. To understand the consequences of the UROOF extensions, an in-depth analysis of the mixed wireless/optical channel is given. Eckhard Grass et al. addresses in his contribution the aspect of 60 GHz communication being discussed in the IEEE802.15.3c. He describes the spectrum aspects, possible applications and possible realizations that could lead to solution in the near future. Data rates of up to 10 Gbps will be addressed. Moreover, the contribution describes the current state of IEEE802.15.3c, its roadmap to complete the standard, different choices for different application areas, as well as the postulated markets. While being the view of a single group, the

Short-range Wireless Communications

253

contribution recognizes the different aspects that can lead to decisions. Marked as ‘work in progress’, this contribution will not cover all aspects in detail. The coming full paper will have authors from all important working groups on 60 GHz communications and will have more and different views outlined. Thomas Kaiser and Emil Dimitrov present in their contribution a short version of the overall work of UWB communications. They describe the channel model for UWB communications, outline the spectrum aspects, and discuss different types of transceiver for pulse-based communications and multiband communications. An innovative aspect of UWB communications is its inherent capability of localization. This aspect has been addressed by IEEE802.15.4a. Using the UWB signal it is possible to detect the first arriving one out of a variety of multipath signals. Accurate distance estimation becomes possible. Furthermore, several antenna aspects are discussed and possible MIMO extensions will be described. Finally, approaches towards low-power hardware are outlined. Being marked as ‘work in progress’, several aspects can only be addressed in a short overview mode. The full work will be presented in the WG5 book on aspects of short-range communications. Our chapter on short-range communication should give an overview of several hot topics currently under discussion in the communication community. We are convinced that future mobile networks will consist of a multitude of individual networks using different technologies. The aspects of their individual advantages for particular applications, as well as the cooperation and integration aspects, will lead to an ever-increasing capability of communication systems for the service of human beings. Nobody is able to operate thousands of mobile devices for their own purpose; all devices together will be automatically controlled by each other, by some overall policies and by some interaction with operators and users. WG5 will continue its work with some new interesting aspects of short-range communication. First, we will address the challenges of ultra high-speed communication links of up to 100 Gbps. We will investigate what roadblocks have to be removed to succeed in coming close to that goal. A second new topic will be machine-to-machine communications. We will address especially aspects of car-to-car communication as a prominent example of this class. Here the reliability aspects are of utmost importance. Last but not least, we will address the challenges of visible light communications. A white paper on this topic is already in preparation and will be ready in June 2008.

7.2 Integrative and Cooperative Aspects of Short-range Communications: Technologies, Designing Rules and Trends This section deals with integrative and cooperative aspects of short-range communication systems. It also stresses the need for high-performance short-range technology for future wireless communication systems. Besides the existing cellular communication systems, short-range communication will play a major role in the future, enabling pure peer-to-peer networks as well as composites combining peer-to-peer and cellular networks. Peer-to-peer networks are used to form social networks and sensor networks. Social networks, where people can exchange digital contents or interests, will spice up the wireless world and generate new kinds of service. In this section, sensor networks are approached from the standpoint of sensors supporting mobile devices with context information for the customer’s environment. The combination of cellular and peer-to-peer networks is also an interesting field as the cellular networks contribute with their richness of services and the ability to ensure secure peer-to-peer

254

Technologies for the Wireless Future – Volume 3

connections. Once a peer-to-peer network is established, the peers can act as a wireless grid supporting each other in a cooperative way. As the peers are owned by different users, mutual cooperation needs to exist, bringing instantaneous benefits to all the collaborating partners. The existing short-range technology is only partially suited to supporting the envisioned networks. Besides basic connectivity and increasing data rate, current technologies fail in many aspects, including high power consumption, lack of fast device discovery, and security. This section motivates the need for high-performance short-range technology and lays out the main needs and designing rules for the next decade. 7.2.1 Motivating the Need for Short-range Communications Current communication systems mainly focus on centralized or cellular networks. Short-range communication is in general seen as less important or secondary technology. Since the inception of Bluetooth, which was introduced as a cable replacement and IRDA, this area has not undergone major progress. Currently, technologies such as WiBree and Bluetooth v2 or v3 are being discussed. The main development direction is increased data rate or lower energy consumption. It seems that short-range technology is less appealing as the market potential for those technologies is not quite understood. But, as we will advocate throughout our manuscript, short-range communication could bring new business models, working either in a standalone fashion or cooperating together with the cellular network. In a nutshell, short range is more than just exchanging digital business cards. 7.2.2 Introduction to Short-range Communications Short-range communication can be defined as any information transfer process taking place with a reach from some millimetres up to few hundred metres. Research and development activities around such small-scale communications are today booming and there are more than enough reasons to believe that in the future short-range communications will become one of the dominant wireless communications approaches. WWRF predicts that by the year 2017 there will be on average some 1000 wireless devices per world inhabitant. The vast majority of this impressive figure, 7 trillion communication links, will certainly correspond to short-range wireless connections. Short-range wireless communication involves a very diverse array of air interface technologies, network architectures and standards. An overview of short-range communications from the standpoint of the typical range is presented in Figure 7.1 Short-range communications encompass a large variety of different wireless systems with a great diversity of requirements. Figure 7.2 illustrates the wide scope of short-range communications by classifying it according to the most common air interface technologies and network architectures, as well as supported mobility and data rates. 7.2.3 Peer-to-peer Networks and Social Networks Peer-to-peer technology has proven its strength already in the wired domain. Bringing it to the wireless world is a major act that could cause concern for network providers, given that the first impression is that that they could lose their ground. However, we encourage network operators to continue to read this manuscript, as we discuss the potentials of short-range from their business perspective.

Short-range Wireless Communications

255

Wireless Sensor Networks

WSN

Car-to-Car Communications

C2C

Wireless Local Area Networks

WLAN

Wireless Personal Area Networks

WPAN

Wireless Body Area Networks

WBAN RFID

(passive)

Radio Frequency Identification

(active)

Near Field Communications

NFC 10−2

10−1

100

101

102

103 Range (m)

Figure 7.1 A classification of short-range communications according to the typical supported range Conventional (narrowband) radio

Air interface

Ultrawide Band (UWB) radio mm-wave communications

Optical wireless communications Short-range wireless communications

Network topology

Centralized (cellular, infrastructure) Distributed (ad hoc, mesh decentralized, infrastructureless)

Mobility

Fixed Nomadic Mobile Low (bps – 100Kbps)

Data rate Moderate (100Kbps – 10Mbps) High (10Mbps – 10Gbps)

Figure 7.2 A general classification of short-range communications

As the world continues searching for the always elusive killer application – without any clear success in the last decade – there is one observation telling us what people are really interested in: other people! This is the reason why voice services and chat tools are so popular. Most of these services connect people over long distances with each other. Social networks are now focusing on short-range communication among people in the short-range communication coverage of their mobile phone. The very first mobile applications were some flirt-oriented services, but nowadays the applications are going in the MySpace direction, such as aka-aki

256

Technologies for the Wireless Future – Volume 3

[1]. Once people start getting connected over the short-range they will also start to exchange digital content of every kind. Such application was realized in the SMARTEX project at Aalborg University in 2004 [2]. In some communities, digital content is becoming more important than real physical items. It is used to enhance the properties of mobile terminals (e.g., ring tones), to form collections (e.g., baseball cards), or for multiplayer role games (e.g., virtual tools). SMARTEX is a novel approach to creating new services and it is based on a communication architecture referred to as cellular controlled short-range communications. The main idea behind SMARTEX is to allow customers to exchange digital content between terminals using the available short-range communication links under the control of the cellular network. Figure 7.3 shows a screenshot of the SMARTEX application [3]. The short-range communication is based on Bluetooth, over which the digital content is exchanged. The cellular link is needed to control the short-range communication (security and digital rights management). In SMARTEX the problem of digital ownership [3] is introduced for this kind of communication architecture.

Figure 7.3 Screenshot on a NOKIA 6600 of the SMARTEX application developed at Aalborg University [3]

7.2.4 Hybrid Mobile Device and Sensor Networks Pervasive computing with mobile devices and wireless sensor networks has been identified as a future area of research. This section motivates the convergence of those two technologies as this will introduce new services to the customer. Wireless sensors have the possibility to sense the surrounding world and communicate wirelessly among each other or with the mobile device within short-range. The mobile device offers the well-known cellular services to the symbiotic partnership. Moreover, mobile devices provide their embedded user interface and display, allowing users to control and monitor the sensor network. It is expected that in the future more and more wireless sensors will surround us to give us more information about our environment. The main idea is to move high-priced functionalities from vehicles to the mobile phone and spice them up. One example is the presented park sensor concept [4]. A car is equipped with ultrasonic distance sensors which are connected wirelessly with a mobile phone. The mobile phone is placed within the car, showing the distances to the nearest objects. Besides the display functionality, the mobile phone can support the driver with audio information (currently we use the text2speech functionalities of S60 phones). A first demonstration can be seen in Figure 7.4. As the service is bundled to the phone, it makes no sense to steal anything from the car.

Short-range Wireless Communications

257

Figure 7.4 Details of the wireless sensor board (collaboration of TU Berlin and Aalborg University) (left) and some application scenarios of the parking sensor application (top right, bottom right)

7.2.5 Wireless Grids In the process of developing the new wireless/mobile generation, referred to as the fourth generation or 4G, some of the key problems to solve include enhancing spectral and power/energy efficiency, keeping the complexity of the end system reasonably low, and developing appealing services for the users. In this section we will introduce a new wirelessly scalable architecture introducing cooperative principles for cellular networks and aimed at solving the mentioned challenges. The architecture is basically a composite or hybrid approach combining centralized and distributed architectures, typical of cellular and ad hoc networks respectively. The interested reader is referred to [5] for further details on this composite architecture. In order to offer more services and capabilities to users, mobile devices are crammed with several functionalities. Such multiplicity of capabilities introduces two major problems: first, the mobile device becomes very expensive, and second, the battery will be drained very quickly, reducing the operational time of the device. As the monolithic (or single-piece) architecture will exhibit many of the aforementioned drawbacks, the very first idea is to enable cooperation among several closely located mobile devices, with their different functionalities forming a heterogeneous wireless grid. This is a clear departure from today’s use, where a given air interface (and corresponding architecture) is used in a given scenario or for a particular service. Dynamic interaction between and across air interfaces is not possible in such a conventional approach. Clearly, cooperation will enable resource sharing or augmentation among the collaborating terminals, with the cellular network being basically the service entry point and the short-range links the gluing element allowing such a wireless grid to have a wirelessly scalable architecture. As given in Figure 7.5, a mobile device has many different capabilities and functionalities. We have grouped those of a mobile (smart) phone into three classes, namely user interfaces, communication interfaces and built-in resources. The user interfaces typically comprise the speaker, microphone, camera, display, built-in sensors and keyboard capabilities.

258

Technologies for the Wireless Future – Volume 3

Figure 7.5 A mobile device broken up into several capabilities, grouped into user interface, communication interface and built-in resources

The built-in terminal resources include the battery, central processing unit and data storage. From our point of view we highlight particularly the communication interfaces, which typically include cellular and short-range capabilities. Instead of having one mobile device hosting all functionalities with the best possible quality, the concept presented here uses and shares cleverly the capabilities of several, in principle different, mobile devices. Indeed, in such a heterogeneous scenario each terminal could feature a particular specialization, from a simple mobile phone for voice calls to advanced terminals with music and imaging devices. In the proposed wirelessly scalable architecture, a wireless grid of terminals is formed where the connecting links among cooperating terminals are implemented by the short-range air interfaces. In general terms, such a scalable concept is not totally new, as Bill Joy, the visionary figure behind Jini Technology and cofounder of Sun Microsystems argued in 1999 for cooperative software: Jini technology is a dislocation in the context of cooperation of subspecies of devices. Once you have lots of different kinds of devices combining in different ways, you can not do monolithic software anymore. Each of these devices has a certain set of functions, and if we have to assume when we build them what they are going to be used for, it is not very flexible.

Inspired by the architect Frank Lloyd Wright’s theory of architecture, Bill Joy rephrased the thought ‘A brick should be a brick. A wall should be a wall. A column should be a column’ for mobile devices as (see Figure 7.5): A cell phone has a microphone. It has a numeric keypad. It has a display. It has a transmitter and receiver. That is what it is. It has a battery with a certain lifetime. It has a processor and some memory. If it is a European model it has a stored value card. It is a collection of things. It is some collection. The fact that it is all one bundle, well, that is pretty artificial.

Short-range Wireless Communications

259

Figure 7.6 Example of cooperative devices of the same user for a mobile phone and a tablet PC

In our context, the main idea is to identify core functionalities of the cooperating mobile devices and reassemble them according to the service to be provided, without necessarily holding the full set of functionalities in one mobile device. An example is given in Figure 7.6. A user may have two different terminals, where the devices have a subset of different capabilities (e.g. cellular communication facilities for the mobile phone) and a subset of the same functionalities but with different realization levels (e.g. the display of the tablet PC has a bigger size and higher resolution). In order to support a wirelessly scalable architecture, the devices need to locally communicate with each other, which is done by wireless communication links, or more precisely the short-range communication link. In the future the short-range link will become far more important than it is today. The number of applications is numerous, for example using remote devices for storage, camera, microphones, etc. This kind of cooperation is envisioned for terminals owned by the same customer and therefore is referred to as altruistic cooperation, inspired by the cooperative nature found in bees or ants. In [9] such a concept is brought to the next level with respect to two main properties. The first is that cooperation is also envisioned for the wireless terminals of different users (known to each other or not). As each customer is basically following egoistic interests and customers may not know about the neighbouring mobile device capabilities, this kind of heterogeneous cooperation seems to be more difficult to establish. The second is that capabilities are not simply accumulated but negotiated beforehand among the devices, to then dynamically adapt/change them according to the upcoming needs. One example of such an adaptation is the air interface that needs to support cellular as well as short-range communication. We have shown that both communication parts can be realized within the same spectrum using a software-defined radio (SDR) approach, where the ratio between cellular and short-range depends on the actual needs for cooperation [6,7].

260

Technologies for the Wireless Future – Volume 3

Base station (access point)

Cellular link (C)

Cellular link (C)

Radio resources time space

Radio resources

frequency

time Short-range link (SR)

power/ energy

Built-in resources

mass CPU/DSP storage

space

frequency power/ energy

Built-in resources

battery

Cooperative cluster or wireless grid

mass CPU/DSP storage

battery

Figure 7.7 Cellular-controlled peer-to-peer network

7.2.6 Cellular Controlled Peer-to-peer Networking Our novel wirelessly-scalable architecture is an extension of a cellular network, where the mobile devices communicate with each other and with the serving base station. This basic architecture is shown in Figure 7.7. Here the main functionalities to group and share are the built-in resources and the communication interfaces. The wireless grid can give rise to new virtual functionalities with enhanced performance, such as a high virtual data rate and a virtual battery, a virtual processing unit and a virtual storage. The virtual functionalities can in principle be used in a more efficient way than the standalone functionalities. In [8] the idea was first proposed of grouping close mobile devices, letting them communicate among each other over the short-range communication link while being simultaneously connected with the base station. By grouping the devices, we form a new cooperative cluster, where each device can contribute with its own capabilities, as given in Figure 7.7. These capabilities may be either homogeneous or heterogeneous in nature. It can be shown that the newly formed cluster has much better communication performance in terms of bandwidth usage, end system complexity and energy consumption. This is based on the fact that the whole cooperative cluster as such is more than the simple sum of its capabilities [9]. Further examples illustrating the benefits of such a cooperative setup can be found in [10–13]. In Figure 7.8 a comparison of monolithic and cooperative communication systems with respect to a requested service and the related costs is given. For the monolithic case the cellular communication link has to serve with a given data rate to support the requested service. This may result in a very complex architecture design. In the cooperative case the same service can be supported, as the same data rate is obtained through a less complex cellular air interface and short-range links. As the short-range link supports higher data rates and has better energy efficiency, the overall costs are less than with the monolithic one. On the other hand there is a need for a cooperative peer device.

Short-range Wireless Communications

261

Figure 7.8 Comparison of monolithic and cooperative communication systems with respect to requested services and the related costs

What makes the proposed approach different from many other cooperative techniques is that each interacting entity or device is in fact gaining at the same time as it cooperates. This is different from many cooperative approaches such as multihop, relaying, or others. Therefore, this kind of cooperation is referred to as egoistic cooperation, inspired by the cooperation strategies found in nature, for instance by monkeys and vampire bats. In most of our previous work we claimed that the benefits of cooperation help all players in the mobile device business, including terminal manufactures, network operators, service providers and the customer. Throughout this section we will see such benefits and map them to the entity receiving them. Following our discussions on the cooperative architecture, we will focus now on service discovery for such architecture. Service discovery has basically two stages. In the first stage we would like to identify possible situations in which mobile devices are available for cooperation. We propose some ideas on how to find such situations and how to report on them. The next stage concentrates on how to encourage cooperation for certain services. Therefore we highlight some cooperative services, such as cooperative video broadcasting and cooperative web browsing, and afterwards give some ideas on how to promote them within the own cooperative cluster. Once the cooperative cluster is established, different services can be provided. Obviously, multicast or broadcast services are well suited to cooperative clusters. For such services the cooperative cluster can accumulate their cellular bandwidth to open up a virtual big pipe. Each terminal receives only a part of the content, which it makes available over the short-range communication links to all cooperative partners. Also on the short-range communication link, every collaborating terminal receives the disjoint information, thus the missing parts. Certainly, such a scenario is ideal for file downloading or video reception. As an example of video services, the benefit of cooperation for DVB-H services was investigated. The main reasons why DVB-H was introduced even though DVB-T was already available are the video format and the battery consumption. In DVB-H the air interface is only active a fraction of the receiving time to receive high data rate bursts. Battery savings are available, as the air

262

Technologies for the Wireless Future – Volume 3

interface is switched off from time to time. The main idea in [5] is to virtually increase the times the RF/BB is switched off. This is done as one mobile device of the group receives the broadcast data and forwards it to its local neighbourhood, which has deactivated the cellular air interface and listens only on the short-range link. As the short-range link is much more energy efficient than the cellular link, energy can be saved for all remote devices. The forwarding device invests slightly more energy as it has to activate two interfaces. Therefore, for fairness reasons, the role of the forwarder is changed within the group from time to time. In this scenario, each cooperating entity is more than welcome to the video service. Here the benefit for the customer is clearly visible. An extension to this scenario is the use of multiple description coded (MDC) video. Here the network provider and the service provider agree to transmit a broadcast video stream in small substreams, where each substream is independently decodable. The more substreams are available at the mobile device, the better the video quality becomes. To improve the video quality, the customer needs a high-data rate connection. One possible way is to achieve a high-class mobile device. This device will cost considerably more than a basic device and will also consume more energy as the realization of high data rate is quite complex over the cellular link. Another option is to exploit cooperation. Here, even basic devices will receive one or a small subset of the total substreams and they will exchange them cooperatively within the group. The customers will gain in power consumption and it will even result in a smaller price, if the network and service providers have agreed on such a cooperative service. The reason why the network provider and the service provider should do so, is the fact that they increase their market share. High-class services can only be used in a standalone fashion by high-class mobile devices. The number of those devices may be small compared to the basic, omnipresent devices. To sell those high-class services even to the largest group of terminals, the network and service provider should agree that the grouping will pay off in some way. Besides multicast services, unicast services can also be exploited in a cooperative manner. The DVB-H scenario can also be extended for the unicast IP services. As the unicast IP services are conveyed in a broadcast fashion with the DVB-H cell, the concepts derived above for video services are still valid. Among other possibilities, we consider now the use of cooperative web surfing. We take the example of two mobile devices with a given cellular data rate, and we assume that the data rate over the short-range link is much larger than that on the cellular one. For illustrative purposes, we assume a GPRS connection for the cellular and a Bluetooth connection for the short-range. The communication architecture is now assumed to be as given in Figure 7.7. If both devices start to surf the web, their traffic can be modeled as ON OFF traffic. If they do not cooperate, their cellular link will be idle for most of the time and active only when a link is clicked or inserted directly. The time they have to wait until the full page can be displayed depends on the content size and the available data rate. In the cooperation case the traffic can be split in two parts. One part is the direct cellular part and the second part is relayed by the neighbouring mobile device. Note that each web page is already divided in many TCP substreams and that therefore this scenario can be easily applied. In the cooperative scenario with two mobile devices, the data rate is nearly doubled and therefore the waiting time should be broken into half. It may happen that both stations click request the medium at the same time. In such a case the performance is (as bad) as in the non-cooperative scenario. The probability of double booking reduces with a larger number of cooperating mobile devices. How the system deals with unequal cooperation awards (if one mobile device uses more resources than the other) depends on the network costs and the relationship of the terminals to each other.

Short-range Wireless Communications

263

Figure 7.9 Possible cooperative application on the mobile device

This section focuses on techniques for service announcement within a cooperative cluster. We believe that the announcement should be made by the entities themselves instead of the cellular base station. For sure, the base station may enhance the service announcement, but the main task should be done by the mobile devices themselves. On each mobile device there should be an application running showing which cooperative services are currently active. In Figure 7.9 such an application is depicted. As seen, the user gets an overview of the ongoing cooperative services. Furthermore, the application has to inform the user about the possible costs in case he or she joins the group. Costs are composed out of the network and the service provider costs. The information needs to be given by the network and/or service provider. Such an application benefits the network operator and the service provider as not only are services that are initialized by the customer themself requested, but already ongoing services attract other users to join. As illustrated in Figure 7.10, the market potential will increase due to new features announcing ongoing services. Note that this kind of advertisement differs from those that could be launched by the network operator. Here the surrounding customers are using the service and indicate that it might be interesting. The customer may spontaneously join the service. This will lead to a significant increase of the market potential. We stress that cooperation increases the market potential due to two main facts. The first is that

Figure 7.10 Increased market potential for cooperative services

264

Technologies for the Wireless Future – Volume 3

the cooperative clusters advertise their services and may encourage other customers to join spontaneously. A good service or a service that is used by others intensively may also be frequented by customers that would not have considered such a service before. The second is that more terminals can consume even high-class services. While a high-class service can be consumed by high-class devices in a standalone fashion, the same service can be used by commercial mobile devices exploiting cooperation.

7.2.7 Scenarios for Social Networking We note that in the three scenarios in which people spend most of their time: home, office and public places, the basic conditions for establishing a cooperative cluster (wireless grid) are met. In these environments we are very likely to find peer users with wireless devices. As the relationship between the users is in principle different in different wireless scenarios, the techniques needed to form clusters, as well as the employed cooperative strategies, could be different in each scenario. The scenarios and associated wireless grids are summarized in Figure 7.11, where the main players and types of cooperative principle are also included. Even though there are several commonalities between the scenarios, the nature of these interacting entities as well as the type of the applied cooperative principles could be quite different. In the home environment, the cooperative cluster can be formed by entities belonging to the personal network, for instance wireless devices owned by the user, home appliances, etc. In addition, wireless devices of other family members sharing the dwelling can be assumed to be natural interacting entities of the wireless grid. Clearly, altruism is the dominant principle enabling cooperation of devices owned by a given user and his or her close relatives. In the office environment also, the wireless devices of coworkers can participate in the cooperative cluster, either spontaneously or encouraged (or even required) by the employer. In public places, cooperation can take place among wireless devices owned by users who are unknown to each other. In such a case cooperation can emerge spontaneously but in general it has to be encouraged by some incentives.

who

public wireless grid

personal wireless grid

private/professional wireless grid

myself

friends, family, colleagues

everybody

where

everywhere

home, office

airports, hotels, restaurants, etc.

cooperation type

altruistic

(non-)/altruistic

non-altruistic

Figure 7.11 Wireless grids in different everyday scenarios

Short-range Wireless Communications

265

BS

BS

Core network

composite access network

composite access network Cellular access

Cellular access

2

1 distributed local access

Pervasive Wireless Social Netwoks -Local social interaction one-to-one, one-to-many, many-to-one, many-to-many - Communications basically through the wireless grid, cellular network has a supporting role - Example (one-to-one) user 1 connected to user 2

- Wide social interaction one-to-one, one-to-many, many-to-one, many-to-many - Communications across the wireless grid, cellular network has a key role

3 distributed local access

- Example (one-to-one) user 1 connected to user 3

Figure 7.12 A pervasive wireless social network exploiting the discussed composite network architecture

Network operators have a clear role here, by creating new services exploiting cooperation between users and also developing exciting incentives to encourage users to cooperate. We highlight here that the considered composite (cellular-short-range) architecture works not only as a framework for providing pervasive wireless connectivity but also as a natural framework for establishing wireless social networks. This is an important point that needs to be capitalized by future development. Figure 7.12 illustrates the described approach, showing a) local social interaction and b) wide social interaction, via the associated wireless grid in the first case, and the corresponding wireless grids and centralized cellular networks in the second case. 7.2.8 Short-range Connectivity Measurement Campaign In this section we summarize the results of a measuring campaign carried out to determine some basic statistics of the mobile devices connected over short-range links. This campaign was motivated by the fact that knowing such cluster statistics will give us valuable hints for designing cooperative strategies. The cluster statistics of interest were mean number of connected devices, type of devices and mean time of connection. Also, the scanning time required to find out how many devices surround a given reference terminal was evaluated. Several wireless scenarios need to be characterized in detail, including public places such as airports, downtown, conference centres, shopping malls, restaurants, bars, etc. In addition, other interesting environments where cooperative clusters can be established include for instance home

266

Technologies for the Wireless Future – Volume 3

Figure 7.13 Screenshots of the Python S60 application

and office places and schools. In this contribution we present only results for a small-size airport (Aalborg Airport, Denmark). 7.2.8.1 Measurement Test-Bed The measurement test-bed consists of one single mobile device (e.g., a Nokia N70 terminal) running a Python S60 script. Figure 7.13 depicts a screenshot of the application on the terminal. The Python script constantly collects information about the BT devices found in its coverage. For each device found, information about address, name, major service class (MaSCl), major device class (MaDCl) and minor device class (MiDCl) are obtained. There are codes used to check the type of device found. The GSM position is also stored by obtaining the mobile country code, mobile network code, location area code and cell ID. After the position, the time information is taken. The time information contains date, day and time (hh:mm:ss). The scanning for neighbouring mobile devices is repeated every 30 seconds until the application is stopped. The interval is set to 30 seconds due to limitations of the cache refreshing process and in order to have good information gathering about the environment. By setting a smaller interval, the cache has no time to be refreshed, resulting in wrongly discovered devices. After each scanning, the information obtained is stored in an XML file, which represents a useful way to collect and store data. A different file is created each time a new measurement campaign is started. The analysis of the information stored in the XML file is done with the Python 2.5 software installed on a computer. All the plots are realized in Python, by using Matplotlib [14]. 7.2.8.2 Methodology The measurements taken, as explained in the previous section, have been used to characterize wireless short-range grids formed over BT connections. This will allow a better understanding of the nature of the cooperative cluster and how to exploit it in the most efficient fashion. Among the total devices, the device type is checked to include only mobile devices in our evaluation. The following investigation values were of special interest. Number of Devices and Time of Connection The number of devices refers to the total number of devices found, and we differentiate between mobile devices that were found in the last scan and new mobile devices.

Short-range Wireless Communications

267

The numbers of overall and newly-found mobile devices are represented in two bars of different colours. If the mobile device within a cluster changes frequently, a considerably large amount of energy will be spent for signaling. Therefore, darker bars can be thought of as an indicator of the amount of signaling needed. Types of Device The analysis of the device types can be done by looking at the MaDCl. In this way, we can identify if the BT device is a phone, a computer, etc. This is useful because our aim is to establish peer-to-peer networks among mobile devices. The specific type within a category can be found by looking at the MiDCl. Scanning Duration vs Number of Devices Our analysis has also been focused on the time spent to scan the mobile devices in the BT coverage. Even though the scanning should be restarted every 30 seconds, whenever a scan takes longer (because of a large number of surrounding devices) the scanning period is increased. We exploit this property of our measurement test-bed to determine how fast the sensing device can understand its surrounding environment. 7.2.8.3 First Measurement Results A set of measurements was taken during 10 days at the Aalborg Airport [131]. A commercial was boxed and installed on a wall of the waiting room at a height of around two metres. By running a specifically designed Python script, the phone was able to collect information of the BT devices present in the gate. In order to collect meaningful statistics the device ran every day from 5 o’clock in the morning to midnight, for 10 days. As the device was placed at the boundaries of the waiting room, it was able to find approximately half of the present devices that a device placed in the middle of the room would have found. The sensing device was located at that position for security reasons. Typical Results of the Measurement Campaign The results are presented for the different scenarios individually. Aalborg Airport As a first result we present the total number of devices discovered, corresponding to the times of departure and arrivals of flights given in Figure 7.14. The upper part of the figure represents the total number of devices found in each search, while in the lower part the departure times of different flights are presented. The bursty characteristic of the number of mobile devices is a result of travellers going to the gate and being embarked on the aeroplane. Even though Aalborg Airport is rather small compared to other international airports, the number of found devices is quite large. In Figure 7.15 the mean values of the number of devices found for each day (the 19 hours of scanning) at the airport are presented together with the corresponding confidence interval (HL). The histogram contains two types of bar. The dark bar (on the left side) is proportional to the mean value of devices found over the 19 hours. The light bar (on the right side) represents the mean number of devices in case the sensing device has found at least one device. As there are times when no one is allowed to enter the gate, the instances with zero devices found were deleted. As can be seen from Figure 7.15, the light-coloured bars are by definition higher than

Technologies for the Wireless Future – Volume 3

Number of Devices

268

Devices in AA Airport on 1/4/2007

18 16 14 12 10 8 6 4 2 0

Total Dev

Departures

5

6

7

8

9

10

11

12

13

14 15 Time [hh]

16

17

18

19

20

21

22

23

24

Figure 7.14 Number of wireless devices (Bluetooth-enabled) discovered at Aalborg Airport

Mean and HL of Devices in AAlborg Airport 6 Mean with 0 Mean without 0 HL

Mean # of Devices

5

4

3

2

1

0 4

1/

4

2/

4

3/

4

4/

4 4 6/ 5/ Days [dd/mm]

4

7/

4

8/

4

9/

/4

10

Figure 7.15 Basic wireless device statistics for Aalborg Airport

the dark-coloured ones. To summarize, the mean value of mobile devices found is around five, but for certain situation up to 16 devices can be found. In Figure 7.16, a distribution of different types of device and types within the phone category is presented. As can be seen from the upper plot, more than 90% of the detected BT devices were typical mobile phones. The reason for that is evident: today almost everyone owns a mobile device, while laptops, or other types of BT device like peripheral, etc., are more rare. After the analysis of the category, in the second subplot, for the phone category, the different types of phone are then analyzed. From the lower plot in Figure 7.16 it can be stated that smart phones and cellular phones represent the most prevalent types in the phone category. In most of the places under investigation, common phones are far more prevalent than smart phones.

Short-range Wireless Communications

269

Types of Dev found in AA Airport

Phone Class

Uncategorized Phone Audio/Video Computer

Uncategorized Cellular Wired Modem or Voice Gateway Smart Phone Cordless Common ISDN Access

94.4% 0.4% 1.4% 1.1%

0.2% 4.9% 0.4%

20.3% 0.5%

Figure 7.16 Basic wireless device classification at Aalborg Airport

As we said before, we are also interested in the scanning time needed to find a certain number of devices. Figure 7.17 shows the scanning time (in seconds) vs the number of devices found. Because the result is a time increasing with the number of devices, it can be stated that the scanning duration is proportionate to the number of devices found. Thus, the more devices are discovered, the more time is required for the short-range link organization. This is one important result of this section, underlining the shortcomings of today’s short-range communication technologies.

Mean Time for Search/Num Devices in AA Airport Mean Time HL

Scanning Duration [sec]

160 140 120 100 80 60 40 20 0 0

1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

Number of Devices

Figure 7.17 Scanning time (mean search to find wireless devices) vs number of devices found

270

Technologies for the Wireless Future – Volume 3

7.2.8.4 Concluding Remarks on the Measuring Campaign The results of the campaign shown here, together with supplementary measurements carried out in other relevant wireless scenarios (not shown here due to space restrictions) are quite promising for the future of peer-to-peer networks. Indeed, a larger number of mobile devices could be found in most of the explored scenarios. So the fundamental precondition for cooperation – being surrounded by enough potential partners to interact with – appears to be fulfilled in typical scenarios where people spend most of their time. Note that the numbers reported here represent somewhat worst-case results, as some people tend to shut down their short-range communication technology. It is expected that in the future even more devices will be equipped with short-range links, including conventional communication devices like mobile phones as well as other equipment with integrated communication capabilities such as home and office appliances, personal gadgets, etc. As more and more potential cooperative entities begin to surround us, novel efficient device discovery techniques need to be developed. Furthermore, not only do conventional services need to be adapted and enhanced in order to support the discussed novel cooperative setups, new innovative genuine cooperative services need to be developed as well.

7.2.9 Designing Rules for Future Wireless Short-range Communication Systems In this section we focus our attention on the designing rules for future wireless short-range technologies. As discussed previously, we distinguish between sensor networks, social networks and cooperative networks (e.g. supporting parallel/simultaneous cellular communications). Those network approaches are briefly compared to each other in terms of features such as low energy consumption, point-to-multipoint capability, quick device discovery and high-data rate support. As a result, we conclude that low energy consumption and quick device discovery are essential for all network types, while high data rate is only needed for cooperative networks. To achieve those goals, clever and sophisticated solutions need to be developed which at the same time are attractive from an implementation standpoint. One promising solution may be the introduction of a slotted medium access scheme, which would be beneficial for the energy

Table 7.1 Key requirements for future wireless short-range networks.

Sensor networks Low energy consumption Point-tomultipoint Quick device discovery High data rate

Social networks

Cooperative networks

Short-range Wireless Communications

271

consumption and the device discovery. A summary of the key requirements for future wireless short-range networks is presented in Table 7.1. 7.2.10 Short-range Communication as the Main Driving Force for Cooperative Networking The short-range communication capability is a must for functional cooperative networking in a Cellular Controlled Short-range Communications (CCSRC) system. Furthermore, the short-range technology-dependent system parameters determine how large the benefit of CCSRC can be. These parameters include the data rate of the short-range technology, the power consumption (sending, receiving and idle), and medium access control (MAC) operations. While the data rate and the energy values have a clear impact on the energy per bit ratio (EpBR), the impact of the MAC may not necessarily be obvious. Therefore, as an example we refer to wireless local area networks (WLAN) and Bluetooth as representative candidates for distributed and centralized short-range communication approaches, respectively. Certainly, wide-area cellular networks could also be considered in the example. While WLAN supports the mobile device providing the cellular information received over the WLAN in a multicast fashion, Bluetooth does not. In a Bluetooth-based communication network, the master terminal is able to control the communication of up to seven active slaves and some more inactive slaves. No direct communication between slaves is possible and therefore the transparent energy per bit ratio (TEpBR), defined as the overall energy needed to convey information from a particular slave to another (via the base station), is larger than the EpBR. To illustrate this concept, we consider the WLAN scenario, with J cooperative devices. In such as scenario, J multicast packets are exchanged among the mobile devices. Each mobile device sends one packet to the cooperating device and expects J – 1 packets to receive, as given in Table 7.2. If one wants to get a better insight into the number of packets sent over a Bluetooth-enabled cooperative cluster, the calculation becomes slightly more complex: first the master can multicast its own packet towards the J – 1 mobile slave devices. The J – 1 mobile slave devices cannot multicast their packets directly to the neighbouring slaves and need to relay them over the master. This ends up in an overall number of packets sent equal to 2 J – 1. It is important to note, as given in Table 7.2, that the master is more active than the individual slaves. While the power consumption of the slave is equal to a cooperating mobile device in the WLAN scenario, the master takes all the burden as it is sending J times more packets than the slaves. Because of this, a round robin of the master role would be fair. In this case the mean value of sent packets is 2 – 1/J . For large values of J , we can say that we need double the number of packets to exchange, which also doubles the TEpBR values in contrast to the EpBR. As one conclusion on the designing principles for future short-range wireless networks, we can Table 7.2 Number of packet exchanges for the WLAN and Bluethooth cases (master and slave separated).

WLAN Bluetooth Individual Mean

Sending 1 Master Slave 1 J 2 − 1/J

Receiving J−1 Master Slave J−1 J−1 J−1

272

Technologies for the Wireless Future – Volume 3

highlight the importance of the three key factors, data rate, power values and the used MAC concept. A lesson learned from the last section is that the MAC should support direct communication between cooperating entities in a point-to-multipoint fashion. The reason is to avoid relaying through a central entity, as would happen in the Bluetooth communication case. Furthermore, to minimize the idle time as well as to reduce collision times, the channel should be slotted in time or frequency. This is not viable in IEEE-based WLAN systems. For power saving, it is crucial to be able to switch on and off the RF/baseband chain in time domain or to dynamically enable/disable frequency bands reducing the complexity, a feature that can be found in DVB-H technology. 7.2.11 Conclusions In this contribution we discussed integrative and cooperative aspects of short-range systems in order to answer the following questions: how will short-range networks be integrated into future highly heterogeneous wireless networks? and how can cooperation between the short-range network and other component networks be exploited? Together with the omnipresent cellular systems (e.g. broadcast systems, wide area cellular, metropolitan networks), short-range networks will be key players in future highly-heterogeneous wireless communication networks. This section discussed the most important emerging approaches, including peer-to-peer networks and their social aspects, hybrid mobile device and sensor networks, wireless grids, and the concept of cellular-controlled peer-to-peer networks. We highlighted the social aspect of cooperation, where behind every mobile device there is a user deciding whether to cooperate of not. In other words, non-altruistic cooperative scenarios are considered of interest, where such egoistic cooperation brings advantages to each user joining a cooperative cluster. This is a clear departure from the conventional (forced) cooperation, for example multi-hop or relaying techniques, where a relaying device, for instance, helps other devices to deliver some information without gaining anything. We also discussed the basic cooperative scenarios and presented the results of a measurement campaign showing the potentially rich cooperative environment surrounding a mobile device. Finally, some designing rules for future wireless short-range communication networks were discussed, highlighting again the central role of these networks in future wireless communication systems.

7.3 Ultra Wideband Radio over Optical Fibre1 We propose a new concept to address the fundamentally short-range limited ultra wideband (UWB) wireless communications. The main idea behind the proposed concept is to deliver UWB radio signals over mixed wireless RF and optical fibre channels, where the optical part is serving as a super efficient transparent medium to carry the radio signal of several GHz bandwidth. The new concept is called ‘UROOF’ (UWB radio-over-optical fibre). Future research should therefore aim at developing the enabling key building blocks for converting UWB signals from electrical domain to optical (E/O conversion) and vice versa (O/E conversion) and to research the basic limitations of the overall UROOF channel. Based on UROOF technologies, one can envision a range of new services and applications as: range 1 This

work is partially supported by the European Commission under project UROOF FP6-2005-IST-5-033615; see http://www.ist-uroof.org/and list of contributors in Section 7.5.

Short-range Wireless Communications

273

extension of wireless personal area network (WPAN) by 2–3 orders of magnitude; new optical/wireless infrastructures capable of delivering broadband multimedia and above 1000 Mbps traffic to and from subscribers in remote areas. Another application is related to homeland security; it collects data from a large number of sensors and cameras equipped with UWB and transmits it over existing optical infrastructures using UROOF technologies.

7.3.1 Introduction Ultra wideband (UWB) communications is a fast emerging technology that offers new opportunities and is expected to have a major impact on the wireless world vision of 4G systems [16,61]. The important characteristics of UWB signals are their huge bandwidth (3.1–10.6 GHz, and more recently 57–64 GHz) and their very weak intensity, comparable to the level of parasitic emissions in a typical indoor environment (FCC part 15: – 41.3 dBm/MHz). The ultimate aim of UWB systems is to utilize broadband unlicensed spectrum (FCC part 15: 3.1–10.6 GHz) by emitting noise-like signals. The major UWB advantages are potentially low complexity and low power consumption, which implies that UWB technology is suitable for broadband services in the mass markets of wireless personal area networks (WPAN). Furthermore, UWB combines the high data rates with capabilities of localization and tracking features, and hence opens the door for many other interesting applications, such as accurate tracking and location, safety, and homeland security. For these reasons, UWB is considered a complementary communication solution within the future 4G systems. However, the current high-data rate UWB systems (e.g. 480 Mbps, see [25]) and future evolving multi-gigabit UWB version IEEE802.15.3c at 60 GHz [63] are inherently limited to short ranges of less than 10 m. This is simply derived from the constraints on allowed emission levels and fundamental limits of thermal noise and Shannon limits [55]. Larger coverage of high-data rate UWB to say 10–10 000 m is most desired for broadband access technology. We propose a new concept for converging between the high-data rate wireless short-range communications based on UWB technologies and the wired optical access technology. The main idea behind the proposed concept is to enable the transmission of UWB radio signals transparently over optical channels. This concept is called UWB radio-over-fibre (UROOF). In this approach, the UWB RF signals of several GHz are superimposed on the optical continuous wave (CW) carrier. This strategy has several advantages: 1) it makes the conversion process transparent to the UWB’s modulation method; 2) it permits avoiding the high costs of additional electronic components required for synchronization and other processes; 3) it makes possible integration of all the RF and optical transmitter/receiver components on a single chip. The overall combination of these features results in a cost-effective broadband system that can easily support 1 Gbps, suitable for the residential markets as WPAN. The UROOF concept is a new paradigm that extends the state-of-the-art radio-over-fibre (RoF) technologies to the short-range communication case. RoF systems in mobile cellular are motivated by the demand for replacing a central high-power antenna with a low-power distributed antennas system (DAS). RoF technologies are successfully deployed for in-building coverage in 2G/3G cellular networks. In this application, many remote access units (RAU) serving as low-cost base stations (BS) are connected to a single central station (CS).

274

Technologies for the Wireless Future – Volume 3

The converged UROOF, proposed in this contribution, is a promising approach in the field of short-range communication applications. One of the most important goals of this approach is to overcome the inherently limited range of high-data rate UWB short-range communications, extending it by 1–3 orders of magnitude compared to the state-of-the-art UWB radio systems in the 3.1–10.6 GHz band. Similar to RoF, UROOF allows separation of low-cost BSs from the CS. The main differences between UROOF and the conventional RoF are: ž The state-of-the-art RoF technologies are used in the backbone of the wireless access systems, whereas UROOF addresses the challenges of range-extended low-cost WPANs. ž In RoF, which targets the 2G/3G cellular systems, the RF signal bandwidth is only a few tens of MHz and its average power is in the range of several hundreds of mW. This requires high-cost photonic components in the CS and medium-cost components in the BS. UROOF, on the other hand, targets the PAN market, which is characterized by very low-cost and low-power (tens of mW) access point. In UROOF, the optical fibre is used to carry extremely wide RF signals (several GHz). Other approaches to extending the fundamentally limited short-range nature of the high-data rate UWB system are either practically not realizable due to link budget considerations, or too expensive for the WPAN market. For example, the free space optical link approach suffers a 41 dB loss over 10 km, roughly the same as 0.375 inch coax cable of 10 km length operating at 10 GHz [22]. In the free-space RF approach, the signal link losses at say 4 GHz centre frequency would be at least 125 dB. These should be compared to the extremely low loss level of optical fibres, about 3 dB at wavelength 1.55 mm. A legacy approach to demodulating the UWB signal and transmiting it digitally as 1 Gbps Ethernet data over single-mode fibre is too expensive for WPAN applications. Furthermore, this solution is tailored to the specific UWB technology being employed. We are looking for much more generic (e.g. agnostic to the specific modulation technology at the access point) and scalable solutions that can be easily applied to other less demanding cases (e.g. range extension of wireless local area networks (WLAN)). Other potential solutions using ad hoc and multihop network topologies to deliver the high data rate between the nodes of WPAN would impose major delay constraints that would prevent the sending of most target WPAN services and applications. Currently, three main flavours of UWB technology are proposed for WPAN communication: impulse radio (IR-UWB), direct sequence (DS-UWB) and multiband OFDM (MB-OFDM); see updated survey in [16]. In IR-UWB, information is carried in a set of narrow-duration electromagnetic pulses. The bandwidth is inversely proportional to the pulse width. Unlike conventional wireless communications systems, which are based on carrier modulation, IR-UWB is essentially a baseband (‘carrier free’) technology. The centre frequency in IR-UWB is dictated by the zero crossing rate of the pulse waveform. In general, waveforms for IR-UWB are designed to obtain flat frequency response over the bandwidth of the pulse and to avoid a DC component. Various ‘monocycles’ waveforms have been proposed to meet these characteristics including Gaussian, Rayleigh, Laplacian and cubic [21]. The time domain representation of Gaussian

Short-range Wireless Communications

275

monocycle pulse is given by: "



pG .t/ D A G 1


t
3:5 ¦

2 #

"

2 # t
3:5 exp
0:5 ¦ 

(7.1)

where the monocycle pulse width Tp is chosen to be T p D 7¦ in order to ensure that 99.99% of the total energy is captured in the Gaussian monocycle. A G is the amplitude of the monocycle, chosen in such a way that it has unit energy. The frequency spectrum of the Gaussian monocycle is given [21] by: SG . f / D A G

p

 2³ ¦ 2 .2³ f /2 exp

.2³ ¦ f /2
2

½ (7.2)

Note that the Gaussian monocycle has an even symmetry, and this can reduce the mathematical analysis of performance of IR-UWB over various channels. Various data modulation formats may be used with IR-UWB. Among them are pulse amplitude modulation (PAM), pulse position modulation (PPM) and pulse shape modulation (PSM) [44]. Typically, the PPM format is combined with time hopping to spread the RF energy across the frequency band and reduce the spikes in the spectrum of the pulse train. Direct sequence UWB (DS-UWB) borrows concepts of conventional DS spread spectrum (DS-SS). By utilizing a spreading code like pseudo noise (PN) sequence, or ternary spreading codes with prescribed correlation properties [28] and a given chip time Tc , a spreading effect is accomplished. The chip duration plays the same role as the pulse width, T p , of the monocycle in IR-UWB. However, the DS-UWB is indeed a completely different concept. It has constant envelope modulation format and the information signal is modulated twice: by a spreading code and by a carrier waveform. Unlike IR-UWB, where the interference suppression mechanism is based on time windowing over Tp , the DS-UWB is based on cross-correlation properties of the spreading code and subsequent lowpass filtering at the information bandwidth at the receiver. The DS-UWB waveform proposed in [28] is based upon dual-band BPSK and 4-level bi-orthogonal modulation (4-BOK), with band limited to baseband data pulse. The mandatory lower band ranges from 3.1 to 4.85 GHz, and the optional upper band from 6.2 to 9.7 GHz. MB-OFDM [34] is based on subdividing the UWB spectrum into five band groups and 14 sub-bands of 528 MHz width (see Figure 7.18). Only band group 1 is mandatory, while 2 to 5 are optional. Each 528 MHz channel is composed of 128 subcarriers OFDM signal, each QPSK modulated. The OFDM symbol period is 312.5 ns and time-frequency code switches the signal between the sub-bands of a given group band. It is the goal of this research to address some of the following challenges: ž What are the key requirements for realizing UWB over fibre for each of the applications: range extension, wireless distribution network, wired (e.g. hybrid fibre coax (HFC)) distribution network? ž Which UWB technology – IR-UWB, DS-UWB or MB-OFDM – meets best the aforementioned applications of UWB over fibre?

276

Technologies for the Wireless Future – Volume 3

Band Group #1

Band Group #2

Band Group #3

Band Group #4

Band Group #5

Band #1

Band #2

Band #3

Band #4

Band #5

Band #6

Band #7

Band #8

Band #9

Band #10

Band #11

Band #12

Band #13

Band #14

3432 MHz

3960 MHz

4488 MHz

5016 MHz

5544 MHz

6072 MHz

6600 MHz

7128 MHz

7656 MHz

8184 MHz

8712 MHz

9240 MHz

9768 MHz

10296 MHz

f

Figure 7.18 MB-OFDM band plan

ž How can we achieve integration of UWB over fibre technology into next-generation systems (4G, next-generation HFC, etc.)? ž Which media access control (MAC) protocols and multiple access schemes best fit UWB over fibre? The rest of this section is organized as follows. In Section 7.3.2 we briefly review UROOF background and state of the art in radio-over-fibre (RoF) technologies. In Section 7.3.3 we present typical user applications for UROOF technologies. In Section 7.3.4 we discuss fundamentals of UROOF technologies, and in Section 7.3.5 we discuss link budget aspects for future UROOF systems. We point out in Section 7.3.6 technology trends to be explored for realizing UROOF systems, and give some concluding remarks. 7.3.2 Background and Motivation 7.3.2.1 Objectives The aims of this contribution are to: 1. Introduce UROOF case scenarios, system and device concepts that can contribute to the 4G vision. 2. Provide detailed performance analysis of typical UROOF applications with different UWB modulation schemes to select the best technique for UWB radio-over-optical fibre distribution, and to determine the optimum RoF distribution configuration. 3. Focus on the challenging problem of low-cost and high-performance conversion of high-data rate modulated communication signals from optical domain (over single mode and multimode fibre) to UWB radio frequency domain and vice versa. 4. Identify future research areas related to the design and performance of UROOF novel building blocks, including: evaluation of the impact of electro-optical device nonlinearities (pulse chirp/chromatic dispersion) on the UROOF distribution network; device suitability studies for multiple-access architecture VL-DAS; uncooled VCSEL-based design; optical-UWB link analysis. 7.3.2.2 State of the Art of Radio-over-fibre Radio-over-fibre (RoF) is an analogue optical link transmitting modulated RF signals. The development of RoF systems is motivated by the demand for replacing a central high-power

Short-range Wireless Communications

277

antenna with a low-power distributed antennas system (DAS) [48]. RoF systems are usually composed of many base stations (BS), which are connected to a single central station (CS). Therefore, much effort has already been devoted to reducing the BS’s cost and moving the complexity to the CS [57]. Recently, RoF-based wireless ‘last mile’ access network architecture was proposed [18] as a promising alternative to broadband wireless access (BWA) network architecture. In such a network, the CS performs all switching, routing and operations administration maintenance (OAM). An optical fibre network interconnects a number of simple and compact antenna BSs for wireless distribution. The BS has no processing function and its main function is to convert the optical signal to wireless and vice versa. Each BS serves as an access point for an area in which many subscriber stations (SS) exist. The proposed architecture assumes a centralized medium access control (MAC) located at the CS responsible for offering a reservation-based, collision-free medium access. Currently, RoF systems are used for several applications, such as: ž 2G/3G cellular systems: ž Radio coverage extension in dense urban environments [46]. ž Capacity distribution and allocation [46]. ž Metropolitan area networks (MAN), both wired (Cable TV) and wireless (IEEE 802.16x) broadband access systems. ž Intelligent transport systems (ITS), road-to-vehicle communication systems using e.g. the 36/37 GHz carrier frequency [36,46]. ž Other applications, such as use of RoF in RADAR systems to isolate the RADAR control station from RADAR antenna. Analogue optical links that can support transmission of the entire RF band of a few GHz over large distance are of a great importance. The fibre link losses are very small as compared to the wireless or wired-coax channel, as mentioned above. The power budget of a UROOF system is determined by the contributions of RF losses, optical fibre losses and conversion losses. The RF–optical conversion losses are dominant in the power budget. However, the total losses can be substantially reduced by using optical fibres instead of a cable or wireless transmission, due to the extremely low losses per unit length of an optical fibre. For this reason, the development of efficient and cost-effective RF–optical links for UWB technology is a main objective of our research. It is important to understand the factors behind the conversion losses and to develop techniques for reducing them. We now briefly overview key building blocks of RoF technologies. Up-conversion from RF to Optical Domain This can be realized by either direct laser modulation or external modulation methods [56]. Direct methods have the advantages of simplicity and low cost. Their main disadvantages are relatively limited bandwidth (10 GHz), high chirp, nonlinear and intermodal distortion, and SNR limited by relative intensity noise (RIN). Common external modulation methods are: ž Mach-Zehnder (MZ) interferometer, characterized by limited bandwidths (2–3 GHz), high linearity, low chirp and high bias voltage. In particular, travelling wave (TW) configuration of the MZ modulator permits overcoming of the bandwidth limitations. ž Electro-absorption modulator (EAM), characterized by high bit rate and compatibility with the advanced photonic technologies. EAMs based on the quantum-confined Stark effect (QCSE) in quantum wells can exhibit advanced performance.

278

Technologies for the Wireless Future – Volume 3

Central unit

RAU Downlink

Laser diode

Optical fiber

RF in/out

EAT

RF in/out

Uplink Photodiode Optical fiber

Figure 7.19 A bidirectional RoF BS containing a remote antenna unit (RAU) based on EAT

Down-conversion from Optical to RF Domain This can be implemented by a range of high-speed photodetectors and PIN diode and avalanche photodiode (APD) photodetectors, and characterized by simplicity and typical bit rates of a few 10’s of Gbps. Travelling wave PIN based on QWs is used to generate microwave power. Frequency response of a TW photodetector reaches – 3 dB at approximately 100 GHz. Bi-directional Conversion based on Electro-absorption Transceiver (EAT) EAT acts as a receiver for the downlink and as a modulator for the uplink (Figure 7.19). This approach is appropriate mostly for mobile and MAN systems, in which the cost of the BS is greatly reduced and the complexity is moved to the CS. However, in this case an optical amplifier such as Erbium-doped fibre amplifier (EDFA) is used in order to compensate the link losses. Since EAT operates mainly in the 1500 nm range, it is compatible only with single-mode fibre and Fabry-Perot cavity or distributed feedback (DFB) lasers. It can be made to cover other wavelengths as well (1550, 1310 nm). However, when it is necessary to amplify the signal with an EDFA, it is necessary to choose 1550 nm wavelength. EDFAs are now not the only method to provide optical amplification in networks, as in recent years significant progress has been made with semiconductor optical amplifiers (SOA), which can be designed to operate either in the 1310 nm band or the 1550 nm band. SOAs have the longer-term potential to be lower-cost devices, so they may be more appropriate to UWB distribution networks with many distribution taps to base stations. EAT can operate at up to 60 GHz and appears to be the most promising candidate for BSs in the future broadband wireless access systems [46,35]. A promising EAT consists of a multiquantum well (MQW) III–V semiconductor-active waveguide. We further discuss these technologies in detail in Section 7.3.4. 7.3.2.3 Potential Impact of UROOF Technologies UWB over fibre permits us to realize convergence of different access schemes: optical-wireless, HFC-wireless, short-range personal area network (PAN)-metropolitan area network (MAN). The need for global availability, performance, increase in data rates, and network-independent services will determine the required technical characteristics of the

Short-range Wireless Communications

279

converged UROOF system. UROOF will greatly simplify the future multilayered approach of access technologies described in [60]. In this approach, five horizontal layers are introduced: distribution layer, cellular layer, hot spot layer, PAN layer and fixed (wired) layer. UROOF will allow, for example, direct and seamless connection between the distribution layer and the personal area network layer. UROOF would influence several areas of 4G wireless systems: ž Broadband access technologies (PHY and MAC layers) and interference enhancements in dense multi-user communications. ž Network layer: access across heterogeneous networks, convergence of distribution networks with short range. ž Terminals (technical capabilities vs. cost). ž Applications and services. 7.3.2.4 Contributions to Standards The UROOF concept and its outcomes might be relevant to the following regulation and standard committees bodies: CEPT ECC TG3, 802.15.4 UWB, 802.15.3c. 7.3.3 UROOF: User Applications and Basic System Configuration Several novel applications scenarios are addressed with the proposed UROOF system concepts: Case 1: WPAN range extension – UWB radio signals are transmitted and received among UROOF nodes over combined wireless-RoF links over distances around 1000 m. Case 2: Very low-cost distributed antenna system (VL-DAS) – several access points equipped with UROOF transceivers, each located in different piconet locations, are connected through RoF links to a central station. UWB signals pass over the wireless/RoF channel among the access points. Case 3: Security and home land applications – a collection of UROOF nodes (e.g. sensor network) capable of transmitting simultaneously through radio-UWB and RoF-fibre are all connected to a control station via optical fibre, enabling low-latency, location-enabled and secure (by beamforming) communications. Case 4: Capacity expansion of hybrid fibre coax (HFC) systems. Case 5: Intelligent transportation systems. 7.3.3.1 Case 1: Range Extension of Wireless PAN based on UWB Technology The UWB technology advantages are well known: i) potentially low complexity, and ii) low power consumption. This implies that UWB technology is suitable for broadband services in the mass markets of wireless personal area networks (WPAN). Inherently, the high data rates, i.e. 480 Mbps, are available over a short range of less than 10 m. Several studies point out that the usability range should be in the 3 m range. The limited UWB range limit has application in home/office-wide communication applications. The range extension application aims to enable this high data rate in extended

280

Technologies for the Wireless Future – Volume 3

O/E E/O Antenna

O/E E/O Antenna

Figure 7.20 UROOF UWB range extension application Table 7.3 UROOF Case 1.A functionalities. Min bitrate required Bidirectional capabilities Location capabilities Latency control Range extension

NO YES NO NO SHORT

Node density Multichannel UWB

HIGH NO

1 node every 1-3 m

100 m fibre span

areas from 10 to 1000 m. This means that high-data rate UWB connectivity is provided in an integrated way. This approach is depicted in Figure 7.20. The arrows indicate seamless UWB connectivity. The UWB range extension application overall performance depends on the number of simultaneous users, the traffic and the modulation type (given the packet length) used. Eventually, the collision probability can be reduced by using the different channels (sub-bands) available in the UWB spectrum (3.1–10.6 GHz). Two different cases of this application can be identified: Case 1.A: In-building UWB Extension The in-building UWB range extension architecture consists of a single optical fibre in a building (house) structure. Several UROOF nodes act as access points to provide seamless in-house UWB connectivity. Figure 7.21 illustrates this concept. Case 1.B: UWB Broadcasting This scenario targets the provision of high-definition multimedia services from an operator node or head-end to a number of subscribers employing UWB technology. This application is cost effective for the operator as no modulation conversion is required, and it is also interesting

Short-range Wireless Communications

281

O/E E/O Antenna O/E E/O Antenna

O/E E/O Antenna

Figure 7.21 UROOF Case 1.A: in-building UWB extension application Table 7.4 UROOF Case 1.B functionalities. Min bitrate required

YES

Bidirectional capabilities Location capabilities Latency control Range extension

NO NO NO LARGE

Depends on video quality, Min. 384 kbps (low-Q compressed) Max. 55 Mbps (HD, uncompressed) Node density Multichannel UWB

LOW YES

3 nodes home C Head-End Depends on the number of users

Max: 10 km

for the costumer, as the information arrives at the home as UWB signals. This application combines the economy and bandwidth advantages of fibre (or HFC) broadcasting with the cost economy and flexibility of UROOF technology. Figure 7.22 shows this application. Table 7.4 shows the functionalities of this scenario. 7.3.3.2 Category 2: Very Low-cost Distributed Antenna System (VL-DAS) When a high number of users are present in a UROOF environment, i.e. when a large area is covered, the access efficiency of the Case 1 application can be heavily affected. Case 2.A: UWB Access Segmentation To solve this issue, a cluster-based structure based on a very low-cost distributed antenna system (VL-DAS) can be adopted. This architecture is also interesting in terms of security, in

282

Technologies for the Wireless Future – Volume 3

O/E E/O Antenna O/E E/O

UWB

Antenna

HFC O/E E/O Antenna

UWB Head End

Figure 7.22 UROOF Case 1.B: UWB broadcasting

Table 7.5 UROOF Case 2.A functionalities. Min bitrate required Bidirectional capabilities Location capabilities Latency control Range extension

NO YES NO NO MEDIUM

Node density Multichannel UWB

MEDIUM YES

1 node every 20 m

order to avoid a direct access from anybody to anyone. This architecture could be leaded by the structure of the building in which the deployment has to be achieved. This application is similar to the segmentation done in LAN networks to optimize the network capability. It is depicted in Figure 7.23. In this architecture, it is clear that all the exchanges of a floor (intra-floor exchanges) are connected directly on the same optical fibre but that the inter-floor exchanges will go through concentrators. In such a structure, the local traffic of one floor has no effect on the local traffic of the other floors; the traffic is cumulated floor by floor and each concentrator (with router function) will secure the access and will act as the real access point (in the definition of the 802.15.3a and 802.15.4a standards). It should be noted that the concentrators in Figure 7.23 are not part of UROOF project, since the project targets the access node (AN) only in this topology.

Short-range Wireless Communications

Concentrator

283

O/E E/O

O/E E/O

Antenna

Antenna

Concentrator

O/E E/O Antenna

Concentrator

O/E E/O

O/E E/O

Antenna

Antenna

Figure 7.23 UROOF Case 2.A: UWB access segmentation

Case 2.B: UWB MIMO Extender Coverage This architecture is also suitable for providing enhanced smart antenna processing. In Figure 7.24 we consider a structure in which the number of converters is increased in order to enable new antenna processing or new triangulation computation at the concentrator level. In this application it is considered that the density of converters is sufficiently large to get the signal from one mobile by using multiple antennas. The consequence is that the concentrator can achieve a kind of ‘macro-diversity’, macro meaning that the distance between the antennas is very large in comparison to conventional diversity schemes, consequently improving the traditional diversity results. This application requires precise control of the signal latency. To go further than typical diversity mechanisms, such a structure can be used to implement MIMO algorithms or beamforming techniques. The condition for this is the ability to manage the amplitudes and phases sent to each antenna. It is a complementary way to extend the coverage or to focus the energy in one particular direction, or to avoid interference in certain directions. 7.3.3.3 Category 3: Security and Homeland Applications Case 3.A: Low-latency Communications Latency control is particularly useful for applications in which the time constraints are mandatory, as is the case for homeland security, security in tunnels, etc. A typical scenario is given in Figure 7.25.

284

Technologies for the Wireless Future – Volume 3

Concentrator

O/E E/O

O/E E/O

O/E E/O

Antenna

Antenna

Antenna

MIMO

Concentrator

O/E E/O

O/E E/O

O/E E/O

Antenna

Antenna

Antenna

MIMO

Concentrator

O/E E/O

O/E E/O

O/E E/O

Antenna

Antenna

Antenna

Figure 7.24 UROOF Case 2.B: smart antenna MIMO processing

surveillance camera

O/E E/O Antenna

UWB transceiver

O/E E/O Antenna

O/E E/O

traffic control

Antenna

Figure 7.25 UROOF Case 3.A: very low-latency applications

In the case of long tunnels where security is a major concern, surveillance cameras are deployed in high density. The simultaneous use of a large number of cameras, covering any point of the tunnel or of the parallel corridors, by the security crew puts exceptional stress on the communication infrastructure. This also applies to rescue teams, who can take advantage

Short-range Wireless Communications

285

Table 7.6 UROOF Case 2.B functionalities. Min bitrate required Bidirectional capabilities Location capabilities Latency control Range extension

NO YES

Node density

NO

Multichannel YES UWB Depends on MIMO algorithm performance. 100 m per concentrator

YES MEDIUM

HIGH

1 node every 10m (MIMO requires at least two nodes detecting emission from 1 UWB terminal of 10 m range). 2 (bidir)

Table 7.7 UROOF Case 3.A functionalities. Min bitrate required Bidirectional capabilities Location capabilities Latency control Range extension

YES YES NO YES LARGE

384 kbps, surveillance camera video Node density MEDIUM 1 node every 20 m Multichannel UWB YES Depends on number of cameras Specified from traffic control standards. 10 km

of the capability to download in real time such video streams. UROOF technology provides an economical way to transfer simultaneously a number of video streams with latency control if DWDM channels are used. UROOF technology also enables latency-critical applications such as traffic control of metro/train convoys. These applications are based on small UWB transceivers attached to the convoys. Traffic control applications require a controlled latency and high availability as mission critical applications. Table 7.7 summarizes the scenario functionalities. Case 3.B: Localization-enabled Communications UROOF technology can also be applied to take advantage of the different signals received by the antenna for triangulation purposes, and the different signals transmitted for beamforming applications. The triangulation is based on the time of arrival (TOA) of the different signals. In this way it measures sequentially the distance between a mobile and a set of transceivers, and its position is evaluated. An accurate estimation can be made as long as the channel is stationary. If adequate latency control is provided, it is possible to control the time delay (in a coarse mode) between the signals radiated by the antennas. In this case it is possible to include proper beamforming to increase the communication reach and to increase confidentiality. The two mentioned applications are depicted in Figure 7.26. 7.3.3.4 Capacity Expansion of HFC Systems Yet another interesting related application is the use of UWB technology to increase the capacity of existing wired broadband technology, e.g. HFC distribution systems. This approach aims

286

Technologies for the Wireless Future – Volume 3

O/E E/O Antenna

O/E E/O Antenna

Beamforming (coarse)

O/E E/O Antenna

O/E E/O Antenna

O/E E/O Antenna

Triangulation

Figure 7.26 UROOF Case communication

3.B: localization-based (triangulation C beamforming) UWB

Table 7.8 UROOF Case 3.B functionalities. Min bitrate required Bidirectional capabilities

NO YES

Location capabilities

YES

Latency control Range extension

YES LARGE

Node density

HIGH Depends on triangulation resolution and beam steering resolution. YES 2 (bidir)

Multichannel UWB Depends on beamforming angle resolution. 10 km

to achieve 100–480 Mbps over existing HFC infrastructure by superimposing a UWB signal onto existing data signals. According to ANSI/EIA-542-1997 US standard, 6 MHz analogue video channels are combined by frequency division multiplex (FDM) using frequencies from 55 to 547 MHz. These frequencies (channels 2–78) are the forward analogue channels in a HFC network. Forward digital channels occupy a frequency range from 553 MHz (channel 79) to 865 MHz (channel 136). In some extended systems, it may reach 997 MHz (channel 158). The frequencies between 5 and 40 MHz are allocated in many countries for the digital reverse channel. The two links, downlink (O/E) from head-end to customer, and uplink (UL) from customer to head-end are combined using frequency division duplexing (FDD). Properly precompensated UWB signals are capable of coexisting with currently deployed HFC spectrum or even with the actual dc- to 3 GHz spectrum of the coax. 7.3.3.5 Intelligent Transportation Systems Road-to-vehicle and vehicle-to-vehicle are the basic components of evolving technologies to increase the safety of people on the roads and enable a comfortable mobile traffic environment.

Short-range Wireless Communications

287

Recently, a new family of IEEE standards, aimed at providing communication between cars and roadside infrastructures has been approved for trial use. This family, known as the IEEE 1609 suit of WAVE (Wireless Access in Vehicular Environments), is based on existing 802.11 short-range communications at 5.9 GHz. Cars with WAVE may be reality in about 2011. UROOF technologies can be integrated into ITS concept and provide the means to add broadcasting and anti-collision radar information. 7.3.4 Fundamentals of UROOF Technologies 7.3.4.1 Introduction Optical communication systems use carrier frequencies of about 1014 Hz in the visible or near-infrared region of the electromagnetic spectrum. Lightwave systems usually employ optical fibres for information transmission. The well-known advantages of the fibre-optic communication systems include, for instance, low losses of transmitted power, compatibility with lasers and semiconductor optical amplifiers (SOAs), possibility of operation as Erbium-doped fibre amplifiers (EDFA) and Raman amplifiers, high bit rate of information transmission (over 40 Gbps), large transmission distances, possibility of a system capacity increase by using the wavelength division multiplexing (WDM) technique, soliton communication systems, etc. [17]. Fibre-optic communication systems can be classified into three broad categories: point-to-point links, distribution networks and local area networks (LAN) [17]. These categories fall into two main applications: ž long-haul applications ž short-haul applications. Long-haul telecommunication systems transmit optical signals over large distances (greater than or equal to 100 km). Transmission distances can be substantially increased up to thousands of kilometres by using optical amplifiers. Short-haul telecommunication applications are used in intra-city and local-loop traffic, i.e. over distances of less than 10 km. Multichannel networks with multiple services requiring high bandwidths can be cost effective in such applications. Only fibre-optic communications can meet such wideband distribution requirements [17]. Most UROOF applications are related to the short-haul case. The information transmitted by fibre-optic communication systems is contained in a series of electrical analogue or digital signals. The UROOF (ultra wideband radio-over-optical fibre) technology is a novel technology for the transmission of UWB signals by using an optical carrier propagating through an optical fibre. UWB high-speed optical link includes an optical fibre, electrical/optical (E/O) converter and optical/electrical (O/E) converter. A key problem of the optical communication system design is the modulation of an optical signal at the input of the system and separation of the electrical signal envelope from the optical carrier, i.e. detection at the output of the system. The essential advantage of analogue fibre-optic communications is the possibility to transmit multi-level modulated radio signals over the fibre as an envelope of the optical carrier. This approach permits us to keep exactly the same original signal format needed for the wireless communication at the access portion of the network. After O/E conversion it is possible to use a standard radio receiver for further detection of the multi-level modulated signals.

288

Technologies for the Wireless Future – Volume 3

In the following subsections we discuss the O/E and E/O processes related to the UROOF concept. 7.3.4.2 Conversion from RF to Optical Domain In order to convey analogue signals over an optical fibre it is necessary to impress the analogue signal on the optical carrier via an optical modulation device [23]. Any of the parameters of the optical carrier can be modulated. In practice, intensity modulation of the optical carrier is used at present. The essential parameters characterizing the performance of an analogue optical link are the following [23]: ž The gain of the intrinsic link is defined as the power gain between the input to the modulation device and the output of the photodetection device. ž Noise figure (NF) is the ratio of the signal-to-noise ratio (SNR) at the link input to the SNR at the link output. ž Spurious-free dynamic range (SFDR) is the SNR for which the distortion terms equal the noise floor. There are two main methods for the intensity modulation of the optical carrier [56]: direct modulation and external modulation. Direct Modulation The diode laser intensity is modulated by an analogue microwave (MW) signal. The advantages of this method are its simplicity and low cost. The rate of the direct modulation is limited to about 10 GHz due to the relaxation resonance of commercial diode lasers. The best laboratory devices known so far have maximum modulation frequency of about 40 GHz [38]. Direct-modulated vertical cavity surface emission lasers (VCSEL) seem to be the most promising candidates for UROOF systems below 10 GHz due to their low cost, efficient coupling with optical fibres, and a small number of longitudinal modes. External Modulation In such a case the laser operates in continuous wave (CW) regime and the intensity modulation is imposed via a device external to the laser. High efficiency of RF-to-optical conversion requires a high slope efficiency, a good matching of impedances and low RF losses. The advantages of the external modulation are mainly in terms of bandwidth approaching 100 GHz, high linearity range and spurious free dynamic range (SFDR). Traditionally, external modulation has been implemented by means of bulk nonlinear crystals such as LiNbO3 [56]. Commonly used external modulators are Mach-Zehnder (MZ) interferometer and modulator. Recently, the focus has been shifted to integrating electro-absorption Mach-Zehnder interferometers (MZIs) or simpler electro-absorption modulators (EAMs) onto the same substrate as the laser diode. Another type of high-speed modulator is based on the quantum confinement Stark effect (QCSE) in multiquantum well (MQW) structures [43]. For instance, an analogue EAM based on InGaAsP MQW has been developed which operates at 60 GHz [37].

Short-range Wireless Communications

289

In this section we will focus on two seemingly promising techniques: i) direct modulation of VCSELs up to 10 GHz, and ii) EAT that can be operated at higher frequencies, up to 100 GHz. Vertical Surface Emitting Laser (VCSEL) Technologies VCSELs are characterized as high-performance and cost-effective lasers and have been widely used in on-off digital optical communication systems at bit rates up to 10 Gbps [38]. However it is not obvious that the same performance can be accomplished for analogue signals. The desired characteristics of RF-to-optical converters are quite different. They include RIN, linearity, impedance and modulation efficiency. However, these characteristics are often achieved at the expense of a decreased gain as well as dynamic range, and also an increased noise figure of the optical link. The common VCSEL structure employs an active region consisting of multiple quantum wells sandwiched between two epitaxially-grown distributed Bragg reflectors. These reflectors are made up of 20–40 alternating quarter-wave layers of semiconductors with different compositions (typically AlxGa1-xAs), providing high reflectivities of larger than 99%. Due to their vertical-cavity geometry, VCSELs offer a number of significant advantages over edge-emitting lasers. Since VCSELs have a relatively small active volume, very low-threshold currents are attainable of about several ¼A. Because of the symmetry in the wafer plane, the laser output is suited for high coupling efficiency to optical fibres with relaxed alignment tolerances, as well as easy focusing into a tight spot for optical storage applications. Thus, VCSEL is a low-cost, easily-packaged, compact light source. The quantum nature of photons and electrons as well as the randomness of the physical process involved in the converting devices results in residual fluctuations of the laser beam even for the case of no modulation signal. Clearly, the residual intensity fluctuations, i.e. noise, at the laser output when no modulation signal is applied will impose a lower limit on the minimum RF signal that can be conveyed by the link. The intrinsic intensity noise of a semiconductor laser is quantified by its relative intensity noise (RIN). VCSELs are nonlinear devices. Major sources of the distortion are the relaxation oscillations and spatial hole burning. These nonlinearities in the VCSELs cause distortion of the analogue signal. The third-order intermodulation (IMD3) distortion is very important for the system applications. The IMD3, together with the modulation response and noise, determines the spurious free dynamic range (SFRD). The maximum signal power that gives the maximum dynamic range is determined by the point where noise and intermodulation are equally large. In order to achieve low levels of both RIN and coupling losses, we are required to use single-mode VCSELs which are characterized by low beam divergence and do not have noise due to mode partition. VCSEL Dynamic The operating characteristics of semiconductor lasers are well described by a set of rate equations that govern the interaction of photons and electrons inside the active region. A rigorous derivation of these rate equations generally starts from the Maxwell’s equations, together with a quantum-mechanical approach of the induced polarization. However, the rate equations can also be obtained heuristically by considering the physical phenomena through which the number of photons and the number of electrons are changed with time inside the active region.

290

Technologies for the Wireless Future – Volume 3

The rate equation for the photons is given by the rate of change of photon density p, defined by the difference between the stimulated emission and spontaneous emission radiation and the cavity losses:   dp a.N
N0 / p þ 0N D 0
ÞT;l Vg p
C ; dt 1 C "p −p −e

(7.3)

where ÞT;l D Þloss C L1 ln R is the total loss coefficient, L is the laser active region length and R is the reflectivity of the mirrors. The rate equation for the carrier density N has the form: dN I .t/ Vg a.N
N0 / N D
p

BN2
CN3 dt qV 1C" p −e

(7.4)

Here I .t/ is the current, and carrier loss rate is due to spontaneous, stimulated, bimolecular and Auger recombination processes. All parameters used in Equations (7.3) and (7.4) are described and listed in Table 7.9. From a physical standpoint, the amplitude modulation in semiconductor lasers is always accompanied by the phase modulation in semiconductor lasers because of carrier-induced changes in the mode refractive index. Phase modulation can be included through the equation: ² ¦ d 1 1 D Þc 0Vg a.N
N0 /
dt 2 −P

(7.5)

where Þc is the amplitude-phase coupling parameter, commonly called the linewidth enhancement factor, as it leads to an enhancement of the spectral width associated with a single longitudinal mode.

Table 7.9 VCSEL parameters. Parameter 0 N0 −p −e V Þc þ Vg ÞT ot B C a "
0

Description

Typical value

Confinement factor Electron concentration Photon lifetime Electron lifetime Active region volume Linewidth enhancement factor Fraction of spontaneous emission coupled into a lasing mode Group velocity Total loss coefficient Bimolecular recombination Auger recombination coefficient Differential gain The gain compression factor The total differential quantum efficiency

0.4 1:1 Ð 1018 cm
3 3 ps 1ns 1:5 Ð 10
10 cm3 5 3 Ð 10
5 8:5 Ð 109 cm/s 65 cm
1 1 Ð 10
10 cm3 =s 3 Ð 10
29 cm6 /s 2:5 Ð 10
16 cm2 5 Ð 10
17 cm3 0.4

Short-range Wireless Communications

291

The time variations of the optical power and the laser chirp are given by: P.t/ D

1 d 0:5 p.t/V
0 h¹ I 1¹.t/ D 0− p 2³ dt

(7.6)

In Equation (7.6),
0 is the total differential quantum efficiency. The photon and electron densities within the active region of the laser are assumed uniform and the linewidth enhancement factor and the nonlinear gain compression parameter are taken to be constant for a given laser. Relative Intensity Noise A common measure of the random fluctuations in lasers is the RIN. In the single-mode operation of VCSEL, RIN is almost constant at lower RF frequencies (several MHz) and peaks at the laser resonance frequency (about several GHz). Rate equations (Equations (7.3)–(7.5)) can be used to study laser noise if we add a noise term, known as the Langevin force, to each of them. Equations (7.3)–(7.5) then become:   a.N
N0 / dp p þ 0N D 0
ÞT;l Vg p
C C FP .t/ dt 1 C "p −p −e I .t/ Vg a.N
N0 / N dN D
p

B N 2
C N 3 C FN .t/ dt qV 1 C "p −e ² ¦ 1 1 d C F .t/ D Þc 0Vg a.N
N0 /
dt 2 −P where F p .t/; FN .t/ and F .t/ are the Langevin forces. They are assumed to be Gaussian random processes with zero mean, and to have correlation function of the form: hFi .t/F j .t0/i D 2Di j Ž.t
t0/; where i; j D P; N or , angle brackets denote the ensemble average, and Di j is called the diffusion coefficient (the Markovian approximation). The dominant contribution to laser noise 0V a .N
N0 / 0V a .N
N / comes from only two diffusion coefficients, DPP D g 0 V p and D D g 0V p 0 , with others to be assumed to be nearly zero. For analogue applications, intensity noise is quantified using a signal-to-noise ratio (SNR) which is linked to the relative intensity noise (RIN): m2 ; SNR D 2RIN Ž where m is the electrical modulation depth and given by m D 1I .I
Ith /. I and Ith are the input and the threshold currents respectively. RIN is defined as: RIN D

hŽ P.t/2 i ; P2

where hŽ P.t/2 i denotes the mean square power fluctuation and P is average power.

292

Technologies for the Wireless Future – Volume 3

Spurious Free Dynamic Range (SFRD) An important measure of analogue link performance is the dynamic range. The dynamic range of the link is defined as the ratio of the largest signal the system can transport to the smallest one. The total lasing power changes linearly in a wide region with operation current. Intrinsic distortions due to the intermodulation in the active region of the laser lead to distortion of the fundamental signal. Due to these nonlinearities, spurious intermodulation products are created. At low modulation depth (small m), intermodulation distortions are still below the noise floor. Increase of the modulation depth leads to distortions rising above the noise floor. These distortions grow faster than the fundamental signal. The largest distortion-free SNR determines the dynamic range and is reached when the amplitude of the distortion is equal to the noise floor. The SFRD is the dynamic range where the maximum signal is limited by the intermodulation product and is often used as a measure of the system performance. Predistortion technique of the fundamental signal can be used to increase the SFRD. Electro-absorption Transceiver (EAT) General Considerations The growing demand for UROOF technologies results in the creation of new millimetre-wave (mm-wave) devices [43]. They are implemented mostly in the Ka band (26.5–40 GHz). Now the V band (50–75 GHz) is of particular interest due to the development of ultra wideband systems with the analogous signal modulation. We consider optically-fed microwave radiolinks in a broadband optical network [43]. In the case of the so-called external modulation, the optical carrier is emitted by a continuous wave (CW) laser and modulated by a distinct device which operates a conversion from microwave region to optics and provides an efficient control of the optically-transmitted data. The modulators must not give rise to additional noise and must be highly reliable [43]. We consider below electro-absorption modulators (EAM). EAM is a compact, efficient and integrable component for fibre-optic communications [32]. It utilizes either the Franz-Keldysh effect in a bulk material, or the quantum-confined Stark effect (QCSE) in a multiquantum well (MQW) structure in order to change the absorption of light according to the applied electric field [32]. In EAMs in linear regime, the light input Pin and output Pout are proportional and the transmitted power depends on an applied voltage or electric field [43]. MQW structures are widely used due to their higher modulation efficiency, but the wavelength dependence is generally higher than for bulk material. It has been shown that EAMs based on QCSE in QWs has better figures of merit than similar devices based on the bulk materials [43]. The advantages of EAMs over other types of modulator are as follows [32]: ž A small device dimension, of the order of magnitude of a few hundred micrometres or even less. ž A lower driving voltage due to the EAM’s highly efficient and nonlinear electro-optical (E/O) transfer function. ž The possibility of monolithic integration with other semiconductor components. For instance, a compact module containing an MQW EAM integrated with a DFB laser has been realized [31]. A critical design issue of the EAM is the trade off between the modulation bandwidth and efficiency. The bandwidth of a lumped-electrode EAM is determined by the resistance-capacitance time constant, − D RC. Consequently, the active waveguide length

Short-range Wireless Communications

293

and width must be short in order to keep the capacitance low and to enable high-speed broadband modulation. On the other hand, a short active waveguide length results in a reduced modulation efficiency and a higher driving voltage. The optical power handling capability for a short device also sharply decreases, which is especially important for radio frequency photonic links [32]. EAMs with traveling-wave (TW) electrodes were developed in order to exclude the tradeoff between bandwidth and device length. In such a case, the bandwidth is not limited by lumped RC parameters and the device length can be longer, with a higher bandwidth. In the TW design, the driving signal propagates through the active waveguide, which is in fact a microwave transmission line. With a longer device length, the modulation efficiency can be kept high in such a way that a low driving voltage is required. Theoretically, the bandwidth and the useful device length of TW-EAM are limited by the microwave loss and the velocity mismatch between the lightwave and the microwave. These factors determine how long the lightwave can be effectively modulated by the microwave driving signal. The increase of optical scattering loss with device length is also a limiting factor, degrading the insertion loss of TW-EAM. A low characteristic impedance of the active waveguide (25 or even less) originating from the trade off between the junction capacitance and optical and microwave losses causes reflections and limits the modulation bandwidth [32]. Low impedance terminations in the range of 12–35  are required in order to optimize bandwidth. For instance, a bandwidth of 43 GHz was measured on a 450¼m device with a 13  termination [32]. EAM using the QCSE in strained InGaAs/AlGaAs MQW waveguide structures has been investigated experimentally [43]. It has been shown that the electrical bandwidth measured in a common 50  system has a cut-off frequency higher than 70 GHz due to an optimum impedance and phase matching. The unique properties of TW-EAMs should be explored and utilized. Consider briefly the unique properties such as: distributed effect due to TW design, generation of photocurrent, nonlinear E/O transfer function. Distributed Effect The most significant implication of the TW design is a finite length inside the TW-EAM [32]. It generates an extra dimension for device optimization, especially when the microwave frequency is large enough that the microwave wavelength is comparable to the device dimension. As a result, the interaction both in time and in space between the lightwave and the microwave should be taken into consideration. This is known as the distributed effect in TW-EAM [32]. Both the co-propagation and the counter propagation of the optical and microwave signals can be realized in TW-EAM. The co-propagation regime is preferred in the terms of output power and the shortest optical pulse width. For lower operation frequencies and shorter device length the difference between two configurations becomes insignificant. Photocurrent Generation and EAT EAMs inherently generate photocurrents [32,43]. For this reason, EAM can be used simultaneously as a modulator and a photodetector. This exceptional property makes it possible to combine several functionalities within a single TW-EAM. Such a device is called an electro-absorption transceiver (EAT). EAT has been studied experimentally for operation at 1:3¼m wavelength. The typical device consists of a slightly Si-doped lattice-matched InAlAs top cladding layer and a highly Si-doped lattice-matched InAlAs bottom cladding layer. The active region is formed by 20 n-i-d QWs with a thickness of 7.7 nm each. By implementing 1% tensile strain in the InGaAs QWs and 1% compressive strain in the InAlAs barrier layers,

294

Technologies for the Wireless Future – Volume 3

polarization incentive operation is achieved. The microwave operation of up to 70 GHz can be realized by using n-i-n EAM with hybrid coplanar-microstrip metallization [43]. For applications in ultra wideband modulation a strong photocurrent inside the active waveguide should be avoided. Such a photocurrent causes a reduction of effective voltage supplied to the active region and degrades the large signal dynamic extinction ratio. Nonlinear E/O Transformation The E/O transfer function of an EAM is determined by the material design, device length, electrode design and waveguide geometries. The polarization dependence may be essentially decreased by using strain-compensated MQWs, as was mentioned above. A typical EAM E/O transfer function dependence on the reverse electric voltage applied consists of three transmission regions: 1. high extinction (<10 %) 2. transition (10% to 90%) 3. flat transmission (>90 %). The transition voltage should be small in order to achieve high modulation efficiency and low driving voltage. The slope of the transfer function varies point-to-point because of the nonlinear character of the electro-absorption process. However, it is possible to find certain points with minimized distortion for high dynamic range RF photonic link. The voltage VH at which transition begins is typically designed close to 0 at the operating wavelength in order to reduce excess heating. If VH is shifted towards higher reverse voltage, more flat transmission region can be used without forward biasing the TW-EAM. Perspectives TW optoelectronic devices can meet the contemporary requirements for ultra-high-speed operation. In particular, TW photodetectors and modulators are not limited by the usual RC time constants. Microwave properties determine the bandwidth, and the input resistance is given by the characteristic impedance of the waveguide system. TW devices are much more flexible with respect to design parameters and permit the monolithic integration in optical MMICs. Simulation and modeling of the devices can be carried out by using equivalent circuits for the optical and electrical domain. The interaction can be described by parametrically controlled elements. There are still challenges for TW-EAM applications. The impedance mismatch between the active waveguide and driving electronics limits the bandwidth. This shortage can be improved by new materials or by design customizing. The insertion loss of TW-EAMs should be reduced. Waveguide spot-size convertors and high-level integrations are both possible solutions. 7.3.4.3 Conversion from Optical to RF Domain The optical link speed is mainly determined by the performance of the modulation and demodulation process at the transmitter and the receiver sides. The signal distortion is especially problematic in the case where the signal to be transferred is analogue and a counterpart of digital 3R regeneration (reamplification, reshaping and retiming) does not exist. Therefore, a highly linear link is required to prevent signal distortion. As a result, the design constraints in

Short-range Wireless Communications

295

analogue links are more stringent. In this section we briefly review RoF techniques that can operate from 1 to 100 GHz, with emphasis on the noise performance and link linearity. High-speed photodetectors (HS PDs) are required for use in high-speed optical interconnects, microwave photonic applications and many modern RoF communication systems. Photodetectors are used to convert the optical signal back into electrical form and recover the data transmitted through the lightwave system [17]. A photodetector should have high sensitivity, fast response, low noise, low cost and high reliability. Ultrafast photonic devices based on the microwave optical interactions are expected to be developed and incorporated in future high-speed and high-capacity lightwave systems [33,53]. High-speed Photodetectors High-speed photodetectors (PDs) based on AIIIBV compounds are used in fibre-optic communications, optical/wireless systems, detection and conversion of optical signals, and other microwave photonics applications. The high speed and operation efficiency of PDs is essential for an optical-electrical conversion of analogue signals in ultra-wide radio-over-fibre (UROOF) technology. To that end, there is an interest in PD bandwidth increasing up to 100 GHz [39]. The main types of high-speed PD are the following: ž ž ž ž ž ž ž

PIN PDs avalanche PDs Schottky PDs resonant-cavity-enhanced PDs waveguide PDs travelling wave PDs velocity-matched distributed PDs.

Based on the method of optical radiation propagation, high-speed PDs are divided into three classes: ž surface-illuminated PDs ž resonant-cavity-enhanced PDs ž edge-coupled PDs. The quality of high-speed PDs is characterized by the bandwidth-efficiency product [39]. For surface-illuminated PDs this parameter does not exceed 20–30 GHz, being limited by the trade off between quantum efficiency and bandwidth. In order to increase the quantum efficiency it is necessary to increase PD absorption layer thickness. On the other hand, an increase of the PD thickness results in a decrease of the bandwidth. In resonant-cavity-enhanced PDs, in which the PD is situated in a Fabry-Perot resonator, optical radiation multiply passes through the thin absorption layer, and quantum efficiency increases at the resonant wavelength. For this reason, the increase of the device thickness is unnecessary. Hence, the bandwidth may be larger. In the edge-coupled PDs the directions of the optical radiation propagation and charge carrier transport are perpendicular to one another. In such a case it is necessary to increase the device length instead of the thickness of the absorption layer. Below we briefly discuss the characteristics and peculiarities of these types of PD. We mainly follow the review [39].

296

Technologies for the Wireless Future – Volume 3

PIN PDs Typical PIN PD is based on pC
InP = i
In0:47 Ga0:53 As = nC
InP heterostructure. Wide bandgap pC- and nC-InP layers are hardly doped, while the absorption layer i – In0:47 Ga0:53 As is not doped and has low background impurity concentration. Under the optical illumination at the wavelength ½ D 1:1–1:6 µm the electron-hole pairs are generated in the absorption i-layer and separated by an internal electrical field of p-n junction by means of the drift mechanism in the depletion region and by diffusion in the neutral region. The high-speed operation is determined by the reverse bias large enough for full depletion of the i-layer. The quantum efficiency
of the PIN PD is given by:
D & .1
R/.1
exp.
Þd//;

(7.7)

where & is the internal quantum efficiency, which is typically close to 1, R is the reflection coefficient of the PD surface, Þ; d are the absorption coefficient of the PD surface and the thickness of the absorption i-layer, respectively. The PIN PD bandwidth 1 f is limited by the transit time and the RC-time of the PIN PD equivalent circuit. It has the form [39]:  1f D

1 1 C .1 f t /2 .1 f RC /2


1=2

" D

2³ d 3:5vd

2

 #
1=2  .Rs C Rl / 2 C 2³ "0 "r S ; (7.8) d

where v d is the average electron and hole drift velocities in the absorption i-layer, "0 ; "r are the permittivity of vacuum and of the absorption i-layer, respectively, S is the PIN PD photosensitive area, Rs ; Rl are the series and the load resistances. It can easily be shown that the optimum absorption layer thickness for a maximal bandwidth for the given photosensitive area S is given by [39]: p (7.9) d ³ 3:5v d "0 "r Rl S The numerical evaluations show that for the PIN PD a bandwidth larger than several ten of GHz requires a small photosensitive area with a diameter of about 10–20µm. The absorption layer should not exceed 1µm. The bandwidth can be enhanced by using only fast electrons for charge carrier transport, which is achieved in uni-traveling carrier (UTC) PDs. The transit-time limited bandwidth of the UTC UPD can be 200 GHz and above due to the large electron diffusion and drift velocities [39]. Avalanche PDs Avalanche PDs (APDs) are used for increasing of photoreceiver sensitivity in fibre-optic links. The increase of the APD sensitivity is due to the internal amplification during an avalanche gain of photocarriers. The process of the avalanche gain is characterized by ionization coefficients Þn ; Þ p for electrons and holes, respectively. The multiplication factor M is given by [39]: MD

.Þn
Þ p / ; Þn
Þ p exp[.Þn
Þ p /dm ]

(7.10)

where dm is the multiplication layer thickness. Low noise, fast response and high multiplication factor can be realized only if the ionization coefficients Þn ; Þ p significantly differ from

Short-range Wireless Communications

297

each other. At small reverse bias voltages APDs operate as usual PIN PD with a gain of about unity. The amplification in APDs occurs only at comparatively large reverse bias voltage of about 30–40 V. Bandwidth of the high-speed APD is limited by the finite avalanche formation time and does not exceed 10–20 GHz. The APD performance, including the bandwidth increase, can be improved by use of InAlGaAs/InAlAs super lattice with ionization coefficient ratio .Þn =Þ p / ³ 3
4. PDs based on Schottky Barrier The basic electrical characteristics of the Schottky PD are very similar to the characteristics of the PIN PD. The main advantage of the PDs based on Schottky barriers is the smaller capacitance per unit area compared to PDs based on p-n junction. For the high-speed Schottky PD creation, the thin barrier-enhanced In0:52 Al0:48 As layer of Schottky barrier height 0.8 V with Ti/Au metal contact is used. The main disadvantage of the Schottky PD is the lower quantum efficiency compared to the PIN PDs. The Schottky PD bandwidth is limited by drift time and RC time. Resonant-cavity-enhanced PDs It is possible to increase the quantum efficiency of the surface-illuminated PDs by means of resonant-cavity-enhanced (RCE) structure. In the RCE PD the surface-illuminated PD is placed in a Fabry-Perot resonator. The RCE PD quantum efficiency is given by:
.½/ D

[1
R.½1 /][1 C R2 .½/ exp[
Þ.½/d]][1
exp[
Þ.½/d]] ; p j1
R1 .½/R2 .½/ exp[
Þ.½/d] exp.
j8/j2

(7.11)

where: 8 D 4³

n 1 d1 C n 2 d2 C nd C ½

1 .½/ C

2 .½/;

(7.12)

R1 ; 1 and R2 ; 2 are the reflection coefficients and phase shifts of the forward and reverse reflectors, n; Þ; d are the refractive index, absorption coefficient and thickness of the absorption layer, n 1 ; d1 and n 2 ; d2 are the refractive index and thickness of layers above and below the absorption layer. The quantum efficiency has a maximum at the wavelength ½0 that satisfies the condition: 4³

n 1 d1 C n 2 d2 C nd C ½0

1 .½0 /

C

2 .½0 /

D 2³

(7.13)

Then we have the maximal efficiency:
max D

[1
R][1 C R2 exp[
Þd]][1
exp[
Þd]] p .1
R1 R2 exp.
Þd//2

(7.14)

and the half-maximum 1½1=2 : 1½1=2 D

½20 .1


p

R1 R2 exp.
Þd// pp 2³.n 1 d1 C n 2 d2 C nd/ R1 R2 exp.
Þd/

(7.15)

298

Technologies for the Wireless Future – Volume 3

Due to multiple propagation of optical radiation at the resonant wavelength through the absorption layer, its thickness can be reduced for receiving necessary bandwidth without any loss of efficiency. In this case the bandwidth-efficiency product can reach hundreds of gigahertz [39]. While various photodetectors such as PIN, metal-semiconductor-metal (MSM) and APD can be used for high-speed applications, each of these devices has certain drawbacks which make them rather unsuitable for use in UWB-over-fibre communications. Lateral PIN and MSM devices are simple to fabricate, but are characterized by low responsivity at wavelengths around 850 nm. APD devices, on the other hand, require relatively high bias voltages and have increased noise. Optically Controlled Microwave Devices Novel photonic techniques for optical signal generation and processing are required in ultra wideband systems [33]. Microwave photonic links may serve efficiently in analogue interconnects in state-of-the-art wideband microwave functions. Optical signals introduced into microwave devices or controlling them may be used for signal detection [53]. This approach possesses the following advantages: ž The absence of extra electronic circuits required to process the detected signals before application to the microwave device. ž The absence of circuit parasitics limiting the response speed. ž Creation of optical control port to the microwave device. ž The optical control signal is not influenced by any microwave electromagnetic disturbances. The optical control of microwave devices has been widely discussed in the last decade (see, for instance, [19,33,47,53]). Consider briefly the optical control of some typical microwave devices [53]. Amplifiers The gain of MESFETs and high-electron mobility transistors (HEMTs) can be controlled by illuminating the gate region and including an appropriate series resistor in the gate bias in order to produce a change in gate bias caused by the optically-generated current. Gain of up to 20 dB in MESFET amplifiers can be achieved using optical powers of about several microwatts. HEMT amplifiers exhibit even larger optical sensitivity that is 7–10 times higher. Oscillators There are three main forms of oscillator optical control: i) optical switching; ii) optical tuning; iii) optical injection locking. In the case of optical switching, the oscillator optical power is changed depending on the optical control signal intensity. The case of optical tuning is similar to optical switching, however the optical intensities used are rather small, and the output power change is also small. In the case of optical injection locking, the optical control is optically modulated at a frequency close to the free-running frequency of the oscillator, one of its harmonics or subharmonics. The modulated optical signal fed into the device active region creates the current at the modulation frequency. These phenomena occur in avalanche diodes, MESFETs and bipolar transistors.

Short-range Wireless Communications

299

Opto-electronic Mixers Two regimes can be realized. In the first case, the signal to be converted in frequency is supplied electrically, and the local-oscillator signal is an intensity-modulated optical source. In the opposite case, an electrical local-oscillator signal is used to down-convert an intensity-modulated optical signal. Optoelectronic mixers have been realized using photoconductive devices, diodes, field-effect transistors and bipolar transistors. Optically Controlled Microstrip Converter (OCMC) A novel concept named OCMC is based on the generation of photocarriers within semiconductor devices. The substrates possess a strongly manifested contrast between dark conductivity and the photoconductivity under illumination by the optical signal, allowing an electrical modulation from the incident light. As a result, in depletion regions a photocurrent occurs, and the built-in potential is changing. In other cases, the conductivity of semiconductors increases due to the photoconductive effect. Consequently, optically-controlled semiconductor devices such as MESFETs, HEMTs, PIN diodes and APDs can be integrated as an optically-controlled load with different types of microstrip and coplanar lines. Such a system combining all the advantages of microwave photonic devices mentioned above can be successfully used as ultra wideband OCMC. Indeed, microwaves can be fed into microstrip and coplanar lines realized on GaAs, Si or novel SiGe substrates. The optical signal modulated with an ultra wideband envelope is introduced into the optically-controlled load that provides the rapid detection of the microwave signal. Different regimes of the signal transmission can be realized, including heterodyning. Microstrip antennas can also be integrated into the system, making possible the detected ultra-wide signal radiation. The design of OCMC with a 60 GHz bandwidth and minimum losses is a substantial challenge from the point of view of the optimization of microstrip and coplanar waveguiding elements, efficient matching of the microwave transmission line with an optically controlled load, and the choice of the high-speed operating PD devices [26].

7.3.4.4 Optical Microwave Mixing Used for UWB-over-fibre Systems All-optical microwave mixing techniques could be used to up-convert a UWB signal around a RF carrier in the 3.1–10.6 GHz bandwidth. Two configurations can be employed. The first possibility is to simultaneously modulate the optical carrier and realize the RF up-converting. The second, after the optical transmission, is to simultaneously detect the baseband UWB signal and up-convert it around the RF frequency. Optical-to-RF Conversion All-optical microwave mixing has been demonstrated by direct nonlinear photodetection using CW signals [40,27]. The simultaneous injection of a microwave signal at the electrical port of the PD and a modulated optical signal permits the mixing of the two signals. The mixing process results from the nonlinearity of the PD current-voltage relationship. Due to the fact that the characteristics exhibit the maximum nonlinearity in the vicinity of 0 V, it is the optimal operation point for efficient mixing. We propose to extend this technique to optical UWB systems by replacing the optical CW signal with an optical UWB one. With this technique, the optical source (for example the VCSEL) bandwidth could be even more reduced, to the value of the baseband UWB bandwidth (500 MHz for MB-OFDM signals).

300

Technologies for the Wireless Future – Volume 3

RF-to-optical Conversion One way to generate optical-microwave mixing is to insert a nonlinear element in the optical transmission link. Different methods have already been demonstrated. Two input microwave signals at fRF frequency containing data and a local oscillator signal at fLO frequency are then converted at mixing frequency f R F šf L O at the output of the optical link. Generally, mixing is achieved by the use of optical-intensity modulation (direct modulation of laser diodes or external modulation) in different configurations: two cascaded linear modulations [54] or one modulation working in a nonlinear regime [62]. Other methods can be used by employing frequency modulation into intensity-modulation conversion achieved by passive optical interferometer [20] or by chromatic dispersion properties [30]. To implement low-cost optical UWB systems, direct modulation of VCSELs in a nonlinear regime could be used to generate optical-microwave mixing. VCSELs are low-cost components but are today limited in modulation bandwidth. Nevertheless, using optical-microwave techniques imposes less stringent requirements on the modulation bandwidth because the VCSEL is modulated by a UWB signal centred around an IF frequency, and by an unmodulated signal at LO frequency, close to the IF one. The mixing process up-converts the signal at IFCLO frequency. More precisely, requirements concerning the VCSELs can be divided into two: 5.5 GHz modulation bandwidth is sufficient to generate signals in the 3.1–10.6 GHz range. 7.3.4.5 Integrated UROOF Transceiver (IUT) The key element in UROOF architectures is the integrated UROOF transceiver. This transceiver is responsible for translation of the UWB signal for the RF domain to the optical domain and vice-versa. Two IUT approaches are investigated: 1. The optically-controlled microstrip converter consists of an open-ended microstrip line implemented on a semiconducting material. This device is intrinsically unidirectional, and produces an O/E conversion. The first IUT approach combines OCMC for the O/E conversions with E/O conversion based on VCSEL. 2. The second approach is the enhanced electro-absorption transceiver (E-EAT). This device is capable of O/E and E/O conversion simultaneously. In UROOF it is implemented in an electro-absorption modulator (EAM) optimized for bidirectional conversion. For optical input signals at a wavelength close to the absorption edge the small voltage changes applied are translated to a shift in the position of the absorption edge (i.e. transparency edge) of the device. Optical signals at a shorter wavelength than the absorption edge are absorbed and it effectively operates as a photodiode. The E-EAT is intrinsically bidirectional and is the most suitable option for bidirectional application scenarios. It must be taken into account that bidirectionality implies the use of two WDM channels, or the use of two fibres (uplink and downlink). The studies presented in this document are based on an OCMC/VCSEL. This approach is employed without lack of generality, as the parameters employed can be directly applied if an E-EAT device is employed. The OCMC/VCSEL IUT structure is depicted in Figure 7.27. The integrated UROOF transceiver (IUT) includes the OCMC and the VCSEL systems in order to convert the UWB signal from the optical to the RF domain (OCMC) and RF to optical

Short-range Wireless Communications

301

Tx OCMC optical VCSEL Rx

electrical

Figure 7.27 IUT structure including amplifier in the front end

(VCSEL). The IUT includes an amplifier that prepares the received RF signal to be converted to the optical domain and transmitted to the fibre and vice versa, and prepares the received optical signal converted to RF to be transmitted again by the TX antenna. The front-end amplifiers, standard in a typical UWB transceiver, can raise the emission power and the system sensibility considerably, and will compensate the propagation losses during free space or fibre transmission. 7.3.5 Link Analysis of UROOF Systems The optical link speed is mainly determined by the performance of the modulation and demodulation process at the transmitter and the receiver sides. The signal distortion is especially problematic in the case where the signal to be transferred is analogue and a counterpart of digital ‘3R regeneration’ (reamplification, reshaping and retiming) does not exist. Therefore, a highly linear link is required to prevent signal distortion. As a result, the design constraints in analogue links are more stringent. We briefly review radio frequency (RF) optical analogue link design that can operate up to 100 GHz. 7.3.5.1 Mixed Wireless-wired UROOF Channel Basic UROOF Point-to-point Link In this section we evaluate the performance of a simple two integrated UROOF transceiver (IUT) nodes point-to-point link. The study of this simplified scenario is the keystone of the multiple-access scenarios. The simple point-to-point link is depicted in Figure 7.28. Where PRx is the received power at the UWB receiver, PT x is the RF power delivered to the TX antenna at the UWB transmitter, PIUTRx is the received power at the IUT receiver antenna, PIUTTx is the power the IUT is going to transmit to the UWB receiver, G FeRx stands for the IUT front-end amplifier gain included in the IUT before VCSEL conversion, G FeT x stands for the IUT front-end amplifier gain included in the IUT after the OCMC conversion and before the IUT transmission, G VCSEL (dB) stands for the electrical (power) to optical (intensity) conversion gain, G OCMC (dB) is the OCMC conversion gain from optical power to RF power, G IUTTx (dBi) and G I U T Rx (dBi) are IUT TX and RX antenna gain respectively,

302

Technologies for the Wireless Future – Volume 3

PIUTTx r

GOCMC GFeTx GIUTTx

IUT d·a

OCMC

GTx

r

GRx

OCMC

Rx

VCSEL

VCSEL

PRx

Tx UWB Rx

GIUTRx G FeRx GVCSEL

UWB Tx

PTx

IUT

PIUTRx

Figure 7.28 Simple point-to-point link

G T x (dBi) and G Rx (dBi) are the UWB device TX and RX antenna gain respectively, d stands for the length of the SSMF fibre, Þ stands for the fibre attenuation at the operation wavelength, dÞ is the attenuation loss due to the fibre propagation, r is the distance the signal travels in free space from the UWB transmitter antenna to the IUT receiver. In order to simplify the calculations, without lack of generality, we will suppose the same distance between the IUT transmitter and the UWB receiver antenna. Power Budget We will summarize the power budget and calculate the expected noise from a system point of view. i.

PI U T T x PI U T Rx

stands for the power gain of the UROOF distribution stand-alone, from electrical to electrical point: G e=e D PI U T T x PI U T Rx . ii. PPTR XX stands for the UWB terminal to UWB terminal power budget, including the wireless link. Let us first calculate (i): PI U T T x G FeRx Ð G V C S E L Ð G OC MC Ð G FeT x D PI U T Rx d ÐÞ

(7.16)

Considering free-space propagation over a range r: PRx D PT x



PI U T T x PI U T Rx



PI U T Rx PT x



PRx



PI U T T x

(7.17)

Then, the overall power budget for the complete link is given by: G FeRx Ð G V C S E L Ð G OC MC Ð G I U T Rx Ð G T x Ð L 2Fs Ð G FeRx Ð G I U T T x Ð G Rx PRx D ; (7.18) PT x d ÐÞ

Short-range Wireless Communications

303

Table 7.10 Typical values for power budget calculation. SYSTEM PARAMETERS IUT front-end transmitter amplifier gain IUT front-end receiver amplifier gain electrical (power) to optical (intensity) conversion gain VCSEL OCMC conversion gain from optical power to RF power UWB device TX antenna gain UWB device RX antenna gain IUT TX antenna gain IUT RX antenna gain SSMF fiber attenuation at the operation wavelength Length of the SSMF fiber r Distance UWB transmitter antenna to the IUT receiver Frequency

VALUE

UNITS

20 20 0 10 2 2 6 6 0.2 0.1 3 5.00EC09

dB dB dB dB dBi dBi dBi dBi dB/km km m Hz

where L Fs are the losses due to free-space transmission given by:  L Fs D

½ 4³r

2 (7.19)

Typical values are shown in Table 7.10. Expected power budget: dB

49,98 PIUTTx GFeRx ·GVCSEL ·GOCMC ·GFeTx = PIUTRx d PIUTRx = GTx ·LFs ·GIUTRx PTx PRx = GIUTTx ·LFs ·GRx PIUTTx PRx P P  P  =  IUTTx  IUTRx  Rx  PTx PIUTRx  PTx PIUTTx 

−47,9635974

−47,9635974

−45,9471947

Carrier-to-noise Ratio Let us consider the degradation of the CNR due to the signal propagation through the UROOF distribution architecture. Several noise sources contribute to signal degradation in a conventional fibre link: thermal noise, laser noise (RIN), photodetector noise and intermodulation

304

Technologies for the Wireless Future – Volume 3

distortion due to the nonlinear response of the system. All these are evaluated in this section from a system point of view. The work presented here covers standard devices (VCSEL, APD/PIN photodetector) as introductory calculations. Thermal Noise The CNR after the IUT Rx antenna is C N R I U T Rx . The C N R I U T Rx at the other side of the transmission link is given by: C N RIU T T x D

f FeRx

C N R I U T Rx ; Ð f V C S E L Ð f f ibr e Ð f OC MC Ð f FeT x

(7.20)

where f FeRx and f FeT x are the noise factor of the RX and TX front-end amplifiers respectively; f f ibr e stands for the equivalent noise factor due to the signal attenuation when travelling down the fibre. In this case it is equivalent to the fibre attenuation along the fibre link; f V C S E L and f OC MC are the noise factor of the laser and the detector. Intermodulation Distortion The UROOF topology supports different UWB analogue channels multiplexed in frequency. This is a particular implementation of a subcarrier multiplexing (SCM) system. However, SCM technology is subject to some important system penalties. Because the channel spacing between subcarriers is small, a high nonlinear distortion may be expected. The nonlinearity is reflected in intermodulation distortion – composite second-order (CSO) distortion and composite triple beat (CTO). The UROOF architecture system has very stringent requirements on the noise and the linearity of the system. The linearity of the two key conversion devices (OCMC, VCSEL) is of paramount importance, so it is necessary to calculate the carrier to interference ratio (CIR) due to nonlinearity. In subcarrier multiplexed systems the various electrical subcarriers are combined and used to modulate the optical signal, which is then directly detected. If the devices used in the electrical-to-optical (E/O) and optical-to-electrical (O/E) conversion are nonlinear, the various subcarriers are mixed to form intermodulation products. Second- and third-order intermodulation products (IMPs) are generated by every combination of two and three input frequencies, respectively. The interference resulting from source nonlinearity then depends strongly on the number of channels and the distribution of channel frequencies. Let us consider transmission of three channels (Figure 7.29) with subcarrier frequencies f1 , f2 and f3 . In the UROOF application, second-order IMPs, f i š f j , will have to be taken into account since the transmission bandwidth occupies more than one octave. The most troublesome third-order distortion products are those that originate from frequencies f i C f j – f k and 2 f i – f j , since they lie within the transmission band, leading to interchannel interference. The interference thus depends strongly on the number of channels and generally on the allocation of channel frequencies. For an N-channel system with uniform frequency spacing the number of IMPs, and of type 2 f i – f j and f i C f j – f k , respectively, coincident with channel r is given by [59]: rIM

N 21

² ¦ 1 1 N r N
2
[1
.
1/ .
1/ ] D 2 2

(7.21)

Short-range Wireless Communications

rIM

N 111

D

305

1 r 1 .N
r C 1/ C [.N
3/2
5]
[1
.
1/ N ].
1/ N Cr 2 4 8

(7.22)

Considering an OFDM-UWB system, and considering only one sub-band, N D 122, Figure 7.30 shows the total number of third-order IMPs as a function of channel number.

Transmission band

f1+f2−f3

f2−f1 f3−f1

Laser modulation response

f1+f2+f3

f2+f3−f1

2f1+f2 2f1 2f f1+f3 3 f1+f2 f2+f3

f f f 2f1−f3 1 2 3 2f3−f1

3f1

2f3+f2 3f3

2f1+f3 2f3+f1

Figure 7.29 Intermodulation products and harmonics generated by a three-tone modulation of a nonlinear device

5600 5400

Total number of thrid-order IMPs

5200 5000 4800 4600 4400 4200 4000 3800 3600

0

20

40

60

80

100

120

140

Channel number

Figure 7.30 Total number of third-order intermodulation products as a function of channel number for an N D 122 channel system

306

Technologies for the Wireless Future – Volume 3

It is seen that the central channel is the one with the largest number of third-order IMPs. N N and r M2I approach the asymptotic value of 3N2 =8 and N =2, respecFor large N, r I M11I tively, whereas the number of second-order IMPs is proportional to N. In order to quantify the effect of the intermodulation distortion on the system performance we define the carrier-to-intermodulation ratio for the rth channel (CIR) as [51], [52], [58]: CIR
1 D m 4 .D111 N 2 C D21 N / C m 2 .D11 C D2 /;

(7.23)

where m is the optical modulation depth and D111 , D21 , D11 and D2 are the distortion coefficients, associated with each type of distortion. For a device with a static nonlinearity the distortion coefficient will depend only on the nonlinear characteristic of the device and so is a measure of its linearity performance. However, if the device possesses a dynamic nonlinearity, sometimes called nonlinearity with memory, which is a nonlinearity that is frequency dependent, the distortion coefficient will also depend on the specific channel frequency allocation. The direct modulation of the VCSEL corresponds to the latter case since the laser dynamics are intrinsically nonlinear. This is a result of the interaction between the carriers and photons in the laser cavity. The OCMC should fall in the first category but further work is required to confirm this. Laser and Photodetector Noise Baseline Let us evaluate now the noise expected in a conventional laser/photodiode configuration. This study must be addressed now, as this very standard configuration gives the baseline for the complete UROOF architecture. The carrier-to-noise results obtained here are considered baseline figures, and they will be checked against the developed devices in order to check their technical advantage. For a standard APD or PIN photodetector, determination of the overall system carrier-to-noise ratio (CNR) is required to take into account the relative intensity noise (RIN), receiver shot and thermal noise. The carrier-to-noise (CNR) at the receiver may then be expressed as: C CNR
1 CNRr
1 D CNR
1 R Xr ; T Xr

(7.24)

where CNRT X is the ratio of the carrier power to the noise power generated by the laser diode and CNR R X is the ratio of the carrier power to the noise power generated at the receiver. These ratios are relative to a specific channel r. For simplicity of notation we will drop the subscript r. The ratios are given by: 2B Ð RIN C C I R
1 m2

(7.25)

Ž 1 2.mg I 2 / ; D .2eI g 2 F C hIr2 i/B

(7.26)

CNR
1 TX D

CNRRX

where g is the APD gain, Ir2 is the receiver noise spectral noise density A2 /Hz, I is the primary dc photocurrent, F D gx is the excess APD noise factor, e is the electron charge, B is the signal bandwidth.

Short-range Wireless Communications

307

The total CNR is then written as: CNR D

hIr2 iB C g 2 I 2 R I N Ð B C

1 2 2 2 2m I g 2eB I g 2 F

C 12 g 2 I 2 .m 6 C1 C m 4 C2 /

;

(7.27)

with: C1 D D111 N 2 C D21 N

(7.28)

C2 D D11 N C D2

(7.29)

CNR may be bipartitely maximized in m and g for a given I , equivalent to balancing the total contributions of signal-independent and signal-dependent noise terms [52]. The maximum CNR for an optimum modulation depth, m opt , and a given I , obtained by differentiation of Equation (7.27), is: 4 2 [D111 N 2 C D21 N ] C 2m opt [D11 N C D2 ]; CNR
1 .m opt / D 3m opt

(7.30)

or, conversely, it may be determined by specifying the required CNR:

2 m opt D


C2 C

q Ž C22 C 3C1 C N R

(7.31)

3C1

The optimum APD gain is given by the usual expression:  gopt D

hIr2 i eI x

½1/ .2Cx/ (7.32)

The maximum CNR for an optimum OMD and APD gain is then: CNR.m opt ; gopt / D

1 2 2 2 2 m opt I gopt 2 2 K 3 B{hIr2 i[1 C x ] C gopt I2RI N}

(7.33)

where: K3 D

2 3m opt C1 C C2

(7.34)

2 2m opt C1 C C2

The necessary primary dc photocurrent to achieve CNR is readily obtained by combination of Equations (7.32) and (7.33): 

IAP D

ex D hIr2 i

½

1 1Cx

"

C N R.m opt ; gopt /K 3 BhIr2 i.1 C 2=x/ 1 2 2 m opt


R I N Ð K3 B

# 1Cx=2 1Cx

(7.35)

308

Technologies for the Wireless Future – Volume 3

For a PIN receiver the gain and the excess noise factor are both unity; the required photocurrent to achieve CNR(m opt / is then: IP I N D with: W D

eC

p e2 C W hIr2 i ; W

2 m opt

2C N R Ð K 3 B


RI N

(7.36)

(7.37)

In practice we are interested in obtaining the receiver sensitivity for a desired CNR and a specific number of channels with the laser biased at a certain point. These last two parameters, number of channels and laser bias current, will determine the levels of distortion at the laser output and are included in the analysis through the distortion coefficients D. Once the system parameters are specified, the distortion coefficients are determined and from these the optimum modulation depth for the desired CNR follows from Equation (7.31). Depending on whether the system uses a PIN or an APD, the receiver sensitivity IPIN(APD) times the photodetector responsivity (R0) is readily obtained from Equation (7.35) and (7.36), respectively. In cases where only either second- or third-order intermodulation effects are significant, considerable simplification of the previous equations is possible. If only second-order distortion is significant, the optimum modulation depth (Equation (7.31)) may be reduced to: m opt D [2C N R.D11 N C D2 /]
1/ 2

(7.38)

Similarly, for third-order distortion effects: m opt D [3C N R.D111 N 2 C D21 N /]
1/ 4

(7.39)

Note that, in accordance with the previous definition for the distortion coefficients, these equations are bias- and frequency-dependent: as the frequency subcarrier allocation gets closer to the resonance frequency the distortion coefficients reach maxima, which forces m opt to correspondingly lower values. Clipping Distortion Implication RIN and intermodulation distortion establish a limit to the maximum obtainable optical modulation depth or carrier-to-noise ratio. It has been shown that clipping effects determine a fundamental limit to the total modulation depth [15,49]. Thus, even if the distortion coefficients are all zero a maximum CNR still exists. This limit imposed by clipping distortion will be determined based on the model by Saleh, which has been revised in [45] and shown to agree p with simulation results to within 2 dB for values of the effective modulation depth, ¼ D m N =2, greater than 0.25. In Saleh’s analysis, the nonlinear distortion is calculated by approximating the sum of multiple randomly-phased subcarriers as a Gaussian probability density of the amplitude. The total nonlinear distortion is then assumed to be proportional to the power in the Gaussian tail that falls below zero and to

Short-range Wireless Communications

309

be distributed uniformly over all channels. The total mean square value of the clipped portion of I .t/ is then [49,45]: . ¼5 1
1 2¼2 2 i D p e (7.40) hIcli p 2³ 1 C 6¼2 Additionally, a calculation by Mazo [41] indicates that over a wide range of ¼ most of the distortion is thrown out of the transmission band. Thus, the carrier-to-clipping distortion noise ratio, CCR, becomes: s . 2 ¼3
1 2¼2
1 e ; (7.41) CC R D 3 ³ 1 C 6¼2 where 3 represents the fraction of the clipping distortion power which falls in the transmission band, which for the US-CATV plan (50–500 MHz) takes the value 3 D 1/2 [41]. Rewriting Equation (7.27) to include clipping noise, we obtain: r . 2 ¼3 KN
1 2¼2
1 e C 4¼4 D111 C 2¼2 D11 ; (7.42) CNR D 2 2 2 C 3 ¼ I g ³ 1 C 6¼2 where K includes all the other noise contributions: K D hIr2 iB C g 2 I 2 R I N Ð B C 2eB I g 2Cx

(7.43)

The CNR at optimum ¼ and APD gain, ¼opt and gopt respectively, now becomes: C N R
1 .¼opt / D 3

e

. 2
1 2¼opt

p

2³

3 5 C 18¼opt ¼opt C 11¼opt 2 .1 C 6¼opt /2

4 2 C 12¼opt D111 C 4¼opt D11 (7.44)

and gopt remains unchanged. The relation between the photocurrent I and ¼opt is determined from the solution of the following equation: 2 2 2 I 2 R I N Ð B Ð N ] D ¼opt gopt I2 ð [hIr2 i.1 C 2=x/B N C gopt " # 2 3 5 C 6¼opt e
1=2¼opt ¼opt C 9¼opt 4 2 C 8¼opt D111 C 2¼opt D11 ð 3 p 2 .1 C 6¼opt /2 2³

(7.45)

It may be advantageous to accept some distortion rather than constraining the total modulation depth to 100% [42]. An adequatepmeasure of the total effective modulation depth is the rms modulation, defined as mrms D m N [42]. Latency The system latency, Td , is given by the maximum time required by any signal received to propagate by the complete topology, i.e. to reach the most far end, IUT. For the latency calculation, it is considered the time delay between the IUT i-node and the k-node. We can calculate the accumulated time delay by: Td D Td

RX

C Td V C S E L C T p C Td OC MC C Td T X ;

(7.46)

310

Technologies for the Wireless Future – Volume 3

where Td
R X and Td
T X stand for the delay (phase response) of the RX and TX antenna respectively, Td
V C S E L stands for the VCSEL group velocity delay, Td
OC MC stands for the group velocity delay and T p stands for the propagation time along the fibre path. TFeRx and TFeT x stand for group velocity delay of the IUT front-end amplifier. The sum given by Equation (7.46) must then be under this upper limit. Electrical/Optical Conversion Gain Let us calculate now the theoretical electrical-optical conversion gain. We evaluate this parameter from a high-level system point of view. This model will also be adequate for an E-EAT implementation. The electrical power is converted to optical power via the VCSEL efficiency ¾ [W=A] p according to the following relation: Pel D I 2 R, and consequently: I D Pel =R. After the conversion, we have: r Pele Popt D ¾ I D ¾ R Expressed in dB, this has the form: r Popt [d Bw] D 10 log.¾ I / D 10 log ¾

Pele R

!

  1 ¾ Popt [d Bw] D 10 log p C Pele [d Bw] 2 R



¾ D 10 log p R



1 C 10 log.Pele / 2 (7.47)

The 1/2 factor is not a problem, as in the OCMC the opposite effect is produced, and it will compensated. From the efficiency values above, the expected E/O conversion loss can be calculated as follows: 1 For example, for ½opt D 850 nm; xi D 0:3; we havePopt [d Bw] D
15:23 C Pel [d Bw]: 2

Pele [W ]

v I

Popt [W ]

Rs = 100 Ω

Figure 7.31 System-level generic electrical-optical converter (VCSEL/E-EAT) model

Pele [W ] Rs = 100 Ω

Popt [W ] I

Figure 7.32 System-level generic photoreceiver (APD/PIN/OCMC) model

Short-range Wireless Communications

311

Table 7.11 Comparison of 100% conversion efficiency for VCSEL at UROOF wavelengths. Wavelength [nm]

100% Conversion efficiency [W/A]

Actual conversion efficiency [W/A]

G(Laser) [dB]

850 1310 1550

1.46 0.947 0.8

0.3 0.28 0.1


13.7
10.58
18.061

For ½opt For ½opt

D 1310 nm; xi D 0:28; we obtain Popt [d Bw] D
15:53 C D 1550 nm; xi D 0:1; we obtain Popt [d Bw] D
20 C

1 Pel [d Bw]: 2

1 Pel [d Bw]: 2

O/E Conversion Gain The same analysis can be applied to calculate the optical-to-electrical conversion gain. The electrical power is converted to optical power via the OCMC efficiency
[A=W ] as follows: p I D
Popt D Pel =R 2 We get Pel D R
2 Popt . The same relation expressed in dB has the form: 2 Pele [d Bw] D 10 log.R
2 Popt / D 10 log.R
2 / C 2 Ð 10 log.Popt /

Pele [d Bw] D 10 log.R
2 / C 2 Ð Popt [d Bw]

(7.48)

This confirms that the 1/2 factor in Equation (7.48) is not a problem, because in the OCMC there is a 2 factor that compensates it. Popt [d Bw] D 18 C 12 Pel [d Bw] Table 7.12 reports actual data we can expect for photodiode efficiency [DSC-R402HR]. For example, for ½opt D 850 nm,
D 0:25, we obtain: Popt [d Bw] D 7:96 C and for ½opt

1 Pel [d Bw]; 2

D 1310 nm and 1550 nm:
D 0:8

Popt [d Bw] D 18 C

1 Pel [d Bw] 2

Table 7.12 Comparison of 100 % conversion efficiency for receiver at UROOF wavelengths. Wavelength [nm]

100% Conversion efficiency [A/W]

Actual conversion efficiency [A/W]

G(PIN) [dB]

850 1310 1550

0.685 1.055 1.249

0.25 0.8 0.8


8.75
2.403
3.869

312

Technologies for the Wireless Future – Volume 3

p(t)

1st cluster

a00 g = ray

a10

a01

Γ = cluster decay factor

th

a20

a11 a30

T0

decay factor

a0

T1

cluster

ak

T

i Figure 7.33 Power delay profile of S-V model; here Þk;` Ð Ð Ð gain coefficients of multipath channel

Modeling the Wireless UWB Indoor Channel Models for propagation through typical WPAN of UWB signals were developed in IEEE802.15.3a [29] based on a modified Saleh-Valenzuela (S-V) model. Like the original S-V double-exponential decay power delay profile model, multipaths arrive in ‘clusters’ of rays. Cluster arrival times and ray arrival times within the cluster are modeled by statistically-independent Poisson processes. A recent survey of UWB channel models is provided in [24]. Modeling the Propagation through the Fibre The dispersive nature of the fibre media is characterized by: ž frequency-dependent refractive index n.¹/ ž absorption coefficient Þ.¹/ ž phase velocity v and propagation constant þ.¹/. Since a pulse lightwave is the sum of many monochromatic waves, each component is modified differently by the medium. As a result, the pulse is delayed and broadened (‘dispersed in time’). The simplest model for a dispersive fibre [50] assumes a pulsed plane wave propagating along the z axis of a linear, homogeneous and isotropic media with known n.¹/; Þ.¹/ and propagation constant þ.¹/ D 2³ ¹n.¹/=c0 . The complex wave function is given by: U .z; t/ D A.z; t/ exp[ j .2³ ¹0 t
þ0 z/]; where ¹0 is the central frequency, þ0 D þ.¹0 / is the central wave number and A.z; t/ is the complex envelope of the pulse. We assume that A.0; t/ is known and wish to determine A.z; t/ at the distance z in the medium. Linear System Description A transfer function linear-system description that relates A.z; t/ to A.0; t/ is based on the following. Let A.0; t/ D A.0; f / exp. j2³ f t/ be a harmonic function with frequency f so that the wave is monochromatic with frequency ¹ D f C ¹0 . Then A.z; f / D A.0; f /H . f / where the transfer function of the linear system is given by: ¦ ² 1 H . f / D exp
Þ. f C ¹0 /z
j[þ. f C ¹0 /]z 2

Short-range Wireless Communications

313

Hence, the complex envelope of the pulse is extracted using the Fourier transform: A.z; t / D F


1

Z1 { A.z; f /} D

A.z; f / exp. j2³ f t/d f
1

This can be equivalently expressed as the convolution A.0; t/ with h.t/ D F
1 {H . f /} as follows: Z1 A.0; t0/h.t
t0/dt0: A.z; t / D
1

Slowly Varying Frequency-dependent Approximation A.z; t/ is slowly varying with respect to the central optical frequency ¹0 and its Fourier transform A.z; f / is a narrowband function of f with width 1¹ − ¹0 . It can be assumed that Þ.¹/ D Þ; that is, absorption coefficient is not frequency dependent. þ.¹/ varies only slightly with frequency. Hence, it can be approximated by three terms 1 2 d2þ Taylor series þ.¹0 C f / ³ þ.¹0 / C f dþ d¹ C 2 f d¹ 2 . The approximate transfer function can be now presented by: H . f / ³ H0 exp.
j2³ f −d / exp.
j³ D¹ f 2 /; 1 where H0 D e
Þz=2 ; −d D z=v; v1 D 2³ where D¹ is the dispersion coefficient.

dþ d¹ ; v

is the group velocity D¹ D

1 d2þ 2³ d¹ 2

D

d d¹


1Ð v ,

Notes 1. When the dispersion is negligible, the third term of propagation constant (function of f
2 / isÐ small and hence the transfer function contains H . f / D H0 exp.
j2³ f d / D exp
Þz 2 exp.
j2³ f d /. The first term is an attenuation factor and the second term is a delay, −d D z=v. In this case the original pulse passed through a linear system (with losses) and is not distorted. Hence, A.z; t / D e
Þz=v A.0; t
−d /. 1 1 ð 2³ 2. For an ideal linear system without losses, Þ D 0 and v1 D 2³ c D c . Hence, v D c, and the pulse envelope travels at the free space light velocity c. 3. In general, the group velocity depends on frequency, so that different frequency compoz is frequency nents undergo different delays. Hence the quantity −d D z=v.¹/ D 2³.dþ=d¹/ dependent, and the pulse is distorted (broadened and attenuated). If a pulse has a spectral width ¦¹ (Hz) then after z metres the pulse width can be approximated by ¦tau D jD¹ j¦¹ z. 7.3.6 Technology Trends to be Explored and Summary The rapid growth of the next-generation mobile communication infrastructure and the hybrid radio-over-fibre systems for personal area networks creates new challenges both in signal processing technologies and in the photonic components. In particular, the successful transmission of UWB signals requires the largest possible bandwidth up to 100 GHz and, at the same time, the mobility of wireless techniques. These requirements can be satisfied by using a UROOF technology combining the advantages of optical signal transmission through the low-loss optical fibres and the E/O and O/E signal conversion carried out by fast, efficient

314

Technologies for the Wireless Future – Volume 3

and low-cost modulators and detectors. As was argued above, the rapidly developing (in the last decade) microwave photonics is the most promising candidate for the realization of some seemingly contradictory conditions. The processing of microwave signals at high frequencies up to 60 GHz, and even above that level, can be successfully realized in an optical domain. To that end, the high-level integration technologies of passive and active optical and electronic components, multichip modules, MMIC, etc. are needed. One of the existing obstacles for the integration of optical and microwave components in the same substrate in the framework of the common functional system is the incompatibility of the GaAs-based optical components technology and the Si components microelectronic technology. As previously shown, while various photodetectors such as PIN, metal-semiconductormetal (MSM) and APD can be used for high-speed applications, each of these devices has certain drawbacks which make them rather unsuitable for use in UWB-over-fibre communications. Lateral PIN and MSM devices are simple to fabricate, but are characterized by low responsivity at wavelengths around 850 nm. APD devices, on the other hand, require relatively high-bias voltages and have increased noise. We believe that the most promising perspective for the low-cost integrated opto-electronic system fabrication is a novel SiGe technology. The devices based on nanostructures such as SiGe MWQ lasers, two dimensional (2D) photonic crystal optical waveguides, high-speed optoelectronic receivers, SiGe/Si phototransistors and SiGe/Si quantum well waveguide EAMs can be successfully integrated with microstrip and coplanar guiding systems for microwaves, as well as with optical fibres. They can achieve exceptionally efficient performance, low cost and design optimization. Other aspects of UROOF technologies that remain for further research are: ž Study of best match UWB technology for UROOF converged system. ž Cost optimization of microwave photonics components and integrated silicon-photonic devices to enable low-cost residential applications. ž MAC enhancements and solving best multiple access for UROOF various multicell architectures. ž Coexistence of UROOF technologies over the combined wireless-fibre channels with legacy inherently digital optical communications.

7.4 Work in Progress 7.4.1 High Data Rate Wireless Communications in the Unlicensed 60 GHz Band2 7.4.1.1 Introduction To satisfy the rapidly growing demand for communication bandwidth, the 60 GHz band becomes more and more attractive to many companies and research institutions. A number 2

This research is partly funded by the German Ministry of Education and Research (BMBF) under reference number 01BU371.

Short-range Wireless Communications

315

of countries have made available 5–7 GHz of unlicensed bandwidth. Such a large amount of unlicensed bandwidth has never before been available. It allows wireless transmission at enormous data rates of up to 10 Gbps. Forthcoming systems operating in the 60 GHz band are believed to meet the expectations which were originally placed on UWB systems. In contrast to UWB systems operating in the 3–10 GHz frequency bands, there are no hard power limitations at 60 GHz. This is because the spectrum doesn’t have to be shared with other wireless technologies. In most regions 10–20 dBm of transmit power are allowed with an equivalent isotropic radiated power (EIRP) of around 40 dBi. This allows high-data rate transmission well beyond 1 Gbps over a distance of about 10 m. The main applications to be targeted by such systems are uncompressed HDTV transmissions. For high resolution modes, data rates in excess of 3 Gbps over a distance of more than 5 m are required. For these applications, quality of service (QoS) is an important requirement. Fast downloads from data kiosks over very short distance of about 1 m are also targeted. These applications are very cost sensitive and would benefit from even higher data rates. Nowadays, cost-efficient technologies are available to implement RF front end circuits operating in the 60 GHz range. In the past, circuits based on costly GaAs or InP technologies had to be deployed. Recently, a number of silicon-based solutions were published. For instance, IBM has demonstrated a complex 60 GHz analogue front end which is suitable both for single-carrier and multicarrier modulation schemes [69]. The developed circuits were implemented on a high-performance SiGe BiCMOS technology. Another 60 GHz SiGe-BiCMOS front end solution, specifically designed for OFDM modulation, was presented by IHP [79]. Recently, the company SiBEAM released information on a 60 GHz CMOS solution [70]. As the 60 GHz front end technology is progressing very rapidly, different standardization activities are also gaining speed. In particular, the IEEE 802.15.3c standard has made significant progress and may get release in mid-2009 [71]. In parallel, there are 60 GHz standardization activities within ECMA. Recently, a study group ‘Very High Throughput’ (VHT) under the working group IEEE802.11 was formed. One possible physical layer mode defined by this study group will most probably operate in the 60 GHz band [72]. In addition to this, some companies of the WirelessHD consortium are pursuing the definition of an industry standard. This is apparently based on already existing prototypes [73]. 7.4.1.2 60 GHz Frequency Regulation and Standardization Frequency regulation for the 60 GHz band in Europe is in progress. Currently, there is only 500 MHz of bandwidth available as an ISM band. Internationally, there is at least 5 GHz of overlapping bandwidth available in many countries, as shown in Figure 7.34. It is hoped that Europe will make a similar amount of bandwidth available in the near future. In some documents of ETSI, the bandwidth from 63 to 64 GHz is earmarked for road transport and traffic telematics (RTTT) applications. A similar bandwidth allocation is apparently planned in Japan. The main RTTT applications are car-to-car (C2C), car-to-roadside and roadside-to-car communication. Furthermore, in Japan the maximum bandwidth to be used by a transceiver must not exceed 2.5 GHz.

316

Technologies for the Wireless Future – Volume 3

63

64

RTTT

Europe?? 59.4

62.9

Australia

Canada

Japan

USA

57

58

59

60

61

62

63

*) Frequency Regulation in Europe is in Progress

64

65

66

f [GHz]

Figure 7.34 Bandwidth in the 60 GHz band in different countries; RTTT D road transport and traffic telematics

For the 60 GHz frequency regulation in Europe, a proposal from five organizations is currently being discussed. These companies are: Atheros Communications (USA), Motorola, OFCOM (UK), RFI (UK) and Intel Corporation. The original timeline for the 60 GHz ISM band regulation in Europe is as follows: 1. 12/2005: Start of work date. 2. 06/2006: Target date for approval of deliverable by TC BRAN. 3. 09/2006: Target date for approval by ERM RM. Obviously, there is some significant delay, since to date the regulation is not completed. However, it is assumed that eventually the frequency regulation in Europe will make available a similar amount of bandwidth to that in other places. IEEE802.15.3c Standardization Roadmap Currently, the 60 GHz standardization within IEEE 802.15.3c (TG3c) is in progress. The status of this work is as follows: ž 26 individuals have issued their intent to contribute to this 60 GHz standard. ž The proposal presentation took place at the IEEE meeting in May 2007. Sixteen standard proposals were presented. ž In TG3c, the minimum data rate to be supported by a terminal was increased from 1 Gbps to 2 Gbps. ž The down-selection process started in July 2007 and was completed in November 2007 by a confirmation vote, with 87% of the of IEEE802.15 members voting in favour of the baseline standard document. ž Between November 2007 and May 2008 the comment resolution took place. ž The first official standard draft D0 was released in June 2008 with the first Letter Ballot starting at the same time. ž The final approval of the standard can take place in September 2009.

Short-range Wireless Communications

317

2080 MHz

520 MHz

1040 MHz

Ch B2

Ch B1

Ch B3

Ch B4

A1 57

58 58.24 GHz

Ch B5

Ch B6

Ch B7

A2 59

60 60.32 GHz

Ch B8

Ch B9

Ch Ch Ch B10 B11 B12 A4

A3 61

62

63

62.40 GHz

64

65

66

fGHz

64.48 GHz

Figure 7.35 Proposed channel plan for use of the 60 GHz band with wideband channels A1–A4 and narrowband channels B1–B12

The ‘channelization’ of the 60 GHz band as discussed in the IEEE 802.15.3c group will most probably be based on about 2 GHz-wide wideband channels (WBC), labeled A1–A4 in Figure 7.35. Furthermore, some narrowband channels (NBC), labeled B1–B12, which can be used as a return channel, audio data or for control information are suggested. In one proposal, the narrowband channels are 500 MHz wide and are distributed as shown in Figure 7.34. The frequency generation for the centre frequencies in Figure 7.35 is based on a 26 MHz crystal. Since this crystal is also widely used for cellphones, the component cost of future integrated 60 GHz front ends can be significantly reduced. The advantages of the proposed channelization in Figure 7.35 are as follows: ž A 4x factor between wideband channels (WBC) and narrowband channels (NBC) gives more system flexibility than a 2x factor: ž The 500 MHz hardware (H/W) can be significantly cheaper than the 1 or 2 GHz H/W. ž There are a reasonable number of NBCs available in all regions (i.e. 9). ž Implementation of the common mode using 500 MHz channels requires less costly H/W. ž The utilization of the available spectrum is very good (75%). ž There are similarities with the W-HD LDR channels. W-HD will probably insist on LDR channels anyway. ž No overlap problems between NBC and WBC. ž The NBC centre frequencies can be easily derived from a 26 MHz crystal. The same principle is possible for centre frequencies based on a 19.2 MHz crystal. ž There is no overlap with UWB data rates and applications since UWB is power limited and, hence, the UWB data rate and/or distance are much lower. The currently published timeline for standardization as of May 2007 is as follows: 1. 01/2007: Call for proposals was issued at the IEEE London meeting. 2. 05/2007: Submission of final proposals due 7 May 2007; presentation of proposals took place in May 2007. 3. 07/2007: Beginning of down-selection procedure. 4. 03/2008: Standard completed.

318

Technologies for the Wireless Future – Volume 3

UM1 Uncompressed Video Streaming U1/U3

TV Point-to-point

TV or Monitor

PC, uMPC, Set top Box (STB)

Figure 7.36 Graphical representation of Usage Model 1 (UM1), ‘Uncompressed HDTV Video Streaming’ (source: IEEE802.15.3c Usage Model Document)

7.4.1.3 Applications and Market Expectations Wireless HDMI Currently, the most dominant application is the replacement of HDMI cables. The costly and difficult-to-handle cables are a weak point of recent HDTV sets. They limit the distance between set-top box and TV screen to about 3 m and are difficult to hide due to their diameter. Therefore, a wireless solution would significantly improve user acceptance. The required data rates reach up to 3.5 Gbps. The distance for the HDR link is up to 5 m. An LDR return channel is used for control information, as depicted by the dashed line in Figure 7.36. Since there are clear benefits in replacing the HDMI cable with a wireless solution, this appears the most important application to date. Connections from laptops to high-resolution beamers are included in this context. Data Kiosk Another important application is the file download from a data kiosk. Customers can download their favourite DVD or audio collection from such kiosks to their portable storage device. Such storage devices can be mobile phones, MP3 players or PDAs. The kiosk makes returning physical DVD media to a video store redundant. Data kiosks can be placed in shopping malls, petrol stations or public areas of airports and railway stations. Here the communication distance can be lower, at about 1 m. However, to allow the download of a 5GB DVD within a reasonable time, the required data rate can be as high as 5–10 Gbps. Next-generation WLAN For corporate computer networks, Gigabit Ethernet becomes more and more standard. For wireless indoor links, customers do typically expect the same data rate and service as with wired connections. Therefore, Gbit WLANs will be required soon. Since IEEE 802.11n-based systems and UWB systems cannot deliver this data rate, industry planners are more and more focusing on 60 GHz technology. The delay in the standardization of IEEE 802.11n and UWB has also slowed down their market entry. It appears that 60 GHz technology can deliver the services which customers originally hoped to receive with IEEE 802.11n and UWB technology. As a result, a new study group ‘Very High Throughput’ (VHT) was formed in working group IEEE 802.11. One possible physical layer under consideration by the study group VHT is supposed to operate in the 60 GHz band.

Short-range Wireless Communications

319

UM5 Kiosk File-downloading

U7/U9

U7/U9

STB, Game Consol

Movie and Game Kiosk

Mobile Storage Device, PDA

Figure 7.37 Graphical representation of Usage Model 5 (UM5) ‘Kiosk File Downloading’ (source: IEEE802.15.3c Usage Model Document) Table 7.13 Comparition of single-carrier and OFDM for 60 GHz.

Spectral efficiency PAPR Multipath robustness Out of Band Radiation Complexity of baseband processor Complexity of AFE

OFDM

Single Carrier

Single Carrier-FDE

CC -CC CC -

C -CC

C C C C C

7.4.1.4 System Architectures Multicarrier vs Single-carrier Modulation Different modulation schemes have different advantages and drawbacks. In Table 7.13. It can be seen that the complexity of the OFDM system is quite high. The main disadvantage of OFDM modulation is the high peak-to-average power ratio (PAPR). This requires a relatively high back off of 6–10 dB for the power amplifier, which in turn does increase the power dissipation. However, for fixed and nomadic applications which are not sensitive to power dissipation this is no real limitation. On the other hand, OFDM has a very high spectral efficiency, is robust against multipath effects and is spectrally well behaved, generating very little out-of-band radiation. Single-carrier modulation schemes with frequency domain equalization (FDE) mitigate the disadvantages of single-carrier schemes, but do require much higher complexity for the digital baseband processing than single-carrier techniques. Pure single-carrier schemes without frequency domain equalization suffer from multipath fading and are only suitable for ultra-short distances and line-of-sight (LOS) communication. Therefore, it seems advantageous to provide both single-carrier schemes and OFDM, depending on the intended application. For cost sensitive, short-range, line-of-sight communication systems, single-carrier modulation schemes appear best suited. For nomadic or fixed systems with no power dissipation constraints and medium range, OFDM seems a good choice. We hence assume that a forthcoming 60 GHz standard (IEEE 802.15.3c) will provide both options.

320

Technologies for the Wireless Future – Volume 3

Front End Architectures There are three different principle options for RF front end implementation. First, a direct conversion front end converts the RF signal in a single step from baseband to RF frequency and vice versa. Direct-conversion front ends lend themselves to monolithic integration due to the avoidance of image filtering. However, DC offset and LO leakage to the antenna prevent a widespread acceptance to date [74]. A variant of the direct conversion system is the low-IF approach. Here the signal is not directly converted to the base band but to a low IF. The data is directly sampled at this low IF and the I/Q separation occurs in the digital domain. For this approach, prohibitively high sampling rates of the A/D converters are required. Therefore, low-IF architectures for 60 GHz systems are currently not in the focus of development. Second, heterodyne transceivers use a two-step approach with an intermediate frequency. The transmit signal is converted from baseband to an intermediate frequency (IF) using a low-frequency local oscillator LO2, and from there to RF using a high-frequency local oscillator LO1, and vice versa. For such a super-heterodyne architecture two important questions must be answered: 1) What IF results in the best RF performance and the lowest cost? 2) In order to select different RF sub-bands between 57 and 64 GHz, which frequency should be tuneable: LO1, LO2 or both? Different IFs have been suggested in the literature: 5 GHz [79], 9 GHz [78] and 12 GHz [81]. The advantage of 5 GHz is compatibility to the 802.11a standard, facilitating the design of a 60 GHz system [72], especially when combined with a 5 GHz backup system. A higher IF facilitates easy on-chip image rejection by the integrated low-noise amplifier (LNA) and PA, which turns out to be difficult (but not impossible) for an IF of 5 GHz. Tuning of LO1 by changing the PLL divider ratio is not recommended, since the PLL divider ratio would be excessive, resulting in a prohibitively large phase noise contribution of the crystal reference as it appears at the PLL output. A programmable LO2, as suggested in [82] and realized in [79], allows a straightforward solution compatible to 802.11a. Alternatively, LO2 can be derived from LO1 by frequency division, as suggested in [81]. Architectures based on this approach are termed ‘sliding IF’. In this case, LO1 and LO2 must be tuned simultaneously. This could be done, for example, by a low-cost low-phase noise fractional-N synthesizer used as reference for both LO1 and LO2. 7.4.1.5 Challenges of 60 GHz Systems Systems design for the 60 GHz band is facing a number of major challenges. Many of these issues are related to the parasitic effects of circuits at such high frequencies. In addition to this, the sheer data rate which is targeted is also demanding for the digital circuitry. In detail, the following main challenges can be identified. The Antenna Design and Packaging for 60 GHz Due to the short wave length of about 5 mm, the antenna can be comparatively small. This makes on-chip or in-package antennas possible. Preferable antenna structures are patch antenna, phased array patch antenna, dipole and Vivaldi antenna. Even though for small volume production the Vivaldi antenna appears the cheapest, it is relatively big and not suitable for beamforming. For low-performance

Short-range Wireless Communications

321

applications requiring a small form factor, a dipole antenna or a single patch antenna are preferable. If required, they can be integrated in the same package as the 60 GHz transceiver chip. To improve the link budget, patch array antennas are useful. For beamforming, phased array patch antennas can be used. Some applications may also benefit from circular polarized antennas, even though their implementation is more costly. In terms of assembly techniques, flip-chip packaging using a chip-on-board technology appears most promising. This way, parasitic inductances and capacitances can be kept to a minimum. In order to integrate the transceiver chip together with the antenna, low-temperature cofired ceramics (LTCC) material is often used. Techniques like ribbon bonding and wire bonding are possible too. Using cavities for the transceiver chips, the length of bondwires can be kept to a minimum. 60 GHz RF Transceiver Circuitry For the RF circuitry, the main challenges are the limitation of phase noise, I/Q mismatch and nonlinear distortions. In order to improve the efficiency of the power amplifier, the peak-to-average power ratio (PAPR) of the transmit signal should be kept low. For the purpose of low component cost, circuits based on silicon technologies are preferable. In particular, SiGe-BiCMOS technologies and below-90 nm RF-CMOS technologies appear suitable. Data Converters In order to achieve a high modulation index, high-resolution data converters are required. For a signal bandwidth of 1 GHz, this results in a very demanding specification for the A/D and D/A converters. The sampling rate has to be in the order of 2 Gsps and a resolution of 8 bit is required for a 64-QAM modulation. Digital Circuit Implementation and Power Dissipation The digital baseband processor for systems operating at 2 Gbps is very demanding too. Complex operations such as FFT and IFFT have to operate at very high speed. The same applies to the forward error correction (FEC) circuitry. One possible solution is the parallelization of algorithms. As with the baseband processor, the MAC processor has to handle a very high throughput and has to operate at low latency. To allow mobile applications, the integral power dissipation of RF circuits, data converters, baseband processor and MAC processor has to be kept to a minimum. Low power system and circuit design techniques have to be used. Battery-powered mobile applications cannot afford more than 1W of power dissipation when operating at full data rate. Advanced power-down, sleep modes and wake-up schemes must be deployed for these applications. In the following section, a brief summary of and outlook for estimated system parameters and power dissipation figures for the main system components are given. 7.4.1.6 Performance Estimation and Outlook On the basis of a current FPGA implementation of the OFDM baseband processor, the chip area can be estimated. A 60 GHz front end demonstrator in 0.25 µm SiGe BiCMOS

322

Technologies for the Wireless Future – Volume 3

Table 7.14 Estimated power dissipation of a 60 GHz OFDM transceiver in 2008. 2008

Transmit

Receive

MAC Processor Baseband Processor Data Converters Analogue Frontend Power Amplifier Total (continuous)

200 mW 200 mW 100 mW 200 mW 150 mW 850 mW

200 mW 350 mW 150 mW 200 mW 10 mW 910 mW

technology is used for estimating and scaling the chip area of the analogue front end. Furthermore, we use the following assumptions: ž ž ž ž

2 Gbps data rate in four parallel data streams 500 MHz digital clock 65 nm digital CMOS process for MAC and baseband processor 130 nm analogue SiGe-BiCMOS process for analogue front end and data converters: ž MAC processor: 10 mm2 (ca. 10 million gates) ž baseband processor: 15 mm2 (ca. 15 million gates) ž data converters: 10 mm2 ž analogue front end (incl. PA): 6 mm2 ž size complete transceiver PCB: CardBus or PCMCIA card module ž patch array antenna (3 ð 4 elements): 30 mm x 40 mm x 2 mm (LTCC) ž Vivaldi antenna (10 dBi gain, 30 grd): 30 mm x 30 mm x 0.5 mm (PCB material).

7.4.1.7 Power Dissipation Table 7.14 is an estimate of the total power dissipation at a data rate of 2 Gbps (65 nm CMOS digital; 130 nm analogue SiGe) for implementations in 2008. It is assumed that the power dissipation of the digital circuitry is reduced to 50% with the transition to the next technology node within two years. Due to advances in circuit design techniques and in process technology, the power dissipation of the analogue modules will reduce to approximately 75% every two years. Based on technology scaling, we predict that in 2010 the power dissipation figures for 60 GHz OFDM transceivers shown in Table 7.15 can be achieved. 7.4.1.8 Conclusions The growing need for wireless transfer rates of several gigabits per second creates a big technical challenge. It can only be tackled by increasing the spectral efficiency of the transmission or by using frequency resources in the millimetre wave band, which have not been used for mobile communication so far. Currently, there is lively discussion about using the 60 GHz

Short-range Wireless Communications

323

Table 7.15 Power dissipation figures for 60 GHz OFDM transceivers. 2010

Transmit

Receive

MAC Processor Baseband Processor Data Converters Analogue Frontend Power Amplifier Total (continuous)

100 mW 100 mW 75 mW 150 mW 125 mW 550 mW

100 mW 175 mW 100 mW 150 mW 10 mW 535 mW

band for applications that require high data rates. Examples are Gigabit-WLAN and wireless multimedia access. For instance, downloading a DVD on a 5 GB memory stick at a net data rate of 5 Gbps will take only 8 seconds. These applications offer a considerable market potential, but they also call for extensive research work. A number of developments have recently generated significant interest in mm-wave communications. First, the FCC has made available 7 GHz of unlicensed bandwidth in the 60 GHz band. Never before has such a huge amount of bandwidth been allocated for wireless communications. This allows for data rates of several Gbps. Second, a standard for mm-wave communications is being developed within Task Group IEEE 802.15.3c. This standard will be released in mid 2008. It will specify PHY and MAC layers for high-data rate systems operating in the 60 GHz band. Third, technological developments such as the implementation of 60 GHz transceiver circuits with cost-efficient silicon-based technologies, as well as the progress in memory technology, enable the development of ultra high-data rate communication systems. 60 GHz technology enables new services and user applications, which cannot be made available with present solutions. 7.4.2 Ultra Wideband Communication 7.4.2.1 Introduction Ultra wideband radio (UWB) is, as illustrated in the last releases of the WWRF Book of Visions (BoV), a physical transmission technique suitable for all kinds of applications, i.e. for communication, localization, imaging, and sensing. Given the strong power emission constraints imposed by the regulatory bodies in the US and the further, more restrictive rules recently released in Europe, UWB has emerged in the communication domain as a particularly appealing transmission technique for applications requiring either high bit rates over short ranges or low bit rates over medium-to-long ranges. The high-bit rate/short-range case involves, for instance, wireless personal area networks (WPANs) for multimedia traffic, cable replacement such as wireless USB, and transmission of HDTV data over the air. The low-bit rate/medium-to-long-range case applies to long-range sensor networks such as indoor/outdoor distributed surveillance systems, non-real time data applications, e.g. e-mail and instant

324

Technologies for the Wireless Future – Volume 3

messaging, and in general all data transfers compatible with a transmission rate in the range of up to 1 Mbps over several tens of metres. Beside these communication applications, UWB signals are also well suited for location-aware applications. On the one hand, time-frequency duality implies that a UWB signal has a very high time resolution, which facilitates accurate time-of-flight and hence range estimation; on the other hand, high bandwidth leads to excellent penetration capabilities. The accurate localization ability of UWB signals facilitates a number of applications, such as noninvasive patient monitoring, personnel tracking, search and rescue operations, and through-the-wall health monitoring of hostages. Some of these areas are related to UWB imaging and sensing, which is of great interest for surface-penetrating radar, surveillance and emergency applications, medical instrumentation, nondestructive testing, microwave imaging, and many others. Their fundamental advantage over other sensors (narrowband RF, optical, etc.) comes from the huge fractional bandwidth of their sounding electromagnetic (EM) waves, which provides high spatial resolution, and the good penetration of UWB signals through materials makes a look through – or even into – nonmetallic materials feasible. In contrast to previous WWRF BoV articles on UWB, here we focus on recent advances and future perspectives of UWB technology by presenting the underlying physics, the feasible signal processing and the far-reaching protocol opportunities. The aim of this white paper is to give an overview for the state of the art, the challenges, and, if somehow justifiable, case-by-case visions for the next decade of wireless UWB. Several worldwide-renowned scientists and engineers have contributed to this survey, and an extended and more detailed overview will be published separately. This contribution is organized as follows. After a brief introduction on UWB, the underlying wireless channel is presented and open research tasks are pointed out. Next, the mono and multiband approaches are discussed, and feasible receiver structures (including synchronization) are investigated. Key topics about UWB hardware and a forecast on the PHY (physical layer) and MAC (medium access control) development follow. Furthermore, an overview of coexistence, regulation and the three related areas, namely localization, imaging and sensing, is given. 7.4.2.2 UWB for Communication In UWB communications, one of the most significant system design decisions includes how to exploit the available bandwidth (e.g. 7.5 GHz for the FCC mask) in a somehow optimum way under regulation, standardization and hardware constraints. Basically, there are two major modulation schemes (and several derivatives and combinations thereof) under discussion – the impulse radio (IR) and the multiband OFDM (MB-OFDM) approaches. In the following, we will first introduce IR, since it facilitates understanding of the impact of the UWB channel. Besides the challenges in UWB channel modeling, the benefits and limits of both modulation schemes are illustrated in order to point out the technically feasible visions. Impulse radio ultra wideband (IR-UWB) [132,133] is a promising approach for UWB wireless communications. Instead of modulating information on a carrier signal, as usual in all today’s wireless communication systems, data are transmitted using a coded series of well separated very narrow pulses, each with a duration of less than a nanosecond. The initially predicted advantages of IR-UWB over conventional communication techniques are significant: the devices are expected to be small and cheap due to fewer hardware components, the

Short-range Wireless Communications

325

foreseen data rates and number of users are large, the transmission method is robust against multipath propagation, the power efficiency is high, and security is inherently present. Besides communications, the pulse-based transmissions of IR-UWB are similar to those used in radar, and can therefore also be used for precise localization with centimetre accuracy. Hence, IR-UWB is also an enabling technology for position-aware devices, with applications such as tagging and tracking of assets and personnel. Although IR-UWB is promising in theory, and reams of contributions are published, the state of the art in IR-UWB implementation can be considered as rather immature. Many of the predicted theoretical advantages have not yet been demonstrated. The main reason is that the standard Rake receiver structure being envisioned in IR-UWB poses an enormous technical challenge. A Rake receiver consists of a bank of so-called ‘fingers’, where each finger correlates the incoming pulse with a locally generated template synchronized to a specific delay. All finger outputs are then superimposed in a possible optimal fashion, which depends on the power that is received by the different fingers. However, there are a few highly significant problems related to this receiver structure. First of all, it is not clear which template should be used. Due to the ultra wideband nature of the pulses, it is unjustified to assume that the received pulse is a superposition of several delayed and scaled copies of the transmitted pulse. In fact, it is a superposition of delayed and distorted copies (see the next section) of the transmitted pulse, where each distorted copy originates from a different propagation path with different propagation characteristics. Hence, the received pulse might be far away from a clone of the transmitted pulse. Next, the delays that are used in the different fingers should be perfectly synchronized to the main reflection paths in the environment, which is a further technical challenge. To alleviate these problems, one could consider an arbitrary template that spans the bandwidth of the signal and uses a large number of fingers with fixed delays. However, such a system is still too complex compared to the data rates that can be achieved. Further, it is unclear how the correlations with the template signal can be implemented. If such a circuit is realized in the analogue domain, then a linear filter has to be constructed that has the impulse response as its template signal – difficult if the template is not constant and hence extracted from the received signal. Other receivers that have been proposed for IR-UWB are all-digital receivers, but they require very high-rate and high-precision A/D converters, which will be for a long time yet too costly in terms of price, size and power consumption. Nevertheless, research and development is not yet finished here and suboptimum Rake receivers, such as partial Rake or selected Rake, may present a technically feasible approach [87,88]. The UWB Channel Understanding the nature of UWB requires a deep insight into the characteristics of the specific UWB propagation environment and its comprehensive and effective model. Any reasonable channel model has to be based on – or verified by – measurement results. Ideally, the measurement setup and parameter extraction technique should not have any impact on the results and the final model. In UWB measurements, the influence of the antenna on the frequency response of the system can be dominant. The most simple form of calibration (for single input single output (so-called SISO) measurements) is to record the output signal with an antenna that is placed in close vicinity, and with a line-of-sight (LOS) connection, to the transmit antenna. The output from this antenna then provides the signal waveform that would be observed in a pure LOS situation. Alternatively, the conventional back-to-back calibration together with a measurement of the

326

Technologies for the Wireless Future – Volume 3

frequency-dependent antenna patterns could be used. However, a true calibration requires that the received signal is calibrated depending on direction and frequency. The situation becomes even more challenging considering the fact that UWB antennas typically have different antenna patterns at different frequencies. Thus, a complete calibration requires the knowledge of the frequency- and direction-dependent transfer function, which should be combined with a directionally resolved measurement of all multipath components (MPCs) [134]. The deterministic description methods for UWB channels are by now well established. The correct channel description is given by: h.− / D

N X

ai i .− /  Ž.−
−i /;

(7.49)

iD1

where N is the number of multipath components, ai and −i are the complex amplitudes and delays of the MPCs, and i .− / denotes the distortion of a single pulse by the frequency selectivity of the interaction. The key problem lies in the identification of the distortion functions i .− /. Electromagnetic computations that are not based on the high frequency assumption seem to be a promising way to establish those values. Measurements of the frequency-dependent reflection coefficients and measurement or modeling of diffraction and scattering coefficients will be required as a basis of (frequency-dependent) ray tracing. The problems here do not lie too much in the fundamental formulations, but rather in ‘crunching the numbers’ and performing a number of time-consuming measurements. For the measurement evaluation, there are two main challenges: ž Generalization of the SAGE algorithm [89] to the UWB case, including the possibility of spherical (and not planar) waves. Attempts have been made first on sub-bands of the UWB spectrum, and then the results of the MPC parameters in the different sub-bands have been combined. However, this is not a completely general approach, as combining the information in the sub-bands suffers from a certain arbitrariness. ž All of the high-resolution algorithms like CLEAN and SAGE [90] assume that the pulse-shape of each single MPC is known, and only the delay, directions and amplitudes are unknown. However, this is not fulfilled in UWB – again, the distortion functions i .− / are unknown. Misestimation of these pulse shapes can lead to significant errors of all the involved parameters. For stochastic channel modeling, there are enormous challenges. The key problem arises from the scarcity of measurements currently available. While good generic stochastic models have been established [135], they are parameterized in only a few selected environments (indoor residential, indoor office, industrial), and even then the parameterization is often based on only a few (or even just a single) measurement campaigns. Thus, a lot of new measurements and evaluations will be required in the future. Other problems that are remaining are the following: ž Are distance and frequency dependence of the pathloss really separable? Up to now, this assumption has always been made, but it is only based on convenience, not on measurements. In other words, is the pathloss exponent independent of the frequency?

Short-range Wireless Communications

327

ž Is the shadowing variance independent of frequency? ž More measurements are needed so that the Erceg-Ghassemzadeh model for the pathloss (treating the pathloss exponent as a random variable, not as a constant) can be fully parameterized in different types of environment. ž Models for the angular dispersion. It can be anticipated that the angular dispersion will show a dependence on the frequency band as well as the delay range (long delays vs. small delays). Another area of stochastic channel modeling is the description of the temporal evolution of the channel. Traditionally, this has been described by the Doppler spectrum. However, the Doppler spectrum is a concept related to the WSSUS model of Bello [91]. But UWB channels are not WSSUS, so completely new description methods may be required [134]. The vision is to have a complete stochastic channel model, based on extensive measurements, for all environments of interest for UWB communications. Furthermore, we aim to develop efficient deterministic channel prediction methods. A large number of measurements will be required in the future in order to gain a better understanding, and statistically viable models, of UWB propagation channels. As a rough estimate, we anticipate that five person years of work are required for a model of each environment, if such a model is to include directional information as well as models for the temporal variability. A total of 10 environments are anticipated to be relevant for future UWB applications. It is obvious that the effort of such extensive measurements cannot be borne by a single industrial or academic entity. It would therefore be advisable to have a coordinated effort by various interested parties. With such a concerted effort, it would be realistic to have a highly reliable model in about 4–5 years time. Faster completion is not anticipated, since much of the effort cannot be parallelized. Feasible Receivers for Impulse Radio Due to the wideband nature of UWB signals, many MPCs are resolvable at the receiver. To efficiently capture the energy contained in the multipath arrivals, the receiver has to efficiently combine the MPCs which are spread over time. The complexity of this problem increases with the signal bandwidth B, and the potential difficulties from the channel estimation and receiver complexity point of view are by no means unique to IR-UWB only. The same happens to UWB transmission based on chirp, on direct-sequence or on other spread-spectrum signals. A coherent receiver, which is able to utilize the energy of dozens of multipath arrivals, will be complex and costly. The hardware itself may consume a lot of power, and the time needed to estimate the corresponding amplitudes and phases wastes capacity. For the above reasons, suboptimal non-coherent as well as autocorrelation receivers have regained popularity. With both types of receiver, it is possible to capture the multipath energy easily, since in both cases multipath combining does not require any knowledge with respect to the phases and polarities. Unfortunately, this advantage can only be obtained at the expense of a rather low data rate and a higher sensitivity to in-band interferences, to noise, and to inter-symbol interferences. Such suboptimal detectors cannot be treated as a replacement of coherent solutions at all, but they are interesting alternatives, especially for low-power systems operating at low or medium data rates [136]. An important breakthrough that could possibly save IR-UWB is the transmit reference ultra wideband (TR-UWB) approach. Basically this is an old idea that was reintroduced by Hoctor and Tomlinson in the UWB context [92]. The approach boils down to sending a pulse

328

Technologies for the Wireless Future – Volume 3

pair or doublet instead of a single pulse, where each doublet consists of a reference pulse followed by an amplitude-modulated data pulse. At reception, this reference pulse shows how the transmitted pulse has been distorted by the propagation environment. Hence, the received reference pulse can be used as a noisy template to collect the data by a very straightforward autocorrelation mechanism. We have to simply correlate the received signal with a delayed version of itself, where the delay is the same as the one between the data and the reference pulse. A major advantage of such a TR-UWB system is that the analogue processing is simple and data-independent, and does not require either synchronization or channel estimation. The data rates after correlation are much lower (related to the frame rate) and easily captured by current A/D conversion technology. Multi-user systems can be created by combining with CDMA at the frame level (the user code has two dimensions and consists of the usual chip code, plus a delay code). Nonetheless, there are still a few major issues that have to be dealt with to make TR-UWB a viable UWB candidate, as discussed below. The crucial component of any TR-UWB system is the delay of the autocorrelation receiver. It can be viewed as the dual of the oscillator in a narrowband communication system. However, building an ultra wideband delay with a high accuracy and a small size and power is a major challenge. Several advances have already been reported, such as analogue delay filters and quantized analogue delays, but it appears that the range of the delays should be limited to a small multiple of the pulse width (a few nanoseconds) in order to satisfy the requirements. Since the environment may spread a transmitted pulse over more than 50 nanoseconds, adopting a small delay will introduce a significant overlap between the received reference pulse and the received data pulse, leading to inter-pulse interference (IPI). Moreover, if a high data rate is required, the doublets should be closely spaced, possibly leading to additional inter-frame interference (IFI). Digital signal processing comes into the picture to combat this IPI and IFI in the digital domain [93]. Note that compared to the all-digital receivers proposed for IR-UWB, which operate at the Nyquist rate, digital equalizers proposed for TR-UWB operate only at a multiple of the frame rate, which is generally much smaller than the Nyquist rate. This also means that the number of unknown propagation parameters required to construct a digital equalizer for TR-UWB is much smaller than that for all-digital receivers for IR-UWB. It is clear that the IPI and IFI will influence the performance, even with an optimal digital equalizer. This is the price we have to pay for the advantages that come with TR-UWB. Another price lies in the ‘squaring’ of the signal in the correlation process, which essentially doubles the noise, introduces many cross-terms, and increases the required dynamic range of the A/D converter. A major problem in TR-UWB, and IR-UWB in general, is the influence of narrowband interference (NBI), caused by WiFi systems, for example. Due to the generated cross-terms in the autocorrelation process, TR-UWB generally suffers more from NBI than IR-UWB. However, solutions have been proposed or are envisioned to solve this problem. First of all, ultra wideband antennas and LNAs are proposed that contain notches at the known NBI frequencies. Further, if a bandwidth from around 3 to 10 GHz is used, frequency down-conversion techniques are envisioned that fold the 5 GHz close to DC and fold the 2.4 GHz out of the desired band. A simple bandpass filter can then be used to remove the WiFi NBIs. The autocorrelation principle is then applied to the in-phase and quadrature components, instead of to

Short-range Wireless Communications

329

the original incoming signal. This has the additional advantage that the delays can be designed for frequencies up to 3 GHz instead of for frequencies up to 10 GHz, with savings on size and power. Finally, NBI can be further reduced in the digital domain, by applying special NBI cancellation schemes, with or without exploiting the knowledge of the NBI frequency [94]. Another interesting version of the TR receiver that avoids the need for an analogue delay line was proposed in [94]: the reference pulse and the data pulse are transmitted at the same time, but with a small frequency shift. Multiband or Monoband? The multiband approach has been proposed to be used in conjunction with several UWB modulation schemes which have traditionally been monoband solutions [95]. Here multiband transmission is defined as having each sub-band (SB) with nonoverlapping bands, and multicarrier transmission as having SBs with overlapping bands. Each subchannel is modulated with a particular modulation method for a separate symbol, resulting in a frequency-multiplexed transmission system. For each time slot the frequency utilization can remain the same or change after a certain time period, whereas the latter resembles the frequency-hopping spread spectrum principle. In general, only one sub-band can be hopped, or alternatively, all sub-bands can be hopped. The first laboratory prototype based on UWB multiband transmission was built in 2001 by General Atomics, according to [95]. Several variations have been proposed for multiband UWB. In addition to a detailed band plan, these variations differ in terms of how the modulation and spreading are performed in each SB. These include, for example, multiband OFDM system [96] and multiband IR system [97]. The motivation to use multiband rather than monoband is many-fold. One of the most significant benefits is that the multiband approach provides attractive means for easier digitalization, as well as controlling the self-interference and the interference between multiple users. While providing several benefits, the multiband approach has some disadvantages compared to monoband UWB occupying the same bandwidth. Multiband schemes often require a complex multiband signal generator which is able to quickly switch between frequencies [98]. Pulse shaping of individual bands needs to be designed to enable maximization of transmission power, and minimization of ICI and the complexity of the signal generator. Synchronization of frequency-hopped multiband signals introduces an additional complexity increase in the receiver. Finally, parallel signal processing of multiband signals results in more expensive digital and analogue parts for the transceivers. It is foreseen that the multiband modulation approach will play an important role especially in high-data rate UWB radios of the future. This is due to the inherent benefits that this technique provides. However, according to the above discussion, the technique introduces some additional challenges which require special attention. Generally, by using less bandwidth per information bit with multiband modulation, the risk of losing the benefits of traditional UWB transmission increases. Furthermore, a rapid frequency-switching multiband signal generation, along with precise switching offset synchronization and low-complexity techniques for parallel signal processing of simultaneously received signals in different sub-bands, must be carefully addressed. Finally, the design of time-frequency codes to minimize multiuser interference while maintaining reasonable complexity is an important research area.

330

Technologies for the Wireless Future – Volume 3

Synchronization While the very large bandwidth of UWB signals makes this technology unique and of great potential (large channel capacity and lack of significant fading), it also brings technical challenges that greatly increase the complexity of UWB devices. More specifically, due to the fine time resolution of UWB signals, the received signal is composed of a very large number of low-energy MPCs that need to be acquired. If the channel coefficients can be obtained, the low energy contained in MPCs can be collected by the Rake receiver, so that the UWB device can achieve a performance close to theoretical limits. In particular, it is known that a major practical implementation challenge for UWB receivers is the design of high-accuracy and rapid-synchronization algorithms. The main reason for the acquisition complexity is the lack of sufficient channel state information, as the receiver has to synchronize with the received symbols before we can perform channel estimation [99,100]. The effects of having no or little knowledge of the state information of the channel can be seen in Figure 7.38, where the probabilities of false lock of three estimators that differ in the amount of a priori knowledge of the state of the channel are plotted. It is seen that there is a significant gap in performance between a theoretical synchronization algorithm assumed to have a priori (and complete) knowledge of the channel and a practical algorithm (based on a maximum-likelihood approach) that corresponds to the case in which the receiver has no knowledge of the channel. It is also seen that the algorithm based on a maximum-a posteriori formulation, which is assumed to have knowledge of the channel statistics (obtained, for example, from a priori channel measurements), has a performance only a few dB away from the theoretically best estimator [99,100]. Hence, one possible way of improving the performance of practical synchronization algorithms is to consider longer observation intervals. 1. Maximum-likelihood estimator with perfect channel estimation (a priori knowledge of the channel realization) 2. maximum-a posteriori estimator (a priori knowledge of the channel statistics) 3. maximum-likelihood estimator (no a priori knowledge of the channel realization or statistics) As previously mentioned, IR-UWB and MB-OFDM schemes have been receiving most of the academic and commercial interest for implementing UWB communication systems. Although the difficulty of performing signal acquisition is inherent to UWB communication systems, each implementation technology has its own peculiarities. In particular, a significant amount of effort has been focused on synchronization for IR-based systems. Due to the use of an extremely short, low duty cycle UWB pulse (frame duration much larger than the pulse duration), the delay uncertainty region contains a large number of potential timing offsets compared with narrowband systems [101]. In addition, timing requirements are stringent because even minor misalignments may result in insufficient energy being captured to make data detection possible [102]. Conventional sliding correlation-based algorithms developed for narrowband systems would require a long search time and an unreasonably high sampling rate [103]. Recent algorithms proposed for UWB systems include coarse bin search, subspace spectral estimation, generalized likelihood ratio test, and cyclostationarity-based approaches (see [101–103] and references therein). As a consequence of the high complexity of the synchronization and channel estimation for IR-UWB, different suboptimum schemes have recently been proposed. Among such schemes, significant effort has been focused on TR-UWB (see Section 7.4.2.2, under ‘Feasible Receivers for Impulse Radio’ and

Short-range Wireless Communications

331

Probability of False Lock

100

10−1

(III)

(I)

10−2

(II)

10−3

10−4

0

5

10

15

20

25

30

Eb/No

Figure 7.38 Probability of false lock for a UWB communication system

[87,104–106]), in which channel information is ‘extracted’ from reference pulses. As previously mentioned, these schemes usually suffer from performance degradation in both data rate and error rate compared to UWB implementations that explicitly account for the overlaid narrowband interferer. The degree of difficulty of accurate synchronization of UWB receivers increases even more with the possible presence of NBIs resulting from the spectral overlay. As the UWB signals have low power and broad spectrum, it is possible, at least in principle, to allow for such signals to overlay narrowband systems with no noticeable interference. Based on this concept, in early 2002 the FCC allocated a 7.5 GHz bandwidth for UWB communication systems that forces its coexistence with narrowband systems, in an attempt to better utilize the spectrum. Because of the critical importance of the acquisition stage, the effect of NBI on the UWB receiver during this stage might be especially harmful, since, if acquisition fails, the desired signal cannot be successfully detected. The probability of false lock when acquisition is performed in the presence of NBI is shown in Figure 7.39. It is seen that the performance severely degrades when an interference mitigation technique is not used. However, the performance closely approaches the performance corresponding to the absence of NBIs when either of two mitigation techniques (covariance matrix estimation and spectral encoding) is used [99,100]. Synchronization is an important issue. Further research into fast and highly accurate acquisition algorithms should be undertaken in order for UWB to become a practical communications technology for high-data rate applications. Key Topics in UWB Hardware: Antennas and Low-power Aspects Beside the required baseband signal processing, two major hardware topics will significantly affect the performance of future mobile UWB systems for communication, localization and sensing: antennas and low-power aspects.

332

Technologies for the Wireless Future – Volume 3

Probability of False Lock

100

10−1

10−2

10−3

10−4

0

5

10

15

20

25

30

Eb/No

Figure 7.39 Probability of false lock for a UWB communication system in the presence of narrowband interference (using a maximum-likelihood estimator). Dots signify an absence of NBI, stars that no interference mitigation is employed, and squares and diamonds that an interference mitigation method (covariance matrix estimation and spectral encoding, respectively) is employed. SIR D
15dB

Antennas In the next 10 years UWB antennas for localization, sensing, imaging and communications will evolve from the present fixed configurations to more flexible ones, the so-called ‘smart-antennas’ (this notion is not directly related to multiple-antenna systems). Frequency-selective surfaces for narrowband applications are currently known in the communication and sensing field. They enable the propagation of electromagnetic waves for certain frequencies and in certain directions, as well as the cancellation of signals at other frequencies or other directions. Frequency-selective surfaces for UWB signals for meaningful applications have not been reported yet, however they shall become available as they significantly increase the system performance. Electronically-steered materials are available even today. They influence the antenna characteristics such as impedance, radiation pattern, polarization, frequency range, etc. It can be envisioned that such materials will provide a certain reduction of the antenna size. Currently, UWB antennas are either two- or three-dimensional. The 3D integration of antennas within a case, i.e. in a way that no additional space is required for the antennas, is one of the major goals to reach in, say, 10 years. Such implementation requires sophisticated 3D electromagnetic simulation tools. Up to now, the polarization of UWB antennas has been completely ignored. Nevertheless, polarization gives the opportunity, in all fields of applications, to significantly enhance the performance of the system, or to increase the efficiency of multiple antenna systems. Again, sophisticated 3D electromagnetic simulation tools are required for the implementation.

Short-range Wireless Communications

333

Low-power Aspects Ultra wideband signaling offers tremendous communication capacity and ranging capability even at the mandated transmit power levels and with the presence of in-band interference. While these constraints limit very high-data rate transmission (e.g. Gbps) to shorter distances, UWB radio also presents an attractive opportunity for both power-efficient (defined as throughput divided by total power consumption) and ultra low-power operation. To date, development generally has focused on MB-OFDM and IR-UWB architectures, but the possible design space is enormous since the physical signaling is not explicitly regulated. Scaling OFDM to UWB operation places strict performance criteria upon the building block circuits that must accommodate fast speeds, large bandwidths and increased computation. Additionally, for applications with a fixed power budget, e.g. battery-powered devices, these constraints must be satisfied with comparable or lower power consumption than existing narrowband solutions. Low cost, and hence a high level of integration, is also a goal, further directing the effort towards CMOS implementations. These block-level requirements are not insurmountable, though. For example, recent investigations into a critical building block, the A/D converter, have posted very promising results [107,108], demonstrating 4 bit GSample/s conversion with power consumption in the order of milliwatts. Similarly, front end matching for low noise amplification has been extended over the whole band: 3–10 GHz [110,111]. Indeed, complete MB-OFDM transceiver chipsets are now becoming commercially available. IR-UWB offers a competing approach to MB-OFDM for power-efficient, high-rate, short-distance communication. It also offers the possibility of ultra low-power and low-rate transmission. The simplicity and duty-cycled nature of pulse-based signaling has inspired several milliwatt-consuming radios based on a correlating receiver capable of Mbps communication and ranging [112,113]. These initial research results at the block and system levels imply that an order of magnitude reduction in power consumption is still possible from current designs. Further, the freedom of physical signaling encourages UWB research at the system and circuit levels in pursuit of more efficient or optimal implementations.

FOM (pj/Conversion Step)

1000 100 ∆Ε 10 Nyquist 1 ITRS2005

ISSCC 0.1

X2 / 3years 0.01 1985 1990 1995 2000 2005 2010 2015 2020 YEAR

Figure 7.40 ADC technology trend (source: the series of Proc. of IEEE Solid-Stated Circuits Conference (ISSCC) and the International Technology Roadmap for Semiconductor (ITRS) [109])

334

Technologies for the Wireless Future – Volume 3

Physical- and Higher-layer Visions Physical Layer It has been observed for more than two decades that data rates of wireless and wireline communication systems increase exponentially [114]. While today’s peak data rates for short-range communications reach 480 Mbps, in 2008/09 we expect data rates exceeding 1 Gbps, and in 2015 a peak rate of 10 Gbps. Major applications are considered to be high-quality video streaming (very high data rates, VHDR) or ultra high-speed data exchange; for instance, almost 100Gb capacity of mobile storages is envisioned in 2010, implying around 40 minutes copying time with today’s state-of-the-art wireless devices. As has been previously stated, a bottleneck for the final breakthrough of UWB is some severe implementation challenges, e.g. the availability of compact antennas, efficient amplifiers, low-power AD converters etc., suggesting that alternatives for boosting data rates are generally welcome. Note that UWB is a short-range and therefore mainly indoor communication technique in an environment characterized by dense multipath propagation or, in other words, by ‘rich scattering’. Fortunately, for such types of environment, MIMO (multiple input multiple output) systems allow for a significant increase of spectral efficiency and are therefore a promising ‘add on’ candidate for enabling future ultra high-speed wireless communications. Simply speaking, a system with two transmit and two receive antennas requires doubling of hardware costs and additional digital signal processing at the transmitter and receiver side, but also doubles, on average, the data rate. Hence, MIMO basically offers a technological trade off for high-data rate UWB systems. Some few years ago such a trade off caused the breakthrough of MIMO-WLANs, mainly because the analogue radio frequency front end became too expensive for modulation orders higher than 64-QAM. A similar situation occurs now for Mobile WiMAX and will also occur very soon for cellular communication systems (3G LTE, IEEE 820.20), where MIMO is considered mandatory on the technical roadmap. In conclusion, ‘MIMO-UWB’ is without doubt worth further investigation and seems to be a feasible approach for highest data rates emerging on the horizon. The challenges certainly still lie in the research and not yet in the development domain. In principal, it is rather unclear from several perspectives how bandwidth affects multi-antenna systems. For instance, how do fading models or diversity gains depend on bandwidth (see for example the comments in Section 7.4.1.2, under ‘The UWB Channel’)? Note that for IR-UWB the array gain may even double on a logarithmic scale (6 dB per doubling of the number of antennas) compared to wideband systems, since the inverse bandwidth of a UWB signal may become less than the travel time through an array [115]. Spatial channel measurements and spatial channel modeling are of particular importance for a thorough investigation of MIMO-UWB systems. A large number of new algorithm challenges can also be foreseen; for instance, in multiband-OFDM, multiple antennas can be exploited to improve the inherent band tracking and crucial packet detection. Hardware challenges only slightly differ from those of SISO-UWB, in the synchronization of the mixers (if needed) for down- and up-conversion, and the antenna array topologies. For localization problems, multipath propagation under non-line-of-sight (NLOS) environments causes difficult challenges: multiple antennas provide additional spatial information and are therefore promising for improving spatial resolution.

Short-range Wireless Communications

335

Higher Layers The most appealing scenarios of UWB application for low rate refer to networks that commonly adopt the self-organizing principle, i.e. distributed networks. Examples of these networks are ad hoc and sensor networks, i.e. groups of wireless terminals located in a limited-size geographical area, communicating in an infrastructure-free fashion, and without any central coordinating unit or base station. Communication routes may be formed by multiple hops to extend coverage. This paradigm can be viewed as different in nature from the cellular networking model, where typically nodes communicate by establishing single-hop connections, with a central coordinating unit serving as the interface between wireless nodes and the fixed wired infrastructure. In a recent approach [114] a UWB-IR tailored MAC design for asynchronous UWB networks is proposed in which asynchronous UWB users are allowed to transmit in an uncoordinated manner: the UWB2 approach. Results of network simulation showed that the probability of successful packet transmission is fairly high for uncoordinated transmission of a reasonable number of users, leading to the conclusion that a UWB-tailored MAC for low-bit rate applications may adopt a pure Aloha approach, the simplest and by excellence decentralized access policy. The other side of the coin is an increased overhead required by the presence of a synchronization trailer in each transmitted packet. This drawback may be an acceptable price to pay though for low-bit rate applications, while an additional advantage offered by the proposed approach is the possibility of collecting distance information between transmitter and receiver during control packets exchange. This information can enable the introduction in the MAC of new functions, such as distributed positioning. Developing a general methodology for the design of UWB-based communication systems is definitely a future challenge. In the network layer, the overall architecture and communication strategies of the system, such as routing, are defined, taking into account the constraints imposed by the application and the physical layer through the interface provided by the MAC layer. In this perspective, the de novo design of a rational communication network should first focus on the network layer, taking into account the goal of the network as well as the fundamental constraints originating from the physical layer. By focusing first on the network layer, the design is driven by the optimization of a properly tailored cost function in the network layer that is flexible enough to subsume a range of possible applications and systems, and to support ad hoc networking. A straightforward example of such an approach is multihop networks, where routing costs label different possible paths. A routing cost might interestingly depend upon a variety of parameters, including battery life, interference patterns, delay, etc., where features may selectively become priorities according to the nature of the network. A major challenge is how to incorporate the above cross-layer approach into the concept of a cognitive radio capable of adapting to the environment and of adjusting its principles of operation as a function of both external and internal unpredictable events [117]. This concept translates into developing smart wireless devices able to sense the environment, whether this refers to channel or interference patterns, and modify accordingly the spectral shape and other features of the radiated signals while maintaining compatibility with regulations on emitted radiations. The final goal is still to form wireless networks that cooperatively coexist with other wireless networks and devices. Given their ultra wide bandwidth, UWB radio signals must in principle coexist with other radio signals (see Section 7.4.2.2, under ‘Coexistence and Regulation’). The problem of possible interference from and onto other communication

336

Technologies for the Wireless Future – Volume 3

systems that must be contained within regulated values is thus intrinsic to the UWB radio principle, and is a mandatory aspect in conceiving the design. The cognitive radio concept focuses on improving the utilization of the wireless resource, that is, the electromagnetic spectrum. As such, it mainly applies to the behaviour of a single node regarding both its transmitter and receiver components, and, as a direct consequence, to the logic ruling communication over a single link. The introduction of the cognitive principle in the logic of the UWB network as regards resource management and routing will be a major challenge. This operation will require extending the cognitive concept to rules of operation that take into account the presence of several nodes in the network as well as their instantaneous configuration. Cognitive principles must be integrated into the rules of interaction between nodes in the network, that is, the set of wireless nodes forms a social network that must be modeled and analyzed as one entity in order to optimize the design. When cognitive principles affect the rules of interaction, cognitive scientists refer to a phenomenon called ‘consciousness’. While consciousness appears as a unique feature of the human brain, it is interesting to map the concept onto our context. Consider a set of UWB nodes forming a self-organizing ad hoc network. Suppose some nodes are cognitive, that is, they are gifted with some sort of intelligence based on adaptive algorithms that take into account the environment and the network in which they operate. Introducing consciousness into the network thus involves modeling nodes that operate in a cognitive way and apply conscious mechanisms to communicate in the network, and adapt their behaviour to the current network topology and status. As already noted above, due to the ultra wide bandwidth of radiated signals, radio devices operating under UWB rules must coexist with severely interfered environments, and must control their behaviour in order to favour coexistence. In other words, these radios must be capable of adapting to ever-changing operating conditions. We foresee that this can be achieved by introducing conscious mechanisms in the analysis process that is used by nodes for determining whether changes in the global network state are appropriate [118]. Coexistence and Regulation Traditionally, existing legacy radio communication systems are designed following the assumption that the received signal is an attenuated replica of the transmitted signal, mainly disturbed by the radio channel characteristics and superimposed by thermal noise in the receiver. This holds if the number of radio services and transmitting devices is limited, which was true for a relatively long period during the last century. Technical realization of system separation was achieved using nonadaptive traditional analogue filters with their inherent low flexibility. Therefore the frequency spectrum regulation policy favoured granting exclusive frequency use rights for dedicated radio services. Thus the current radio frequency regulation paradigm is based on frequency separation. With the start of the UWB regulation process in the US, Europe, Asia, and on ITU level, it became obvious that there are a number of challenges associated with the regulation of such a modern example of spectrum sharing technology based on coexistence and cognitive radio principles. The technical developments are challenging in several areas and still need a considerable amount of research and development in various basic disciplines considered as

Short-range Wireless Communications

337

precompetitive activities. Here we just mention some of these challenges, without claiming completeness: ž Designing radio communication systems based on the cognitive radio principle and having no exclusive spectrum usage rights while still ensuring certain quality of service for the application. ž Providing extremely low protection limits to passive radio astronomy services. ž Providing sufficient high protection to live critical services such as aviation radar. ž Providing sufficient protection to fixed wireless access terminals operating close to the sensitivity limits without introducing harmful interference. ž Protecting strategic services such as military radar and military sensing and communications without disclosing details of their system specification. ž Coordinating co-located radio systems (several radio platforms within a single user device, such as PDA, PC or mobile phone). ž Coping with many potential victim services (due to the inherently huge spectrum band used), which by now are partly or even completely unspecified. From the current point of view, the challenges listed above can be dealt with by applying certain techniques, such as mitigation techniques, and attempting to create common signaling carriers that are the basic up-to-date visions. First steps have been taken to verify the effectiveness of those techniques, and intensive research work is going on to further develop these visions and turn them into reality. Today’s spectrum sharing in Europe is based on a traditional regulation approach and in the near future this trend needs to be changed. In Europe it seems necessary now to have an evolution from the traditional incumbent radio services protection paradigm towards a promotion of viable coexistence scenarios, not creating harmful interference and based on cognitive principles and/or absolute low radio emissions. This new regulation approach shall allow a friendly sharing environment and will avoid radio resource allocation based on exclusive usage rights. UWB regulation in Europe is a good example of the need to build new regulation rules for allowing innovative services and technologies to develop. The regulation and standardization innovative rules could be the following: ž Definition of realistic coexistence scenarios taking into account real usage (indoor/outdoor usage, estimated number of devices, likelihood of worst case scenarios). ž Definition of new radio architectures in order to support coexistence mechanisms. This includes flexible air interfaces enabled to work on different frequencies. The detection-and-avoid mechanisms which are under studies for UWB in the lower band are good examples of the need to have new mechanisms on the air interface. The coexistence mechanisms can also be based on activity factors such as the low duty cycle mechanisms defined for UWB low-data rate systems. Introduction of new mechanisms at MAC and higher layers. This could be the possibility of knowing what kinds of radio are in the vicinity of the user and then putting in place mechanisms in order to avoid time slots or frequencies used by the ‘victim services’ (at PHY and MAC layers).

338

Technologies for the Wireless Future – Volume 3

These new rules, necessary for an evolution of the regulation, would have a great impact on the definition of new wireless standards. One of the difficulties of building a consensus, for example in CEPT ECC TG3 today, is due to the lack of recognition of this innovative approach, conducted in the group, to building new coexistence mechanisms.

7.4.2.3 Related Areas: Localization, Imaging and Sensing Localization In addition to communication applications, UWB signals are also well suited for location-aware applications due to their high time resolution and penetration capability [119]. Since a UWB signal occupies a large frequency spectrum that includes low frequencies as well as high frequencies, it has a higher probability of passing through or around obstacles. Moreover, time-frequency duality implies that a UWB signal has a very high time resolution, which facilitates accurate time-of-flight, and hence range, estimation. The accurate localization capability of UWB signals facilitates a number of applications, such as noninvasive patient monitoring, personnel tracking, search and rescue operations, and through-the-wall health monitoring of hostages. Although UWB signals can potentially provide very accurate location information even in harsh environments, there are a number of challenges in practical systems related to technology limitations and non-ideal channel conditions. Specifically, multipath and/or NLOS propagation, multiple-access interference (MAI) and high time resolution of UWB signals cause practical difficulties for accurate location estimation. Therefore, special attention should be paid to mitigation of errors due to the aforementioned error sources. Theoretical analysis of UWB signals promises very accurate localization capability. Although it is not always possible to get close to the lower limits of the localization accuracy with the current technology under practical constraints, it is expected that future UWB localizers will be able to perform location estimation with sub-centimetre accuracy for LOS scenarios, and with sub-decimetre accuracy for NLOS scenarios, by consuming around a few tens of milliwatts. One of the main technology improvements that would facilitate such low-power and accurate localization is related to the design of low-power analogue-to-digital converters (ADCs) operating at the order of several GHz. In addition, ranging and localization algorithms that perform joint optimizations of power consumption, localization accuracy and time constraints should be developed. Imaging and Sensing Sensors based on UWB technology are of great interest for a vast number of applications, such as surface penetrating radar, surveillance and emergency applications, medical instrumentation, nondestructive testing in civil engineering and the food industry, industrial sensors and microwave imaging, and many others. Their fundamental advantage over other sensors (narrowband RF, optical, etc.) comes from the huge fractional bandwidth of their sounding electromagnetic (EM) waves. This fractional bandwidth, often approaching the 200% limit, ensures high spatial resolution due to the large absolute bandwidth, and good penetration in materials due to the presence of low frequencies. It means that we are capable of looking into or through nonmetallic materials, and detecting and localizing objects or other phenomena of interest with an excellent precision. In contrast to conventional narrowband solutions, UWB sensors can obtain much more information about the material properties and structures of

Short-range Wireless Communications

339

scanned media or bodies. This is of major importance for short-range radar, with applications such as object recognition, environmental imaging, through-wall detection of persons, ground penetrating reconnaissance, landmine detection, etc. UWB sensors provide the advantages of high accuracy and robust operation even in multipath-rich propagation environments. So they still work in the case of bad optical view and obstructed line of sight connection. Imaging performed by electromagnetic waves is well known from nondestructive testing, ground penetrating radar, through-wall radar, medical diagnosis, etc. These methods exploit the scattering of EM waves in an unknown medium and involve some form of back propagation, back projection, or time reversal for image reconstruction [120–123]. Time domain imaging methods using broadband or UWB excitation signals are usually referred to as migration [121,122]. They are usually based on linear operations and take into account a number of simplifying assumptions. An example is the Kirchhoff migration, a well-known migration method [123]. It uses a ray optical model of wave propagation, excluding multiple reflections. It assumes Rayleigh or specular scattering of waves from objects. This requires the size of the objects to be clearly smaller or larger than the wavelength of the excitation signal. Note that in the case of baseband UWB stimulation, the relative span of wavelengths is very wide. If the object size is in the order of any wavelength involved, this gives rise to structural resonances or geometric-induced dispersions of waveforms, which causes image blurring. Moreover, Kirchhoff migration assumes a constant wave velocity, which must be a priori known. Despite these limitations, Kirchhoff migration is widely used due to its relatively low computational complexity. However, even though the computational complexity becomes a less critical factor with the progress in computer technology, some problems still remain open, e.g. nonlinear processing and avoiding over-simplification will be introduced in the field of UWB imaging. Some examples are presented in [124,137,138] for the cross-correlated back-projection algorithm, and in [125,126] for the modifications. Here, the nonlinear processing is used to improve the quality of focused images. Another approach, of inspecting the environment using ‘waves’, was inspired by nature. Some animals, e.g. bats, are able to navigate in the darkness without crashing against each other or obstacles, using only ultrasound cries and two ears. While flying, bats take ‘snapshots’ of the environment and combine them to obtain a focused image of the environment. Why not follow this approach with a single roaming robot equipped with UWB sensors? In this case, the position and orientation of the robot cannot be retrieved from measurements by multiple cooperating sensors. Imaging of the environment and localization of the robot within the environment have to be managed at the same time. An equivalent solution already exists for an autonomous robot equipped with a (stereo) camera instead of UWB sensors. The process typically comprises three steps: ž Feature extraction – features are characteristic elements within an image. The images of a stereo camera are analyzed for features that can be easily identified in subsequent images [127]. ž Feature correspondence search – features extracted from different images, taken at different times and from different viewpoints, need to be checked to make sure they belong to the same feature of the environment, e.g. a corner at a door or at a picture on a wall, etc. [127]. ž Feature tracking for motion analysis and image building – for imaging and navigation, features need to be tracked within a sequence of images and correctly positioned within the environment.

340

Technologies for the Wireless Future – Volume 3

For good results, all the three steps are typically built upon a model of the camera’s focal system. An interesting research question arises: whether a similar approach can be adopted in the UWB roaming robot scenario. The very initial results showing main challenges in this area were published in [128]. Using the imaging algorithms discussed above, the sensor networks can obtain an image of the environment in terms of shape and location of dominant objects such as walls and other static objects. In order to perform some meaningful action in the specific environment, these static objects must first be recognized and described in more details. The fundamental advantages of UWB sensors come from their huge bandwidth. The bandwidth of the sensor is a very important feature that determines a number of its parameters. By extending the sensor’s bandwidth, it is possible to obtain additional gains about the target under test, e.g. an improvement of its range and azimuth resolution for localization systems, an improvement of its target recognition capabilities for GPR radar, etc. However, the extreme bandwidth is not the only benefit offered by the UWB sensor. Apart from the above, other important features possessed by UWB sensors are: ž A high data recording rate to match the time variance of the medium under test. ž Multichannel arrangement capability, in order to allow localization. ž A high degree of hardware configuration flexibility to adapt the system performance to the actual requirements of the individual user. The features and capabilities of a sensor strongly depend on the signal used by the sensor to stimulate the medium under test. There are more possibilities for UWB excitation signals, such as: ž chirp signals ž short-impulse signals ž pseudo noise (PN) binary sequences, and others. However, not all of these excitation signals are suitable to be used for a specific application. For example, chirp signals can easily meet the UWB bandwidth demand. The signals can be generated in a stable manner, with low jitter and drift. The multichannel arrangement is also not a big problem. However, sensors using chirp excitation signals are not suitable for real-time operation, such as sensors using short impulses or binary sequences. This is prohibited because of the slow measurement rate. On the other hand, the existing short-impulse systems are often subjected to a number of constraints such as limited bandwidth, susceptibility to jitter and drift, complex electronics and others. All of these prevent the effective application of UWB sensors in e.g. laboratory measurement devices such as network analyzers. There are some interesting properties of the system using PN binary sequences. These sequences can be easily generated with up to a tenth of 1 GHz of bandwidth by a digital shift register, which is clocked by a stable single-tone radio frequency oscillator. Besides the advantage of having a reasonable correlation gain, these signals are characterized by small binary voltage amplitudes, which allow extremely fast digital switching in integrated circuit technology to meet the demanding requirements on bandwidth and low jitter. Implementation examples of sensors using binary sequences are given in e.g. [129].

Short-range Wireless Communications

341

Thus, it is evident that the key to a powerful UWB sensor for a specific application is the use of an appropriate stimulation signal, since the device hardware and signal processing depend upon it. The main technological challenges in the development of a flexible UWB sensor lie in its monolithic integration, increasing its efficiency, decreasing its power consumption, and in the development of cost-effective sensing principles. 7.4.2.4 Conclusions The state of the art, the challenges and several visions are outlined in the previous sections for different technologies and applications based on UWB. Spectral overlay remains a major concern for UWB systems and currently deters their worldwide adoption. Concerns about possible interference of UWB on GPS systems, for example, led the FCC to protect the GPS band by limiting UWB transmissions to the 3.1–10.6 GHz range. More recently, the European Commission regulators have issued a set of draft rules to define regulations that would require UWB devices to implement certain mitigation techniques (such as ‘detect-and-avoid’ and a cap on device activity) in order to reduce the likelihood of interference with other systems [130]. However, the visions of UWB are apparent: highest data rates for short-range communication, accurate indoor localization (even under non-line-of-sight conditions) and a multitude of applications in the imaging and sensing domains [Proceedings of the IEEE, Special Issue on UWB, to appear spring 2009]. To somehow quantify such visions, we here access empirical observations, such as Edholm’s law of data rate, shown in Figure 7.41. It can be forecasted that UWB will break the 1 Gbps barrier before 2010, while future developments, as with today’s narrowband and wideband communications, likely strive for UWB systems equipped with multiple antennas. To be more specific, we expect data rates with MIMO-UWB to exceed 10 Gbps within 10 years. Regarding the localization accuracy, or imaging or sensing, similar empirical laws are not available yet. However, having in mind advanced accuracy bounds derived by estimation theory (e.g. the Cramer-Rao lower bound), we can conclude a sub-centimetre accuracy for line-of-sight environments in 10 years, while in non-line-of-sight environments sub-decimetre accuracy should be feasible with highly sophisticated signal processing and continuative technology progress. Note that this present and rather qualitative survey will be validated and extended by a more detailed contribution, being part of a future book to be released by the same publisher.

5 Gbit/s 500 Mbit/s 50 Mbit/s 802

5 Mbit/s 2000

.11b

802

.11a

2004

MMO802.11n /UWB

2008

MMOUWB

2012

Figure 7.41 Edholm’s law of data rate for short-range communications

342

Technologies for the Wireless Future – Volume 3

The wealth of applications for short-distance, low-power or power-efficient wireless devices stimulates development at all system levels. Current designs are only beginning to realize the potential of UWB for communication, localization, imaging and sensing. We are therefore looking forward in curiosity to the next decade of UWB research and development.

7.5 Acknowledgements We would like to thank Marcos Katz (VTT, Finland) and Frank H.P. Fitzek (Aalborg University, Denmark) for editing Section 7.2. For contributions to Section 7.2 we would like to thank Beatrice Pietrarca and Giovanni Sasso, especially for the measurement campaigns carried out in their master project ‘Cooperative Technologies for Wireless Grids’. Furthermore, we would like to acknowledge the students of the SMARTEX project for the support in the research work. We would like to thank Moshe Ran, Yossef Ben Ezra, Motti Haridim and Boris I. Lembrikov (Holon Institute of Technology (HIT)), Ronen Korman (Wisair, Israel), Manoj Thakur (University of Essex, UK), Isabelle BUCAILLE (Thales group, France), Roberto Llorente (Universidad Politecnica de Valencia, Spain), Henrique Salgado (INESC Porto, Portugal), Beatrice Cabon (INPG, France) and Norbert Gungle (TES, Germany) for editing Section 7.3. We would like to thank Eckhard Grass, Frank Herzel, Maxim Piz, Klaus Schmalz, Yaoming Sun, Srdjan Glisic, Milos Krstic, Klaus Tittelbach-Helmrich, Marcus Ehrig, Wolfgang Winkler, Christoph Scheytt and Rolf Kraemer (IHP GmbH, Frankfurt (Oder), Germany) for editing Section 7.4.1. Also we grateful for the support from all partners of the WIGWAM project (http://www.wigwam-project.com) for Section 7.4.1. Also we would like to thank for the editing of Section 7.4.2 by Thomas Kaiser and Emil Dimitrov (Institute of Communications Technology, Leibniz University of Hannover, Germany) Co-authors: Antti Anttonen, Robert W. Brodersen, Isabelle Bucaille, Claudio da Silva, Maria-Gabriella di Benedetto, Gunter Fischer (IHP GmbH, Frankfurt (Oder), Germany), Sinan Gezici, Thomas Kaiser, Geert Leus, Aarne Maemmelae, Laurence Milstein, Andreas Molisch, Lorenzo Mucchi, Ian O’Donnell, H. Vincent Poor, Juergen Sachs (Institute of Information Technology, Ilmenau University of Technology, Germany),Werner Soergel, Waho Takao, Reiner Thomae (Institute of Information Technology, Ilmenau University of Technology, Germany), Alle-Jan van der Veen, Mike Wolf (Institute of Information Technology, Ilmenau University of Technology, Germany), Sven Zeisberg, Rudolf Zetik (Institute of Information Technology, Ilmenau University of Technology, Germany).

References [1] http://www.aka-aki.com. [2] http://mobiledevices.kom.aau.dk/projects/smartex/. [3] M. Stini, M. Mauve, F. H. P. Fitzek, “Digital Ownership, “From content consumers to owners and traders”, IEEE MultiMedia, 13(4), pp. 1, 4–6, Oct.–Dec. 2006. [4] F. H. P. Fitzek, M. V. Pedersen, G. P. Perucci, S. Rein and C. G¨uhmann, “Convergence of mobile devices and wireless sensor networks”, Wireless World Research Forum (WWRF) 17, WWRF, Heidelberg, Germany, 2006. [5] F. H. P. Fitzek and M. Katz (eds), “Cooperation in wireless networks: principles and applications: real egoistic behavior is to cooperate!”, Springer, Apr. 2006.

Short-range Wireless Communications

343

[6] J. M. Kristensen and F. H. P. Fitzek, “Cooperative Wireless Networking Using Software Defined Radio”, International OFDM Workshop, Aug. 2006. [7] J. M. Kristensen and F. H. P. Fitzek, “The application of software-defined radio in a cooperative wireless network”, Software Defined Radio Technical Conference, SDR Forum, Orlando, Florida, USA, Nov. 2006. [8] F. H. P. Fitzek, G. Schulte and M. Reisslein, “System architecture for billing of multi-player games in a wireless environment using GSM/UMTS and WLAN services”, Proc. First Workshop on Network and System Support for Games (NetGames 2002), pp. 58– 64, Braunschweig, Germany, Apr. 2002. [9] Q. Zhang, F. H. P. Fitzek and M. Katz, “Cooperative power saving strategies for IP-services supported over DVB-H networks”, IEEE Wireless Communications and Networking Conference (WCNC) 2007 – Networking, Hong Kong, Mar. 2007. [10] Q. Zhang, F. H. P. Fitzek and Marcos Katz, “Evolution of heterogeneous wireless networks: cooperative networks”, 3rd International Conference of the Center for Information and Communication Technologies (CICT) – mobile and wireless content, services and networks – short-term and long-term development trends, Nov. 2006. [11] M. Katz and F. H. P. Fitzek, “Cooperative techniques and principles enabling future 4G wireless networks”, International Conference on EUROCON 2005, 1, pp. 21–24, Nov. 2005. [12] J. M. Kristensen, F. H. P. Fitzek, P. Koch and R. Prasad, “Reducing computational complexity in software defined radio using cooperative wireless networks”, International Symposium on Wireless Personal Multimedia Communications (WPMC’05), Aalborg, Denmark, Sep. 2005. [13] F. H. P. Fitzek, M. Katz and Q. Zhang, “Cellular controlled short-range communication for cooperative P2P networking”, Wireless World Research Forum (WWRF) 17, WWRF, WG5, Heidelberg, Germany, Nov. 2006. [14] http://matplotlib.sourceforge.net/. [15] K. Alameh and R. A. Minasian, “Optimum optical modulation index of laser transmitters in SCM systems”, Electron. Lett., 26, pp. 1273– 1275, Aug. 1990. [16] B. Allen (ed.), “Ultra wideband: technology and future perspectives”, WWRF white paper V3.0, Mar. 2005. [17] G. P. Agrawal, “Fiber-optic Communication Systems”, John Wiley & Sons, Ltd, New York, 2002. [18] H. Bong Kim and A. Wolisz, “A radio over fiber based wireless access network architecture for rural area”, 14th IST Mobile and Wireless Commun. Summit, Dresden, Jun. 2005. [19] B. Boyer, J. Haidar, A. Viloct and M. Bouthinon, “Tunable Microwave Load Based on Biased Photoinduced Plasma in Silicon”, IEEE Trans. on MTT, 45(8), pp. 1362– 1367, Aug. 1997. [20] B. Cabon, Y. Le Guennec and G. Maury, “Optical microwave mixing techniques for broadband and low cost radio-over-fiber applications”, Proc. Photonics North Conference, Sep. 2005. [21] J. T. Cornoy, J. L. LoCicero and D. R. Ucci, “Communication techniques using monopulse waveforms”, MILCOM’99, 2, pp. 1185– 1191, 1999. [22] C. H. Cox, “Analog Optical Links Theory and Practice”, Cambridge University Press, 2004. [23] C. H. Cox, “Analog optical links: models, measures and limits of performance”, in A. Vilcot, B. Cabon and J. Chazelas (eds), Microwave Photonics, pp. 210– 219, Kluwer, Boston, 2003. [24] B. M. Donlan, D. R. McKinstry and R. M. Buehrer, “The UWB indoor channel: large and small scale modeling”, IEEE Trans. on Wireless Commun., 5(10), Oct. 2006. [25] ECMA-368, “High rate ultra wideband PHY and MAC standard”, Dec. 2005. [26] Y. M. El-Bataway and M. J. Deen, “Analysis and circuit modeling of waveguide-separated absorption charge multiplication-avalanche photodetector (WG-SACM-APD)”, IEEE Transactions on Electron Devices, 52(3), pp. 335–344, Mar. 2005. [27] R. D. Esman, “Microwave functions enabled by photonics”, in A. Vilcot, B. Cabon and J. Chazelas (eds), Microwave Photonics. pp. 375–399, Kluwer, Boston, 2003. [28] R. Fisher, M. Mc Laughlin and M. Welborn, “DS-UWB physical layer,’” IEEE P802.15.3a Working Group for Wireless Personal Area Networks (WPAN), IEEE P802.15-04/0137r3, Jul. 2004. [29] J. Foerster, “Channel modeling sub-committee report – final”, IEEEP802.15 Working Group for Wireless Personal Area Network (WPAN), IEEE P802.15-02/490r1-SG3a, 7 Feb. 2003. [30] Y. Le Guennec, G. Maury, J. Yao and B. Cabon, “New optical microwave up-conversion solution in radio-over-fiber network for 60 GHz wireless applications”, IEEE J. of Lightwave Technology, 24(3), Mar. 2006.

344

Technologies for the Wireless Future – Volume 3

[31] H. Takeuchi, K. Tsuzuki, K. Sato, M. Yamamoto, Y. Itaya, A. Sano et al., “NRZ operation at 40Gb/s of a compact module containing an MQW electroabsorption modulator integrated with a DFB laser”, IEEE Photonics Technology Letters, 9(5), pp. 572– 574, May 1997. [32] H.-F. Chou and J. E. Bowers, “High-speed OTDM and WDM networks using traveling-wave electroabsorption modulatrors”, IEEE Journal of Selected Topics in Quantum Electronics, 13(1), pp. 58–69, Feb. 2007. [33] D. Jager, “High speed photodetection”, in A. Vilcot, B. Cabon and J. Chazelas (eds), Microwave Photonics, pp. 82–100, Kluwer, Boston, 2003. [34] J. Kelly, “MB-OFDM physical layer submission”, IEEE P802.15.3a Working Group for Wireless Personal Area Networks, IEEE 802.15-04/0122r4, Mar. 2004. [35] K.-I. Kitayama, A. Stohr, T. Kuri, R. Heinzelmann, D. Jager and Y. Takahashi, “An approach to single component antenna base stations for broadband millimeter-wave fiber-radio access systems”, IEEE Transactions on Microwave Theory and Techniques, 48(12), pp. 2588– 2595, 2000. [36] S. Kim, H. Jang, S. Yonghoon and J. Jeong, “Performance evaluation for UWB signal transmission with different modulation schemes in multi-cell environment distributed using RoF technology”, UWB Conf., 2004. [37] J. Kim, Y.-S. Kang, Y.-D. Chung, K.-S. Choi, “Development and RF characteristics of analog 60-GHz electroabsorption modulator module for RF/optic conversion”, IEEE Transactions on MTT, 54(2), pp. 780–787, Feb. 2006. [38] A. Larsson, C. Carlson, J. Gustavsson, A. Haguland, P. Modh and J. Bengtsson, “Direct high frequency modulation of VCSELs and applications in fibre optic RF and microwave links”, New J. of Physics, 6, pp. 176–190, 2004. [39] Sergei Malyshev and Alexander Chizh, “State of the art high-speed photodetectors for microwave photonics application”, Microwaves, Radar and Wireless Communications, 2004 MICON 15 International Conference on, 3(17– 19), pp. 765– 775, May 2004. [40] G. Maury, A. Hilt, T. Berceli, B. Cabon and A. Vilcot, “Microwave-frequency conversion methods by optical interferometer and photodiode”, IEEE Transactions on Microwave Theory and Techniques, Special Issue Microwave and Millimeter-Wave Photonics II, pp. 1481– 1485, Aug. 1997. [41] J. E. Mazo, “Asymptotic distortion of clipped, dc-biased, gaussian noise”, IEEE Trans. on Commun., 40, pp. 1339– 1344, Aug. 1992. [42] F. Mendis and B. Tan, “Overmodulation in subcarrier multiplexed video FM broadband optical networks”, IEEE J. on Select. Areas in Commun., 8, pp. 1285– 1289, Sep. 1990. [43] C. Minot, “Electroabsoprtion modulators and photo-oscillators for conversion of optics to millimeter waves”, in A. Vilcot, B. Cabon and J. Chazelas (eds), Microwave Photonics. pp. 73–81, Kluwer, Boston, 2003. [44] I. Opperman, M. Hamalainen and J. Linatti, “UWB Theory and Applications”, Chapter 3, John Wiley & Sons, Ltd, 2004. [45] M. R. Phillips and T. E. Darcie, “Numerical simulation of clipping-induced distortion in analog lightwave systems”, IEEE Photonics Technol. Lett., 3, pp. 1153– 1155, Dec. 1991. [46] H. Al-Raweshidy and S. Komaki (eds), “Radio over Fiber Technologies for Mobile Communications Networks”, Artech House, 2002. [47] S. Saddow and C. H. Lee, “Optical control of microwave-integrated circuits using high-speed GaAs and Si photoconductive switches”, IEEE Transactions on Microwave Theory and Techniques, 43(9), pp. 2414– 2420, Sep. 1995. [48] A. M. Saleh, A. J. Rustakoand and R. S. Roman, “Distributed antennas for indoor radio communications”, IEEE Trans. Commun. COM-35, 12, Dec. 1987. [49] A. A. M. Saleh, “Fundamental limit on number of channels in subcarrier multiplexed lightwave CATV system”, Electron. Lett., 25, pp. 776– 777, Jun. 1989. [50] B. E. A Salehi and M. C. Teich, “Fundamentals of Photonics”, John Wiley & Sons, Ltd, 1991. [51] H. M. Salgado and J. J. O’Reilly, “Performance assessment of FM broadcast subcarrier-multiplexed optical systems”, IEE Proc.-J, 140, pp. 397–403, Dec. 1993. [52] H. M. Salgado and J. J. O’Reilly, “Performance assessment of subcarrier multiplexed optical systems”, in B. Wilson and I. Darwazeh (eds), Analogue Optical Fibre Communications, Peter Peregrinus, Ltd, 1995. [53] A. L. Seeds, “Microwave photonics”, IEEE Transactions on Microwave Theory and Techniques, 50(3), pp. 877–887, Mar. 2002.

Short-range Wireless Communications

345

[54] C. K. Sun, R. J. Orazi, S. A. Pappert, “Efficient microwave frequency conversion using photonic link signal mixing”, IEEE Photonics Technology Letters, 8(1), pp. 154–156, 1996. [55] K. Siwiak, “UWB propagation notes to P802.15 SG3a”, IEEE 802.15-02/328r0, Jul. 2002. [56] M. Varasi, “Fast modulators”, in A. Vilcot, B. Cabon and J. Chazelas (eds), Microwave Photonics, pp. 57–73, Kluwer, Boston, 2003. [57] D. Wake, “Radio over fiber systems for mobile applications”, H. Al-Raweshidy and S. Komaki (eds), “Radio over Fiber Technologies for Mobile Communications Networks”, pp. 217–241, Artech House, 2002. [58] S. D. Walker, M. Li, A. C. Boucouvalas, D. G. Cunningham and A. N. Coles, “Design techniques for subcarrier multiplexed broadcast optical networks”, IEEE Journal on Selected Areas in Communications, 8, pp. 1276– 1284, Sep. 1990. [59] R. J. Westcott, “Investigation of multiple FM/FDM carriers through a satellite TWT operating near to saturation”, Proc. IEEE, 114, pp. 726–740, Jun. 1967. [60] WWRF, “Book of Visions”, Dec 2001, http://www.wireless-world-research.org/index.php?id D 107. [61] L. Yang and G. B. Giannakis, “Ultra wideband communications”, IEEE Signal Processing, Nov. 2004. [62] T. Young, J. Conradi and W. R. Tinga, “BER characteristics of ³/4 DQPSK microwave subcarrier signals on optical fiber using Mach-Zehnder modulator nonlinear upconversion”, IEEE Photonics Technology Letters, 8(11), pp. 1552– 1554, 1996. [63] http://www.ieee802.org/15/pub/TG3c.html. [64] E. Grass, F. Herzel, M. Piz, Y. Sun and R. Kraemer, “Implementation aspects of Gbit/s communication system for 60 GHz band”, Proc. 14th Wireless World Research Forum (WWRF14), San Diego, USA, Jul. 2005. [65] F. Herzel, M. Piz and E. Grass, “Frequency synthesis for 60 GHz OFDM systems”, Proc. 10th International OFDM Workshop (InOWo’05), pp. 303– 307, Hamburg, Germany, Aug. 2005. [66] Y. Sun, S. Glisic and F. Herzel, “A fully differential 60 GHz receiver front-end with integrated PLL in SiGe:C BiCMOS”, Proc. 1st European Microwave Integrated Circuits Conference (EuMIC2006), Manchester, UK, pp. 198–201, Sep. 2006. [67] K. Schmalz, F. Herzel and M. Piz, “An integrated 5 GHz wideband quadrature modem in SiGe:C BiCMOS technology”, Proc. 36th European Microwave Conference (EuMC2006), Manchester, UK, pp. 1656– 1659, Sep. 2006. [68] J. Stott, “The effects of phase noise in COFDM”, BBC Research and Development, EBU Technical Review, Summer 1998. [69] B. Floyd, S. Reynolds, U. Pfeiffer, T. Beukema, J. Grzyb and C. Haymes, “A silicon 60 GHz receiver and transmitter chipset for broadband communications”, ISSCC Dig. Tech Papers, pp. 184– 185, Feb. 2006. [70] http://www.sibeam.com/company/pr 06.27.2007.html. [71] http://www.ieee802.org/15/pub/TG3c.html. [72] http://www.ieee802.org/11/ (search ‘VHT’). [73] http://www.wirelesshd.org/. [74] P. F. M Smulders, “60 GHz radio: prospects and future directions”, Proc. IEEE Benelux Chapter, 10th Symposium on Communications and Vehicular Technology, Eindhoven, The Netherlands, pp. 1–8, Nov. 2003. [75] Y. Shoji, K. Hamaguchi and H. Ogawa, “Milimeter-wave remote self-heterodyne system for extremely stable and low-cost broad-band signal transmission”, IEEE Transactions Microwave Theory and Techniques, 50, pp. 1458– 1467, Jun. 2002. [76] K. Ohata, K. Maruhashi, M. Ito, S. Kishimoto, K. Ikuina, T. Hashiguchi et al., “A wireless 1.25Gb/s transceiver module at 60 GHz-band”, ISSCC Digest of Technical Papers, pp. 298–299, San Francisco, USA, Feb. 2002. [77] K. Maruhashi, S. Kishimoto, M. Ito, K. Ohata, Y. Hamada, T. Morimoto et al., “Wireless uncompressed-HDTV-signal transmission system utilizing compact 60-GHz-band transmitter and receiver”, Microwave Symposium Digest, 2005 IEEE MTT-S International, Jun. 2005. [78] S. K. Reynolds, B. A. Floyd, U. R. Pfeiffer, T. Beukema, J. Grzyb, C. Haymes et al., “A silicon 60-GHz receiver and transmitter chipset for broadband communications”, IEEE Journal of Solid-State Circuits, 41, pp. 2820– 2831, Dec. 2006. [79] E. Grass, F. Herzel, M. Piz, K. Schmalz, Y. Sun, S. Glisic et al., “60 GHz SiGe-BiCMOS radio for OFDM transmission”, IEEE International Symposium on Circuits and Systems (ISCAS), New Orleans, USA, May 2007.

346

Technologies for the Wireless Future – Volume 3

[80] B. Razavi, “A 60-GHz CMOS receiver frontend”, IEEE Journal of Solid-State Circuits, 41, pp. 17–22, Jan. 2006. [81] J. A. Howarth, A. P. Lauterbach, M. J. Boers, L. M. Davis, A. Parker, J. Harrison et al., “60 GHz radios: enabling next-generation wireless applications”, Proc. IEEE TENCON05, pp. 2354– 2359, Melbourne, Dec. 2005. [82] F. Herzel, M. Piz and E. Grass, “Frequency synthesis for 60 GHz OFDM systems”, Proc. 10th International OFDM Workshop (InOWo’05), pp. 303– 307, Hamburg, Germany, Aug. 2005. [83] F. Herzel, S. Glisic and W. Winkler, “Integrated frequency synthesizer in SiGe BiCMOS technology for 60 and 24 GHz wireless applications”, Electronics Letters, 43, pp. 154– 156, Feb. 2007. [84] Manabe, Miura and Ihura, “Effects of antenna directivity and polarization on indoor multipath propagation characteristics at 60 GHz”, Journal on Selected Areas in Communications, 14(3), Apr. 1996. [85] Xu, Kukshya and Rappaport, “Spacial and temporal characteristics of 60-GHz indoor channels”, Journal on Selected Areas in Communications, 20(3), Apr. 2002. [86] “France telecom – IHP joint physical layer proposal for IEEE 802.15 task group 3c”, IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs), May 2007. [87] J. D. Choi and W. E. Stark, “Performance of ultra-wideband communications with suboptimal receivers in multipath channels”, IEEE Journal on Selected Areas in Communications, 20, pp. 1754– 1766, Dec. 2002. [88] S. K. Yong, C. Chong and S. Lee, “UWB-DCSK communication systems for low rate WPAN applications”, Proc. IEEE International Symposium on Personal Indoor and Mobile Radio Communications (PIMRC 2005), Sep. 2005. [89] B. H. Fleury, M. Tschudin, R. Heddergott, D. Dahlhaus and K. I. Pedersen, “Channel parameter estimation in mobile radio environments using the SAGE algorithm”, IEEE Journal on Selected Areas in Communications, 17(3), pp. 434– 450, Mar. 1999. [90] J. M. Cramer, R. A. Scholtz and M. Z. Win, “Evaluation of an ultra-wideband propagation channel”, IEEE Transactions on Antennas and Propagation, 50(5), pp. 561–570, May 2002. [91] P. A. Bello, “Characterization of randomly time-variant linear channels,” IEEE Trans. Commun., 11(4), pp. 360–393, Dec. 1963. [92] R. Hoctor and H. Tomlinson, “Delay-hopped transmitted reference RF communications”, Proc. IEEE Conference on Ultra Wideband Systems and Technologies, pp. 265– 270, 2002. [93] Q. H. Dang, A. Trindade, A. J. van der Veen and G. Leus, “Signal model and receiver algorithms for a transmit-reference ultra-wideband communication system”, IEEE Journal on Selected Areas in Communications, 24, pp. 773– 779, Apr. 2006. [94] Q. H. Dang and A. J. van der Veen, “Narrowband interference mitigation for a transmit-reference ultra-wideband receiver”, Proc. 14th European Signal Processing Conference (EUSIPCO 2006), Florence, Italy, 2006. [95] G. Aiello and G. Rogerson, “Ultra-wideband systems”, IEEE Microwave Magazine, pp. 36–47, 2003. [96] A. Batra, J. Balakrishnan, G. Aiello, J. Foerster and A. Dabak, “Design of a multiband OFDM system for realistic UWB channel environments”, IEEE Transactions on Microwave Theory and Techniques, pp. 2123– 2137, Sep. 2004. [97] S. Paquelet, L.-M. Aubert and B. Uquen, “An impulse radio asynchronous transceiver for high data rates”, Proc. International Workshop on Ultra Wideband Systems, joint with Conference on Ultrawideband Systems and Technologies, pp. 1–5, May 2004. [98] C. Mishra et al., “Frequency planning and synthesiser architectures for multiband OFDM UWB radios”, IEEE Transactions on Microwave Theory and Techniques, pp. 3744– 3756, Dec. 2005. [99] C. R. C. M. da Silva and L. B. Milstein, “Coarse acquisition performance of spectral-encoded UWB communication systems in the presence of narrow-band interference”, Proc. Asilomar Conference on Signals, Systems and Computers, pp. 1099– 1103, Nov. 2005. [100] C. R. C. M. da Silva and L. B. Milstein, “Spectral-encoded UWB communication systems: Real-time implementation and interference suppression”, IEEE Transactions on Communications, 53, pp. 1391– 1401, Aug. 2005. [101] J. Ibrahim and R. M. Buehrer, “Two-stage acquisition for UWB in dense multipath”, IEEE Journal on Selected Areas in Communications, 24, pp. 801–807, Apr. 2006.

Short-range Wireless Communications

347

[102] L. Yang and G. B. Giannakis, “Ultra-wideband communications: an idea whose time has come”, IEEE Signal Processing Magazine, 21, pp. 26–54, 2004. [103] Z. Tian and G. B. Giannakis, “A GLRT approach to data aided timing”, IEEE Transactions on Wireless Communications, 4, pp. 2956– 2967, Nov. 2005. [104] H. Zhang and D. L. Goeckel, “Generalized transmitted-reference UWB systems”, Proc. IEEE Conference on Ultra Wideband Systems and Technologies, Reston, Virginia, pp. 147– 151, 2003. [105] C. Carbonelli, S. Franz, U. Mengali and U. Mitra, “Semi-blind ML synchronization for UWB transmitted reference systems”, Proc. Asilomar Conference on Signals, Systems and Computers, Pacific Grove, California, pp. 1491– 1495, 2004. [106] R. Djapic, G. Leus, A. J. van der Veen and A. Trindade, “Blind synchronization in asynchronous ultra wideband UWB networks based on the transmit-reference scheme”, Eurasip J. Wireless Comm., ID. 37952, DOI:10.1155/WCN/2006/37952, 14 pages, 2006. [107] G. V. der Plas, S. Decoutere and S. Donnay, “A 0.16pJ/conversion-step 2.5mW 1.25GS/s 4b DC in a 90 nm digital CMOS process”, Proc. IEEE Int. Solid-State Circuits Conf., pp. 566– 567, Feb. 2006. [108] S. W. M. Chen and R. W. Brodersen, “A 6b 600MS/s 5.3mW asynchronous ADC in 0.13um CMOS”, Proc. IEEE Int. Solid-State Circuits Conf., pp. 574– 575, Feb. 2006. [109] http://www.itrs.net/. [110] A. Bevilacqua and A. M. Niknejad, “An ultra-wideband CMOS LNA for 3.1 to 10.6 GHz wireless receivers”, Proc. IEEE Int. Solid-State Circuits Conf., pp. 382– 383, Feb. 2004. [111] A. Ismail and A. Abidi, “A 3 to 10 GHz LNA using a wideband LC-ladder matching network”, Proc. IEEE Int. Solid-State Circuits Conf., pp. 384–385, Feb. 2004. [112] I. D. O’Donnell and R. W. Brodersen, “A 2.3mW baseband impulse-UWB transceiver front-end in CMOS”, IEEE VLSI Cicruits Digest of Tech. Papers, pp. 248–249, Jun. 2006. [113] T. Terada, S. Yoshizumi, Y. Sanada and T. Kuroda, “A CMOS impulse radio ultra-wideband transceiver for 1 Mb/s data communication and C/-2.5cm range findings”, IEEE VLSI Circuits Digest of Tech. Papers, pp. 30–33, Jun. 2005. [114] S. Cherry, “Edholm’s law of bandwidth”, IEEE Spectrum, 41, pp. 58–60, Jul. 2004. [115] S. Ries and T. Kaiser, “Ultra wideband impulse beamforming: it’s a different world”, Signal Processing in UWB Communications, 86, pp. 2198– 2207, Sep. 2006. [116] M.-G. Di Benedetto, L. De Nardis and G. Giancola M. Junk, “(UWB) 2: Uncoordinated, Wireless, Baseborn, medium access control for UWB communication networks”, Mobile Networks and Applications, 10, pp. 663–674, 2005. [117] S. Haykin, “Cognitive radio: brain-empowered wireless communications”, IEEE Journal on Selected Areas in Communications, 23(2), pp. 201–220, 2005. [118] M.-G. Di Benedetto, G. Giancola and M. D. Di Benedetto, “Introducing consciousness in UWB networks by hybrid modelling of admission control”, Mobile Networks and Applications, 11, pp. 521– 534, 2006. [119] S. Gezici, Z. Tian, G. B. Giannakis, H. Kobayashi, A. F. Molisch, H. V. Poor et al., “Localization via ultra-wideband radios”, IEEE Signal Processing Magazine, 22, pp. 70–84, 2005. [120] G. Bal and L. Ryzhik, “Time reversal and refocusing in random media”, SIAM Journal on Applied Mathematics, pp. 1475– 1498, 2003. [121] N. Bleistein, J. Cohen and J. Stockwell, Jr, “Mathematics of Multidimensional Seismic Imaging, Migration and Inversion”, Springer-Verlag Inc., New York, 2001. [122] L. Borcea, C. Tsogka, G. Papanicolaou and J. G. Berryman, “Imaging and time reversal in random media”, Inverse Problems, 18, pp. 1247– 1279, 2002. [123] D. J. Daniels, “Ground Penetrating Radar”, 2nd edition, IEE, London, UK, 2004. [124] S. Foo and S. Kashyap, “Cross-correlated back projection for UWB radar imaging”, Proc. IEEE Antennas and Propagation Society Symposium, pp. 1275– 1278, 2004. [125] R. Zetik, J. Sachs and R. Thom¨a, “Modified cross-correlation back projection for UWB imaging: numerical examples”, Proc. IEEE International Conference on Ultra-Wideband (ICU 2005), Z¨urich, Switzerland, Sep. 2005. [126] R. Zetik, J. Sachs and R. Thom¨a, “Modified cross-correlation back projection for UWB imaging: measurement examples”, Proc. DSPCOM, Kosice, Slovakia, 2005.

348

Technologies for the Wireless Future – Volume 3

[127] J. Shi and C. Tomasi, “Good features to track”, Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR 94), Jun. 1994. [128] J. Sachs, R. Zetik, J. Friedrich and P. Peyerl, “Autonomous orientation by ultra wideband sounding”, Proc. 9th International Conference on Electromagnetics in Advanced Applications (ICEAA 2005), Torino, Italy, Sep. 2005. [129] J. Sachs, M. Kmec, R. Zetik, P. Peyerl and P. Rauschenbach, “Ultra wideband radar assembly kit”, Proc. IEEE international Geoscience and Remote Sensing Symposium (IGARS 2005), Seoul, Korea, Jul. 2005. [130] S. Wood, “UWB standards”, WiMedia Alliance, white paper, Jun. 2006. [131] B. Pietrarca, G. Sasso, G. P. Perrucci, F. H. P. Fitzek and M. Katz, “Measurement Campaign on Connectivity of Mesh Networks formed by Mobile Devices. 2007. in MeshTech ’07 - First IEEE Internation Workshop on Enabling Technologies and Standards for Wireless Mesh Networking. IEEE. Pisa, Italy.” [132] M. Z. Win and R. A. Scholtz, “Impulse radio: How it works”, IEEE Commun. Letters, 2, pp. 36–38, Feb. 1998. [133] M. Z. Win and R. A. Scholtz, “Ultra-wide bandwidth time-hopping spread-spectrum impulse radio for wireless multiple access communications”, IEEE Transactions on Communications, 48, pp. 679–691, Apr. 2000. [134] A. F. Molisch, “Ultrawideband propagation channels – theory, measurement and models”, IEEE Transactions on Vehicular Technology, (invited paper), 54(5), pp. 1528– 1545, Sep. 2005. [135] A. F. Molisch, D. Cassioli, C. C. Chong, S. Emami, A. Fort, B. Kannan, J. Karedal, J. Kunisch, H. Schantz, K. Siwiak and M. Z. Win, “A comprehensive model for ultrawideband propagation channels”, IEEE Transactions on Antennas and Propagation, 54(11), pp. 3151– 3166, Nov. 2006. [136] IEEE Working Group 802.15.4a, “Draft specifications for IEEE 802.25.4a standard”, D7, Jan. 2007, at www.802wirelessworld.com, D. Goeckel and Q. Zhang: “Slightly Frequency-Shifted Reference Ultra-Wideband (UWB) Radio”, IEEE Transactions on Communications, 55(3), pp. 508–519, 2007. [137] R. R. S. Thom¨a, O. Hirsch, J. Sachs and R. Zetik, “UWB sensor networks for position location and imaging of objects and environments”, in Proc. European Conference on Antennas and Propagation (invited paper), Nov. 2007. [138] R. R. Zetik, J. Sachs and R. S. Thom¨a, “UWB short range radar sensing”, IEEE Instrumentation and Measurement Magazine, 10(2) pp. 39–45, 2007.

8 Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks Edited by Prof. Panagiotis Demestichas, George Dimitrakopoulos and Yiouli Kritikou (University of Piraeus, Greece)

8.1 Introduction Wireless communications attract significant research and development effort, reflected in the progress of work performed in international projects [1], as well as in the discussions in international fora [2]. This work results in a powerful, high-speed infrastructure that offers versatile solutions to the digital information society. In this context, the technological focus is on the cooperation and coexistence of legacy radio access technology (RAT) standards with currently emerging ones. The current wireless landscape is characterized by a plethora of RATs, which can be roughly classified into two major categories. The first one is the wireless wide area networking (WWAN) technologies, which includes, among others, 2G/2.5G/3G mobile communications [3], the IEEE 802.16 suite [4], WiMAX [5] and broadcasting technologies (DAB, DVB) [6]. Wireless short-range networks (WShRNs) fall within a second family, which includes wireless local and personal area networks (WLANs/WPANs), as well as wireless sensor networks (WSNs) [4], [7], [8]. This situation is depicted in Figure 8.1. Regarding the backbone network architecture, legacy [3] or modern paradigms [9] can be followed. Moreover, the evolution of wireless access networks is frequently referred to as B3G (beyond the 3rd generation) systems [1], [2]. In the B3G era, network operators (NOs) will have to address increased complexity with respect to today. Complexity derives from two main sources: on the one hand, there is the inevitable heterogeneity of the network and terminal infrastructure, and on the other, the user requirements that associate the B3G era with advanced services/applications, provided seamlessly and ubiquitously. Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

350

Technologies for the Wireless Future – Volume 3

B3G world

Services and applications

New Radio Interface Digital Broadcast IP based Core Network Cellular 2G/3G

Reconfigurable Segments

WLAN type WMAN

Figure 8.1 Overview of the wireless world in the B3G era

To meet these objectives, NOs have to deploy complex network topologies of heterogeneous nature. The different RATs will have to coexist, and be complementarily (and efficiently) exploited. Each RAT has different capabilities, in terms of capacity, coverage, mobility support, cost, etc. Therefore, each RAT is best suited for handling certain situations. In this respect, an NO will have to rely on different RATs for raising customer satisfaction, assigning the appropriate RAT to perform respective particular task, so as to achieve the required quality of service (QoS) levels, cost-effectively. QoS refers to performance (e.g. bit-rate, delay, etc.), availability (e.g. low blocking probability), reliability (e.g. low dropping or handover blocking probability), as well as security/safety (also indicated in [10]). An option for handling this complex situation is to design wireless B3G infrastructures by exploiting ‘cognitive networking’ capabilities [11], [12]. In general, cognitive systems are able to retain knowledge from previous interactions with the environment and determine their behaviour according to this knowledge, as well as other goals and policies, so as to adapt to external stimuli and optimize their performance. In the case of wireless networks, cognition expresses their ability to dynamically select their configuration, through management functionality that takes into account the context of operation (environment requirements and characteristics), goals and policies [13] (corresponding to principles), profiles (capabilities) and machine learning [14], [15] (for representing and managing knowledge and experience). As can be deduced from the above, cognitive networks consist of reconfigurable platforms and intelligent management functionality. In particular, they benefit from the existence of reconfigurable platforms that enable dynamic changes in their configuration, and they also require management mechanisms capable of finding the most appropriate configurations. Accordingly, the scope of this chapter is twofold: on one hand, to provide the basic characteristics of reconfigurable platforms and explain how they can support cognitive networks; on the other, to provide policy-based management and autonomic [12] radio resources for B3G infrastructures that operate in accordance with the cognitive networking paradigm. This chapter consists of five parts. After this introduction (compiled by P. Demestichas, G. Dimitrakopoulos and Y. Kritikou), Section 8.2 (written by P. Demestichas, G. Dimitrakopoulos, Y. Kritikou, D. Bourse, M. Muck and K. Moessner) provides an overview of cognitive wireless networks. Next, management is detailed in Section 8.3 (written by

Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks

351

E. Adamopoulou, K. Demestichas, Y. Kritikou, P. Goria, A. Trogolo, E. Buracchini, A. Dejonghe, J. Cranickx, L.V. der Perre, D. Bourse, M. Muck and K. Moessner). Then supplementary knowledge features in support of cognition are given in Section 8.4, before the conclusion is presented in Section 8.5.

8.2 Overview of Cognitive Wireless Networks 8.2.1 Operation Principles Cognitive networks have been proposed as a facilitator of the quest to offer seamless mobility to users while meeting their ever-increasing demands. To do so, they act differently to legacy systems, since they are able to adapt their operation by (proactively or reactively) responding to external stimuli. This is achieved, as mentioned earlier, by utilizing mechanisms that observe external conditions, retain valuable knowledge from interactions with the environment and plan their future actions accordingly. Their operation can be reflected in a feedback loop (see also 0), like the one shown in Figure 8.2. A basic cognition principle foresees that the network continuously monitors the environment, looking for potential changes that can affect its operation. Observations form the basis for initiating machine-based reasoning to see if the reconfiguration process should be invoked. Once the decision is taken, the network acts accordingly. This loop is repeated inside a machine learning process, which leads to cognition. The loop is guided by a set of goals, which take the observations into account in planning actions. At this point in time, a reasonable question may arise: ‘Which is the optimum way to manage the diverse entities that form part of a cognitive network?’ The answer to this question might be complex. A potential answer is provided in the next subsection. 8.2.2 High-level View of Management Functionality The radio access networks have been classically designed and deployed to cover the traffic demand of the planned services in a static approach and by means of manual configuration of network elements, considering the busy hour traffic in each geographical zone. However, the continuously increasing demand has also raised the need for the deployment of new technologies and networks, which have to be optimally planned and managed by choosing between finding new sites, co-locating sites or migrating to reconfigurable transceivers. Additionally, in the case of cognitive networks, novel functionality should efficiently plan and manage an ever-changing network, since it should adapt to external requirements that also change over time and space. Furthermore, since a cognitive network consists of numerous elements and terminals of highly heterogeneous natures, located in different places, a centralized management Goals & Policies Observation

Analysis

Planning

Feedback from gained knowledge and experience

Figure 8.2

Cognitive networks operation loop

Action

Technologies for the Wireless Future – Volume 3

Context

Optimization Process Behavior Configuration

Input Parameters

352

Profiles

Policies

Knowledge gained through experience

Figure 8.3 Problem description for the operation of cognitive networks

approach becomes prohibitively complex and inappropriate. Hence, distributed management approaches, relying on pertinent technologies, e.g. autonomic computing, are currently in focus (e.g. see [11]). This approach can offer scalability, stability and modularity (which provides low complexity). In this respect, this chapter aims to provide scalable answers to the question of managing (supporting) cognition. In light of the above, each element may be multi-standard. A subset of technologies is used, namely those that are most appropriate for the context of operation. Network layer reconfigurations accompany the changes at the lower layers. At this layer, each configuration includes the algorithms and parameters for routing and congestion control, and in general, the pattern for interconnection with other elements of the network. Additionally, reconfigurations can be extended to the applications layer, through specifying the QoS levels of the applications. In general, reconfigurations are software-defined. Specifically, each element is controlled by management functionality that has to solve a problem of cross-layer flavour. The general definition of cognitive networks implies some very advanced capabilities, which spring from the necessity to encompass reconfiguration (change in the behaviour of the segment, reflected in parameters/infrastructure variations) features, enhanced by cognition capabilities. On the other hand, such changes should be performed in the best possible way for NOs, and also for the end users. In this respect, regarding the management part, a modern research direction, in order to increase scalability and decrease complexity, is to comply with self-management paradigms, or in other words, to develop the management functionality in accordance with the autonomic computing principles [16], [17], [18]. Figure 8.3 provides the overall description of the management functionality proposed for managing a cognitive network segment. The proposed management mechanisms may refer to the various parts of the input, to the optimization process, to the knowledge features (which accompany all parts) or to the outputs of the management problem. In light of the above, the remaining sections of this chapter present several approaches for all parts of the problem, as shown in Figure 8.3.

8.3 Management Mechanisms for Cognitive Wireless Networks Beginning from the analysis of the previous section with respect to the management functionality for cognitive wireless networks, this section aims to analyze the parts of the problem as described before, concentrating on their operability, as well as on their knowledge features (see also Figure 8.3).

Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks

353

8.3.1 Context Acquisition This subsection presents the learning and adaptation method for robustly estimating the probability that the selection of reconfiguration c is associated with a certain achievable bit rate and coverage capability. Details of the method have been published in [23]. 8.3.1.1 Formulation through Bayesian Networks The Bayesian network that can be proposed for modeling the specified problem utilizes some random variables, e.g. ABR and COV, which represent the achievable bit rate and coverage, respectively. CFG is another random variable, representing configuration. CFG is the Bayesian network’s predictive attribute (node), while ABR and COV are the target attributes. The goal is the computation of the maximum value of the joint conditional probability. Thus, given a configuration CFG, we search for the values of ABR and COV that maximize the aforementioned joint conditional probability. These values constitute the robustly estimated (i.e. the most probable) values of ABR and COV for this configuration. With reference to the above, the desired probability is equivalent to the product of the conditional probabilities. Hence, for performing the computations, two independent conditional probability tables (CPTs) can be organized, one for each random variable, which in this case are ABR and COV. Each CPT refers to a particular RAT. Each column of the CPT refers to a specific configuration (i.e. RAT and carrier frequency). Each line of the CPT corresponds to an ABR value, i.e. a discrete set of potential ABR values has been defined. Each cell (intersection of line and column) provides the probability that the configuration (corresponding to the column) will achieve the potential ABR value (corresponding to the line). Given a configuration, the most probable value of ABR is the value that corresponds to the maximum conditional probability. 8.3.1.2 Solution: Learning and Adaptation In the previous subsection, we stated that the capabilities of configurations are modeled through the conditional probability tables (CPTs). The next step is to describe how to update the CPTs for illustrating this learning and adaptation process, which yields the robust methods for discovering the performance capabilities of candidate configurations. We focus on ABR, since the analysis for COV remains the same. The process takes into account the system’s measurements and, more specifically, the ‘distance’ (absolute difference) between each candidate value and the measured value. The parameter nf is a normalizing constant whose value can be computed by requiring all the ‘new’ probabilities to sum up to 1. The system converges when the most probable candidate ABR value (i.e. the one with the maximum probability) is reinforced, while the probabilities of the other candidate ABR values are either reduced or reinforced less. After convergence, we limit the number of consecutive updates that can be carried out on the probability values associated with each ABR value. This is done to assist fast adaptation to new conditions. For the same reason, we do not allow that a probability falls under a certain threshold. In such cases, the normalization factor, nf , is computed by requiring all the other ‘new’ probabilities. This method’s goal was to show how a cognitive radio system could acquire interference and capacity estimations. Secondly, by enhancing the above with a learning system, which is essential for obtaining a truly cognitive process, the proposed approach was to develop

354

Technologies for the Wireless Future – Volume 3

a robust probabilistic model for optimal prediction of the capabilities of alternative configurations, in terms of achievable bit rate. A short-term related future plan is to enrich the basic Bayesian model that has been described, by adding more nodes (random variables), including ‘coverage’ and ‘context’ (i.e. ‘traffic’ and ‘user mobility’). The overall future plan is to further employ probabilistic relationships and autonomic computing principles in the direction of realizing cognitive, wireless-access infrastructures. The goal is to develop an autonomic manager which will encompass the robust estimation scheme. The manager will consist of policies, context perception capabilities, reasoning algorithms, learning functionality and knowledge engineering, technologies for the representation of ontologies and semantics. All these will yield a system that hypothesizes on causes of a problem, and subsequently validates or falsifies the hypothesis.

8.3.1.3 Knowledge Features Context information is obtained through interactions with the environment, which lead to reasoning and perception, through appropriate machine learning techniques. A managed network element is thus able to gain knowledge from those interactions and be aware of the optimum way to handle a given context. Specifically, an NO could know whether at a certain time of a day a service should be provided through a specific RAT within the element. Alternatively, traffic conditions at a certain time or at a certain spot within the coverage area of an element could also direct a certain service provision manner. All in all, future decisions can be significantly facilitated through storing context information.

8.3.2 Profile Management 8.3.2.1 Overview of Profile Management Strategies This section contains some preliminary considerations for managing the profiles of end users, network elements and terminals in future systems, especially cognitive ones. Regarding the access point’s profile, an access point is assumed to utilize a set of transceivers. Each transceiver is capable of operating a set of RATs, though there is also a set of spectrum bands and/or specific frequency carriers with which each RAT can operate, due to regulation or for technological reasons. So, each transceiver has a set of candidate configurations, i.e. a combination of RATs and frequencies. Let it be noted that this part can also provide the access point’s profiles (capabilities) which are dispersed within the access point, in terms of permissible RATs/spectrum, services, etc. However, in this paper all terminals are assumed capable of applying the self-management of cognitive access points’ decisions. Regarding the users’ profiles, users are grouped into classes, each of which is characterized by specific preferences and requirements. These preferences are kept in special log files in the system’s database. Such log files store important information about the users’ preferences (such as software requirements, skills, level of knowledge, goals and expected outcomes from specific actions), their activity in the system, as well as the hardware (equipment) they most frequently use. Consequently, the log files are recovered every time the user makes a request for a service (application). This means that the system is in a position not only to recognize the user, but also to be aware of the user group they belong to and their personal history.

Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks

Servers Applications Services

355

Terminal Network Infrastructure

Figure 8.4 User–service communication, exploiting a system’s cognition for efficient and effective service delivery

Therefore, it is able to predict at some point the preferences of this specific user and the potential requests they are going to address to the system, and thus provide them with a certain set of services through network infrastructure, as depicted in Figure 8.4, aiming at satisfying their personal needs. Each service has a target QoS level and a set of RATs through which it can be offered. In addition, each service provided at a specific QoS level is associated with a level of importance (utility). 8.3.2.2 Knowledge Features The constant updates in the information provided by the ‘profiles’ part of the functionality lead to significant knowledge gains with respect to user, and also network element behaviour. This is essential in providing services of maximum quality, tailored to individual user needs. For example, network element capabilities can be associated with users of a certain class that require some service. This implies that the process of serving users in future cases may be facilitated and optimized through experience. 8.3.3 Policy-based Management 8.3.3.1 Policy Management Strategies In general, policies designate rules and functionality that should be followed in context handling. Policies are decided by the NO. Indicatively, a policy-based management approach can take the form shown in Figure 8.5. High-level NO policies refer to business strategies for achieving some preset goals. These goals are usually related to the maximization of the NO revenues, on either a short-term or a long-term basis. Moreover, NO agreements with other, cooperative NOs are also taken into account. For instance, as shown in Figure 8.5, the NO may select the offered services as well as the RATs through which they will be offered. This information usually depends on the general NO strategies. On the other hand, low-level NO policies are targeted at an optimal resource consumption. Indicatively, an NO may provide a policy to the cognitive infrastructure that prohibits the provision of a service at a certain quality level for users away from the transceivers, so as to reduce cost. Indicatively, the NO may select not to offer a specific service at some rate, for short-term reasons, usually associated with resources. High-level as well as low-level policies are designated by the NO, through the disposal of the requisite management functionality, a sample of which is provided in Figure 8.5, which depicts the capability of the NO to ‘add’, ‘edit’ and also ‘delete’ parts of a service, or the complete service offered to users, at a certain time period or location.

356

Technologies for the Wireless Future – Volume 3

Figure 8.5

Representation of policy-based management

8.3.3.2 Knowledge Features Cognitive features also lie in the ‘policies’ part of the management functionality. Specifically, the suitability and efficiency of different policies possess great importance in handling versatile contextual situations. For example, an NO may get to know, through experience, that the provision of a service to users at certain locations may lead to overloading, or to increased cost. Additionally, an NO could also exploit knowledge on a potential increase in revenues when providing users of a certain class/location with very high QoS levels. Consequently, learning the most optimum policy and the most appropriate goals to be achieved may become valuable for NOs in successfully (transparently, quickly and securely) handling difficult situations. 8.3.4 Configuration of Behaviour of Cognitive Infrastructures 8.3.4.1 Introduction The configuration of behaviour of cognitive infrastructures includes several aspects, such as the selection of the optimum configuration pattern, the download of software components, their validation and installation, as well as other issues that fall in the realm of the implementation of reconfiguration. This section contains exemplary data on the selection procedure for the optimum reconfiguration pattern, as well as for the download procedure of the necessary software components, during the process of reconfiguration.

Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks

357

8.3.4.2 Selection of Optimum Configuration Pattern The configuration of cognitive infrastructures should encompass several decisions targeted at the optimization of QoS provisioning. Indicative actions to be decided include the selection of operating RAT, spectrum band and/or carrier frequency, as well as other parameters (e.g. transmission power, modulation type). In addition, there will be actions related to the network elements’ interconnection, routing and congestion control. Finally, there will also be actions that involve the allocation of applications to the optimum QoS levels. Several solution approaches can be considered for tackling such a problem. A phased approached that can be utilized is described in the sequel. The first phase includes the division of the problem into several subproblems, which are subject to parallel processing. The subproblems depend on the available configurations (mostly allocation of RAT and spectrum to the transceivers of the CgAP, but also other operating parameters). The second and third phases aim at exploring the capabilities of each configuration in terms of achievable bit rates. This is performed in two stages, i.e. i) through a first allocation of the demand to the transceivers of the CgAP, according to given policies, and ii) by attempting to provide the highest possible quality to the demand. Finally, the fourth phase includes the selection of the most appropriate configuration to handle the given contextual situation. This is performed by rating the available configurations through their potential provision of desired QoS levels. Specifically, the best configuration is decided by utilizing an objective function (OF) associated with the maximization of the total users’ utility requiring the least changes on the already established configuration. Once the decisions are taken, there might be software components that need to be efficiently, transparently and securely downloaded. This is the subject of the next subsection. 8.3.4.3 Software Download The overall software download procedure is divided into distinct phases and each phase is characterized by a duration time; thus, in order to evaluate the efficiency of the overall radio software download procedure, it will be necessary to estimate the duration of the complete process. To this end, in this subsection a recall to the OTA download technique and protocol description is reported. This will help to focus attention on the particular aspects that could guarantee a spectrum-efficient download of the software. Over-the-air (OTA) Download The over-the-air (OTA) download is the software download technique with the maximum versatility. In this case, the software download operation occurs using a radio channel of a pre-existent cellular network or an additional infrastructure. Three different working modes are thus defined: 1. The traffic and control channels of the legacy cellular networks (GSM/GPRS or UMTS) are used. In this case the terminal has to be active on one of these systems and, using the channels provided by such standards, can receive the operative software related to another system. 2. A signaling or bootstrapping ‘universal channel’ is introduced. In this case, at switch on the terminal automatically tunes itself on this ‘universal channel’ and performs the operative software download related to the system present in that place.

358

Technologies for the Wireless Future – Volume 3

3. A combination of the two previous modes may also be envisaged: the negotiation phase of the download is performed, for example, using a legacy standard (first mode), while the software download procedure occurs with a specific channel dedicated to this operation (second mode). Download Procedure The radio software download procedure is divided into three distinct phases: 1. Pre-download Phase – this phase ensures that the radio software modules can be securely transferred on the basis of the current configuration, capabilities of the device and user requirements. This phase includes service discovery, mutual authentication, capability exchange and download acceptance exchange. 2. During-download Phase – this phase includes the transfer of software, the verification of its integrity and retransmission requests in case of errors. 3. Post-download Phase – this phase includes the installation of software, the in-situ testing, the device reconfiguration, the non-repudiation exchange and the recovery efforts in case of reconfiguration failures. All the above OTA download methodologies refer to the client/download path/server architecture. Below, the main steps of a generic procedure for download with any of the previously described methods are reported. The generic protocol for software download is client-server oriented, with the terminal assuming the client role and a generic node (software repository) assuming the server role. The main steps of the procedure are graphically reported in Figure 8.6 and described below: 1. Initiation – the terminal (or the server) triggers the start of the download procedure. In case the network or the terminal cannot operate the download (due to lack of resources or because of other actions with higher priority) the procedure is interrupted. 2. Authentication – in this phase, the terminal and the server authenticate each other. 3. Capability Exchange – the server communicates the information related to the software to be downloaded; the terminal verifies that software can be charged in the memory, installed and started, on the basis of its own characteristics and parameters. 4. Download Acceptance – the server communicates to the terminal the characteristics of the download (e.g. dimension and number of segments/blocks), of the installation and of the billing; the terminal, eventually with user interaction, states whether the server indication can be accepted. 5. Software Download and Integrity Test – the download of the operative software and data checks takes place; as the software is downloaded, an integrity test is run: the terminal requires the retransmission of radio blocks not correctly received. 6. Installation – during the installation step, billing and licensing data are provided by the server. 7. In-situ Testing – before running the new operative software, the terminal runs a test. 8. Non-repudiation – the terminal confirms to the server the correct installation; after the reception of the confirmation by the server, billing procedures are initiated.

Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks

359

Initiation

Authentication

Capability Exchange

Download Acceptance

Software Download and Integrity Test

Installation

In-situ Testing

Non-Repudiation Exchange

Failure

Figure 8.6 Software download procedure

8.3.4.4 Concluding Remarks In conclusion, the configuration of the behaviour of cognitive wireless infrastructures is of critical importance and should be carefully analyzed when considering future systems. This is justified through the fact that wireless communications are indeed migrating towards the B3G era, which can be efficiently realized through the exploitation of cognitive networking technologies. Intelligent, self-management functionality is necessary for directing networks that operate in accordance with the cognitive paradigm. In this respect, this subsection aimed at paving the way for detailed studies in the realm of decisions about the most appropriate configuration pattern, encompassing learning techniques. In conclusion, for the UMTS system the software download procedure has good performance in both patch and RAT download, thanks to the power control algorithm that permits

360

Technologies for the Wireless Future – Volume 3

Gap Energy requirement Energy available in battery Time

Figure 8.7 The energy gap is growing

maintenance of a constant C/I target at the receiver. In case of GSM/GPRS, the software download procedure strongly depends on the RRM scheme adopted. Such remarks should be carefully considered by network infrastructure designers. 8.3.5 Configuration of Behaviour of Cognitive Terminals Anything, anywhere, anytime: the idea is not new. Still, today ubiquitous broadband wireless communication bringing multimedia services is not yet available. One of the major bottlenecks is the need for software-defined radios (SDRs) for wireless terminals, which could enable the user to have access to a large variety of standards at low cost. The combination of the increasing need for functional flexibility in communication systems and the exploding cost of system-on-chip design will indeed make implementation of wireless standards on multipurpose reconfigurable radios the only viable option in the coming years. These devices being battery-powered, the performance requirements are coupled with severe constraints on energy efficiency. This is becoming a key concern; there exists a continuously growing gap between the available energy, resulting from battery technology evolution, and the steeply increasing energy requirements of emerging radio systems (Figure 8.7). Technology scaling, platform improvements and circuit design progress are not sufficient to bridge this energy gap. A clear need for holistic system-level strategies exists. Given the energy gap discussed above, a major challenge is to enable low-energy reconfigurable radio implementations, suitable for handheld multimedia terminals and competitive with fixed hardware implementations. To make such terminals a reality, a two-step approach is advocated. First, effective energy scalability is enabled in the design of the radio baseband and front end. Secondly, the scalability is exploited to achieve low-power operation by a cross-layer controller that follows at runtime the dynamics in the application requirements and propagation conditions (Figure 8.8). 8.3.5.1 Design Step to Enable Flexibility To enable the translation of functional flexibility into energy scalability, the reconfigurable radio (algorithms, architectures, components and circuits) should first be designed accordingly. For the reconfigurable digital baseband engine, one has to carefully trade off flexibility and energy efficiency: flexibility should only be introduced where its impact on the total average power is sufficiently low or where it offers a broad range of control options that can

361

Propagation conditions

Application requirements

Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks

t

t

Cross-layer optimized Run-time performance/ Energy manager Energy-scalable SDR Baseband Front-end

Figure 8.8 Energy-scalable SDRs achieve low-energy operation through cross-layer QoS and energy management

be exploited effectively later in the control step (targeted flexibility). The required subfunctions of the wireless modem should be designed according to their nature (i.e. control or data processing) and flexibility/energy efficiency requirements. This calls for heterogeneous multiprocessor system-on-chip (MPSoC) platforms. In [20] for instance, a heterogeneous MPSoC platform is proposed that builds on this concept of targeted flexibility. For the reconfigurable analogue front end, architectures and circuits should be designed for a broad range of requirements in carrier frequency, channel bandwidth and noise performance, with minimal penalty in power consumption, while also offering energy scalability. In [21] for instance, such a reconfigurable zero-IF analogue front end was implemented. All building blocks in the RF front end are equipped with configuration ‘knobs’ that allow them to adjust their performance to the requirements of the considered standards, but also to scale their energy consumption to the actual requirements. An example of such an energy-scalable component is the power amplifier introduced in [22]. The proposed circuit flexibility enables significant tradeoffs between the transmitter output power and linearity and the corresponding energy consumption. The resulting tradeoff between the link signal-to-noise-and-distortion ratio and the transmitter power consumption is shown in Figure 8.9 (left part), for path losses ranging from 60 to 90 dB. 8.3.5.2 Control Step to Exploit Flexibility To exploit flexibility for saving energy, it is mandatory to control the reconfigurable radio system as a function of the operation conditions. The key observation here is that wireless communications systems typically face very dynamic conditions (in terms of propagation environment and application requirements). By carefully adapting the system to these dynamics at runtime, capitalizing on the energy scalability discussed above, much energy can be saved compared to a conventional design. This problem has to be addressed from a cross-layer perspective, as measuring performance requires taking into account the characteristics of the protocol stack, whereas optimizing energy expenditure assumes detailed knowledge of the low-level radio hardware. As an example, the above energy-scalable radio was combined with smart cross-layer control for data transmission over an 802.11a WLAN link. An energy-efficient runtime radio link controller was designed based on a generic cross-layer optimization methodology. The

362

Technologies for the Wireless Future – Volume 3

Required SiNADlink

1 for Rate [Mbps]: 6 9 12 18 243648 54

85

PL 90dB PL 80dB PL 70dB

0.8

PL 60dB

Pdc [W]

0.7 0.6 0.5 0.4 0.3

PL 60dB PL 70dB PL 80dB PL 90dB

80 Average energy per bit (nJ)

0.9

75 70 65 60 55 50 45

0.2 −20 −15 −10 −5 0

5

10 15 20 25 30

SiNADlink [dB]

40

0

5

10

15

20

25

Average rate (Mbit/s)

Figure 8.9 Left: power consumption vs. link SINAD tradeoff enabled by scalable front end; Right: average net MAC data rate vs. energy efficiency tradeoff enabled by cross-layer energy management

resulting tradeoff between the average net data rate (on top of the MAC) and the energy efficiency is reported in Figure 8.9 (right part). When compared with traditional WLAN radio link control schemes, where the data is transmitted at the maximum achievable data rate and the transmitter is shut down when no data has to be sent, the proposed scheme can improve the energy efficiency by up to 40 %, by properly adapting to the rate requirements and channel conditions.

8.4 Supplementary Knowledge Features in Support of Cognition The solution of problems related to the configuration of behaviour of cognitive infrastructures could be facilitated if knowledge features were integrated therein. This has been constructively described regarding the individual parts of the problem anticipated. Additionally, this section shows why knowledge feature are needed and how they could facilitate problem solutions. The problems described in the previous section could be solved either at the network segment level (more centralized manner) or at the element level (more distributed manner). A significant advantage of the centralized functionality is that it anticipates problems through a holistic view. However, it requires an increased amount of time to solve them, since it has to tackle high complexity levels. Complexity derives from the fact that the configuration of behaviour of numerous elements (access points) falls within the realm of a centralized functionality and thus the situation to be resolved is of multiple difficulty, compared to the one faced at a single element level. Therefore, an only-realtime solution of the problem sounds almost unfeasible. The knowledge provided through the solutions achieved at the element level should be effectively exploited. This can be achieved by retaining a rating of the different configurations in a matrix. Searching within this matrix will reveal the optimum solution. A process that conforms to the above, which is based on machine learning, is analyzed below (and depicted on figure 8.10). The first step is to read the current contextual situation. This situation is then compared to some reference context situations that are available in the matrix. The matrix describes the

Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks

363

Current context

Context matching (pattern recognition, case-based reasoning, hash functions)

Do not find similar context

Find best configuration (solution from MCgWNS)

Find similar context

Find best available configuration from matrix (exploitation of SMCgAP solution)

Apply configuration

Apply configuration

Update matrix

Figure 8.10 Approach for obtaining knowledge useful for decisions at the network segment level

performance of several available configurations (e.g. utility-based OF values, as described before) in a number of given contexts. In other words, the matrix exploits the solutions at the element level. These solutions have revealed the efficiency of the alternative decisions to be taken regarding the element behaviour. The comparison procedure for the identification of a clearly- or almost-identical context situation can be based on techniques such as case-based reasoning or pattern recognition. If the current context comprises aspects that cannot be matched to the available reference contexts, the problem must be faced without the support of machine learning, just by finding

364

Technologies for the Wireless Future – Volume 3

the optimum configuration at the network segment level, applying the method presented for individual elements several times. At this point, the complexity increases significantly and this calls for the consideration of signaling and other cost factors, so as to reduce the overall number of alternative configurations that need to be tested. In any case, after the optimum configuration has been found, it can be implemented. Finally, the matrix has to be updated with the information retrieved by the current context-handling manner. In doing so, its information is gradually improved, leading to more efficient decisions. On the other hand, if the current context has been faced in the past, information has been kept in the matrix and is matched with the reference contexts available. This allows the most appropriate configuration to be selected. This might not be the ‘real’ best solution in terms of performance, but a lower-priority one, due to the fact that the decision made takes into account the previous state of the network segment. In addition, context matching will not always result in a level of 100 % similarity. In general, learning techniques can help the wireless network gradually obtain knowledge, so as to improve the efficiency of the decisions. Learning techniques are therefore currently under intense research focus and are expected to significantly improve.

8.5 Summary This chapter has presented some emerging concepts to support the advent of cognitive networks. In this respect, it has gone through the basic features of cognitive networks and their operation principles. It has focused on their management, by splitting the overall management process into some autonomous components, i.e. the context acquisition, the profiles management and the policy-based management. These components constitute the inputs to the adaptation process of cognitive infrastructures. The output lies in the configuration of the behaviour of the infrastructure (or the mobile terminal). Last but not least, it has outlined the knowledge features of the management approach components. In general, cognitive networking capabilities are a facilitator for the B3G vision, mostly targeting the minimization of complexity associated with heterogeneous environments (and how they can be managed). To this effect, work on cognitive networks is ongoing and is expected to address even more innovative aspects during the coming years.

8.6 Acknowledgements The following individuals contributed to this chapter: George Dimitrakopoulos, Panagiotis Demestichas and Yiouli Kritikou (University of Piraeus, Greece) Eugenia Adampoulou and Kostantinos Demestichas (National Technical University of Athens, Greece) Didier Bourse and Markus Muck (Motorola, France) Klaus Moessner (University of Surrey, UK) Paolo Goria, Alessandro Trogolo and Enrico Buracchini (Telecom Italia Labs) Antoine Dejonghe, Jan Craninckx and Liesbet Van der Perre (IMEC, Belgium)

References [1] FP6/IST project E2 R (End-to-End Reconfigurability), http://www.e2r.motlabs.com. [2] Wireless World Research Forum (WWRF), http://www.wireless-world-research.org.

Emerging Technologies to Support Reconfigurable Cognitive Wireless Networks

365

[3] Third (3rd) Generation Partnership Project (3GPP), http://www.3gpp.org. [4] [5] [6] [7]

Institute of Electrical and Electronics Engineers (IEEE), 802 standards, http://www.ieee802.org, 2007. WiMAX Forum, http://www.wimaxforum.org. Digital Video Broadcasting (DVB), http://www.dvb.org. Bluetooth, http://www.bluetooth.com.

[8] ZigBee Alliance, http://www.zigbee.org. [9] “Wireless mesh networking: theories, protocols and systems”, IEEE Wireless Commun. Mag., 13(2), special issue, Apr. 2006. [10] W. Hasselbring and R. Reussner, “Towards trustworthy software systems”, IEEE Computer, 29(4), Apr. 2006. [11] P. Demestichas, D. Boscovic, V. Stavroulaki, A. Lee and J. Strassner, “m@ANGEL: autonomic management platform for seamless wireless cognitive connectivity to the mobile Internet”, IEEE Commun. Mag., 44(6), Jun. 2006. [12] R. Thomas, L. DaSilva and A. MacKenzie, “Cognitive networks”, Proc. 1st IEEE Symposium on Dynamic Spectrum Access Networks 2005 (DySPAN 2005), pp. 352–360, Baltimore, USA, Nov. 2005. [13] J. Strassner, “Policy-based Network Management: Solutions for the Next Generation”, Morgan Kaufmann, 2005. [14] T. Mitchel, “Machine Learning”, McGraw-Hill, 1997. [15] R.E. Neapolitan, “Learning Bayesian Networks”, Prentice Hall, 2002. [16] J. Strassner, “Autonomic networking – theory and practice”, Proc. 9th IFIP/IEEE International Symposium on Network Management (IM 2005), Nice, France, May 2005. [17] J. Kephart and D. Chess, “The vision of autonomic computing”, IEEE Computer, 36(1), pp. 41–50, Jan. 2003. [18] M. Hinchey and R. Sterritt, “Self-managing software”, IEEE Computer, 39(2), pp. 107– 109, Feb. 2006. [19] C. Lepschy, G. Minerva, D. Minervini and F. Pascali, “GSM-GPRS radio access dimensionino”, Vehicular Technology Conference, 2001. [20] L. Van der Perre, B. Bougard, J. Craninckx, W. Dehaene, L. Hollevoet, M. Jayapala et al., “Architectures and circuits for software defined radios: scaling and scalability for low cost and low energy”, ISSCC 2007, San Francisco, USA, Feb. 2007. [21] J. Craninckx, M. Liu, D. Hauspiel, V. Giannini, T. Kim, J. Lee et al., “A fully reconfigurable software-defined radio transceiver in 0. 13 ¼m CMOS”, ISSCC 2007, San Francisco, USA, Feb. 2007. [22] B. Debaille, B. Bougard, G. Lenoir, G. Vandersteen and F. Catthoor, “Energy-scalable OFDM transmitter design”, Design Automation Conference (DAC), 2006. [23] E. Adamopoulou, K. Demestichas, P. Demestichas and M. Theologou, “Enhancing cognitive radio systems with robust reasoning”, International Journal of Communication Systems, Sep. 2007.

9 Methods for Spectrum Sharing Edited by Sudhir Dixit (Nokia Siemens Networks) sudhir.dixit@{ieee.org, nsn.com}

9.1 Introduction It seems evident that there will not be enough spectrum exclusively available for all wireless systems currently under investigation and development. Furthermore, the current utilization of the spectrum is quite inefficient; consequently, if properly used, there is no shortage of the spectrum that is at present available. Therefore, it is anticipated that more flexible use of spectrum and spectrum sharing between radio systems will be key enablers to facilitate the successful implementation of future systems. However, in order to manage and mitigate interference in both existing and future systems, and to ensure good user experience, it is essential to employ rules and etiquettes between systems sharing the same frequency band in a given geographical area. In addition, technical concepts need to be developed as means for the participating systems to obey the agreed rules. 9.1.1 Drivers for Spectrum Sharing and Spectrum Etiquette The mobile telecommunication industry has dramatically developed in the past decades. A trend is that the market has become heavily dependent on dominating operators and manufacturers, while small and medium size enterprises (SMEs) can hardly push their new technologies and products into the main market if they are not assimilated into the large companies. A potential method to change the current industrial structure is the emergence of alternative spectrum management regimes [1], such as the introduction of some relaxed spectrum etiquettes to let different radio access systems share spectrum. The most notable initiative in this area is the new regulatory framework published by the Federal Communications Commission (FCC) in 2002 [2]. The technical driver for spectrum sharing may come from the view to improve the efficiency of spectrum usage. The current spectrum management is that a specific system uniquely occupies a certain frequency band. Other mobile systems are not allowed to use the frequency band even though the frequency band is not well used at certain times, in certain areas. This leads to poor utilization of existing spectrum, as verified by many measurements. Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

368

Technologies for the Wireless Future – Volume 3

Measured Spectrum Occupancy Averaged over Six Locations PLM, Amateur, others: 30–54 MHz TV 2–6, RC: 54–88 MHz Air traffic Control, Aero Nav: 108–138 MHz Fixed Mobile, Amateur, others: 138–174 MHz TV 7–13: 174–216 MHz Maritime Mobile, Amateur, others: 216–225 MHz Fixed Mobile, Aero, others: 225–406 MHz Amateur, Fixed, Mobile, Radiolocation: 406–470 MHz TV 14–20: 470–512 MHz TV 21–36: 512–608 MHz TV 37–51: 608–698 MHz TV 52–69: 698–806 MHz Cell phone and SMR: 806–902 MHz Unlicensed: 902–928 MHz Paging, SMS, Fixed, BX Aux, and FMS: 928–960 MHz IFF, TACAN, GPS, others: 960–1240 MHz Amateur: 1240–1300 MHz Aero Radar, Military: 1300–1400 MHz Space/Satellite, Fixed Mobile, Telemetry: 1400–1525 MHz Mobile Satellite, GPS, Meteorologicial: 1525–1710 MHz Fixed, Fixed Mobile: 1710–1850 MHz PCS, Asyn, Iso: 1850–1990 MHz TV Aux: 1990–2110 MHz Common Carriers, Private, MDS: 2110–2200 MHz Space Operation, Fixed: 2200–2300 MHz Amateur, WCS, DARS: 2300–2360 MHz Telemetry: 2360–2390 MHz U-PCS, ISM (Unlicensed): 2390–2500 MHz ITFS, MMDS: 2500–2686 MHz Surveillance Radar: 2686–2900 MHz ©Shared Spectrum Company, 2005

0.0%

25.0%

50.0% Spectrum Occupancy

75.0%

100.0%

Figure 9.1 Spectrum occupancy measurement results

Figure 9.1 is the average result measured over six locations in the US by the Shared Spectrum Company [3], which shows that the average spectrum occupancy is less than 6 %, with the maximum being 13.1 % in New York City. Therefore, development of new methods to improve the spectrum occupancy, such as spectrum sharing, might be more important than creation of new frequency bands. 9.1.2 A High-level Functional Model of Spectrum Sharing Management Currently there are a number of applications trying to realize the concept of spectrum sharing [4]. Cognitive radio, however, is known as the most intelligent and promising technique in solving the problem of spectrum sharing. Functionally, cognitive radio can be generalized in four steps: sense radio environment, characterize rapid waveform, react to the environment, and adapt to transit network [5], [6] (see Figure 9.2). Find ranges of frequencies not in use (white spaces) (Characterize)

Dynamically scan and sense the frequency range (Scan and Sense)

Negotiate with the controller which set of frequencies to use (Negotiate)

Adapt the network to new spectrum allocation (Adapt)

Decide on the best course of action (Decide)

Synthesize waveform corresponding to agreed white spaces (Synthesize)

Figure 9.2 A functional illustration of the steps involved in spectrum identification and allocation

Methods for Spectrum Sharing

369

Table 9.1 Spectrum sharing methods Spectrum Sharing Method

Distributed Assignment

Centralized Assignment (Network Sharing)

Within RAT

Within a RAT

Distributed Opportunistic Identification spectrum use:

ž Underlay (interference temperature) ž Use of white spaces

Centralized Control channel Identification containing information about locally available spectrum (e.g. white spaces)

Between Multiple RATs

Between Multiple RATs

Opportunistic spectrum use:

Based on Based on monitoring and monitoring and databases that ž Underlay (inter- databases that can be accessed can be accessed ference temperby all networks by all networks ature)

ž Use of white

spaces(See Section 9.2.4.2 for details on interference temperature and white spaces) Control channel containing information about locally available spectrum (e.g. white spaces)

Based on a Based on a dynamic dynamic channel channel allocation table allocation table common to all common to all access access networks networks

At a more general level, a spectrum sharing system is typically made up of two key functions: identification and allocation. Both can be designed as distributed or centralized, resulting in four different types of spectrum use concept. This can be illustrated by Table 9.1.

9.2 Spectrum Sharing Categories Based on Centralized and Distributed Approaches There are many ways to categorize the various techniques. In this section we take the approach that the spectrum sharing is relevant only for the radio access networks (RANs) and not for the backhaul core networks, and also that the existing identified concepts for spectrum sharing between RANs can be categorized based on the amount of inter- and intra-system coordination they require from the participating RANs [6]. As a separate category we could add spectrum sharing for uncoordinated systems, covering the current systems operating in the license-exempt bands (e.g. WiFi, Bluetooth) as well as UWB. It is worth noticing, though, that no technical concepts for spectrum sharing exist for those systems. The different spectrum sharing concepts for each of the categories above are separately described in the following sections of this chapter.

370

Technologies for the Wireless Future – Volume 3

9.2.1 Centralized Spectrum Sharing between Cooperative Access Networks The spectrum sharing concepts in this category are based on a separate entity above the radio access networks in the hierarchy, i.e. the spectrum manager. Depending on the concept, the spectrum manager can operate in different timescales. The regulator sets the limits in which the spectrum manager may operate, and these limits are not expected to vary in time. The concepts belonging to this category aim at dynamically managing the radio spectrum between overlapping radio access networks in a given area. The aim is to balance the spectrum use between the networks and improve the overall spectrum efficiency in the area. In this type of spectrum sharing it is required that the spectrum assignment between operators or even different access technologies is not fixed but may change in time. The radio access networks may or may not be based on the same access technology, and from that fact they can be further divided two subcategories as follows: 1. Dynamic spectrum management between different RATs. 2. Dynamic spectrum management between collaborative RANs based on a single RAT. In the first subcategory, the spectrum is not dedicated to any particular access technology, but it can be dynamically assigned (reallocated) to different systems depending on need. The spectrum is shared between operators and/or between different RATs. In the second subcategory, the spectrum licensed by the different network operators for a single RAT is pooled entirely or partly, and all networks utilize the common pool of resources, managed by the spectrum manager. It is therefore required that the operators have the right to lease their spectrum to be used by the other operators. The two subcategories are described separately below.

9.2.1.1 Spectrum Sharing between Different Radio Access Technologies This category of spectrum sharing concepts relates to radio communications in a heterogeneous radio environment, in which several RATs exist and the equipment is capable of operation in multiple frequency bands. In the optimal case the terminals and base stations are also capable of communication using multiple different RATs, and can be controlled to select the RAT to be used. The different RATs are, for example, GSM, WCDMA and WLAN. The basic processes required in this spectrum sharing category are illustrated in Figure 9.3. The involved RANs provide information about their current load and predicted near-future capacity requirement to the spectrum manager, which assigns spectrum to the networks based on certain predefined rules and the interference situation in the considered area. The spectrum manager has all the relevant information about the participating radio networks, i.e. locations of the BTS, transmit powers, antenna gains, service type, etc. The spectrum manager also has the information about the amount of average spectrum that should be assigned to each of the networks to ensure fair decisions. Alternatively, in an extreme case, the spectrum can be sold in online auctions to the RAN that is willing to pay the most. Related Activities This type of flexible spectrum use has been studied previously in the IST DRiVE and IST OverDRiVE projects. A brief overview of the two projects is presented below.

Methods for Spectrum Sharing

371

Spectrum manager

Spectrum management layer Resource requests

Resource requests

Network control layer RNC

Resource assignmentes AP

Resource assignmentes

BTS

UE Network access layer

Figure 9.3 Dynamic spectrum management between different radio access technologies

IST DRiVE (2000–2002) The European project DRiVE (Dynamic Radio for IP Services in Vehicular Environments) [7] investigated novel methods for dynamic spectrum allocation (DSA) in a multiradio environment. The overall objective of the DRiVE project was to enable spectrum-efficient high-quality wireless IP in a heterogeneous multiradio environment to deliver in-vehicle multimedia services. The DRiVE project addressed the convergence of cellular and broadcast networks. The work carried out in the DRiVE project serves as a foundation for the research on sharing and flexible spectrum use done in WINNER WP6 [8]. IST OverDRiVE (2002–2004) The European research project OverDRiVE (Spectrum Efficient Uni- and Multicast Over Dynamic Radio Networks in Vehicular Environments) [9] was the follow-on project of DRiVE and aimed at UMTS enhancements and coordination of existing radio networks into a hybrid network to ensure spectrum-efficient provisioning of mobile multimedia services. An IPv6-based architecture enables interworking of cellular and broadcast networks in a common frequency range with dynamic spectrum allocation. The project objective is to enable and demonstrate the delivery of spectrum-efficient multi- and unicast services to vehicles. The key OverDRiVE research issues were: 1. To improve spectrum efficiency by system coexistence in one frequency band and by dynamic spectrum allocation (DSA). 2. To enable mobile multicast by UMTS enhancements and multiradio multicast group management. 3. To develop a vehicular router that supports roaming into the intra-vehicular area network [9].

372

Technologies for the Wireless Future – Volume 3

IST E2 R (2004–) The Work Package 5 (WP5) of the European research project E2 R (End-to-End Reconfigurability) aims at developing mechanisms for dynamic allocation of radio resources. This requires research into combining reconfigurable technology and support structures (e.g. cognitive radio, joint radio resource management (JRRM) and flexible network planning) with novel resource management techniques that are capable of controling the complete spectrum in a local area. Deployment of such technology requires a new approach to regulation and economics of spectrum. Hence the second major aim of this task is to develop, based on the results of the research (which will be in tight collaboration with national regulatory bodies and operators), new options and mechanisms to enable more progressive spectrum regulation and market-based approaches, and to facilitate a more efficient resource usage [10]. Benefits and Drawbacks of this Type of Concept This spectrum sharing category is independent of radio access technology, which makes it an attractive solution for a heterogeneous multiradio environment. The clear benefit of centralized dynamic spectrum management between radio access networks based on different RATs is that under-utilized spectrum can be taken into use, choosing always the most spectrally efficient transport system. Examples of potentially feasible load balancing schemes achieved by different types of services are cellular vs. short-range (spatial) and cellular vs. broadcast (temporal). The greatest challenge of this concept is the increased interference. Due to previously nonexistent spectrum neighbourhoods, new adjacent channel interference scenarios need to be considered. There is a need for spectral separation (guard bands) or spatial separation (coordination distances) between RANs employing different RATs. Difficulties in interference mitigation arise especially in area borders. Another challenge of this concept is the operators’ unwillingness to cooperate unless a single operator operates the different RATs. Business models between different types of service provider need to be considered. 9.2.1.2 Spectrum Sharing within One Radio Access Technology This spectrum sharing concept relates to a situation in which there are several RANs in an overlapping area based on a single access technology. The spectrum licensed to the different operators is pooled either fully or partially so that part of the spectrum is still dedicated to each operator. A spectrum manager controls the usage of the pooled spectrum. Figure 9.4 illustrates this flexible spectrum use (FSU) concept. The figure is similar to that of the previous category, but a significant difference is that part of the radio network controller (RNC) duties can be assigned to the central controller, and the RNCs of the RANs are transparent to the central network, controlled so that it has the full information of all the participating RANs. Also, RANs without a centralized network controller can be involved as long as they have access to the centralized network controller. This makes deployment of small networks feasible. Benefits and Drawbacks A significant benefit of this approach over the previous one is that the interference becomes much easier to control due to the similar air interfaces of the networks. To avoid excess guard bands the spectrum use should be coordinated among the operators. Another benefit of this concept is that the network architecture need not be changed, but the networks of different operators can be combined as one big network in the network control

Methods for Spectrum Sharing

373

Central network controller

Super network control layer Resource assignments

Resource requests

Network control layer RNC

RNC

BTS

Operator 1 RAN

Operator 3 RAN

Operator 2 RAN

Network access layer BTS

BTS

Operator 3 Relay link

UE UE

UE

Figure 9.4 Dynamic spectrum management between operators inside a single RAT

layer. It is possible to further extend this concept to virtual networks where common base station sites (or even hardware) are shared among the operators. The challenges of this concept are the operators’ unwillingness to cooperate and the need of a responsible entity to maintain the central network controller. 9.2.2 Distributed Spectrum Sharing between Cooperative Access Networks Distributed spectrum sharing between cooperative access networks is another concept for flexibly sharing the spectrum, which utilizes a separate spectrum manager entity in its operation. Architecture-wise, this approach is similar to other frequency manager-based approaches, but conceptually it can be seen to be different. The essential part of this spectrum use category is a spectrum manager, which will not assign frequencies directly, but acts as a spectrum resource status registry and information facilitator. This is somewhat different from the actual spectrum manager-based scenarios, where the spectrum manager very actively participates in frequency assignment, and in fact orchestrates the whole flexible spectrum concept. In other words, the spectrum manager in this context only receives spectrum utilization status updates, processes information, and provides it to underlying networks or radio systems. The networks and systems themselves autonomously identify available spectrum resources based on the information provided by the spectrum registry, and assign frequencies in a cooperative fashion. In this spectrum sharing category, the network control layer of each individual network is active and has an essential role in frequency assignment. This responsibility is distributed between different networks, which can be, but are not necessarily, of the same system or radio access type. The spectrum manager itself in this context can be implemented in very many different ways, but can be mostly seen as a network database server. The information could be transferred via a restricted access network with some common, transparent network protocol such as TCP/IP.

374

Technologies for the Wireless Future – Volume 3

9.2.2.1 Spectrum Sharing between Different Radio Access Technologies Related Activities IST WINNER (2004–2007) The European research project WINNER (Wireless world INitiative NEw Radio) [8] has developed methods for efficient and flexible spectrum use and spectrum sharing in a new B3G radio system. Part of the work targeted at developing mechanisms for resource sharing between the different radio access modes, but also for sharing between WINNER network operators and with other systems with WINNER functioning either as primary or secondary system [11], [12]. As noted before, spectrum sharing can be considered from several possible points of view. In WINNER, the different types of sharing are distinguished by the types of system involved, i.e. either intra-system sharing or inter-system sharing. Of these two, inter-system sharing is in the scope of this section, and intra-system sharing is in the scope of Section 9.2.2.2. Both scenarios are considered in WINNER, and the current WINNER approach is treated in more detail in Section 9.3.1. Inter-system sharing involves systems employing different RATs belonging to possibly very diverse applications. The sharing capabilities of the systems may differ substantially and may be very limited. Inter-system sharing is seen in WINNER to provide an important option for using shared spectrum as capacity enhancement for the system. However, the operation in shared spectrum is subject to a number of constraints which cannot be influenced. To meet the B3G system requirements, especially QoS guarantees, operation in inter-system shared spectrum is considered in WINNER only for capacity enhancements. Benefits and Drawbacks Benefits of spectrum sharing among compatible systems providing diverse services can be easily recognized. If two or more dissimilar systems are to be operated in the same frequency band, the basic possibilities for peaceful coexistence on the same band are separation in either time or space. Other approaches, e.g. from an information-theoretical perspective, have to rely on very strong assumptions, which might not be feasible in practice [13]. Another limiting assumption is that no sophisticated signaling between the B3G and legacy systems is possible in general. Depending on the regulatory rules governing the shared bands, different sharing scenarios are possible. We focus on so-called vertical sharing, where the B3G system is in the secondary and the other system in the primary position. In such a scenario, the B3G system has to control its emissions to avoid interference with the primary system. This scenario may occur when a B3G system is deployed in a band already allocated to a legacy system, and the new system may not cause interference towards the established one. To attain the envisioned gains in practice, spectrum sharing needs to be efficient and reliable. This depends critically on radio resource management (RRM) that controls the sharing. 9.2.2.2 Spectrum Sharing within One Radio Access Technology Related Activities IST WINNER (2004–2007) Intra-system sharing, also referred to as flexible spectrum use (FSU) in WINNER, involves RANs employing the same RAT with dedicated functions for inter-RAN coordination.

Methods for Spectrum Sharing

375

The sharing capabilities are part of the system requirements and thus are woven into the system design. In the context of this chapter, such sharing occurs between multiple WINNER RANs. The spectrum manager corresponds to the central database in the WINNER terminology. In addition to the previously described architecture (Figure 9.2), there is also direct signaling between the WINNER RANs for the FSU. The considered intra-system sharing approach is distributed, whereby the spectrum manager does not have control over the networks but is centralized such that there is a central entity within a network that controls the flexible spectrum use. Spectral resources assigned to the WINNER networks are divided into two categories: i) resources assigned to a certain network with indisputable priority, to guarantee basic operation of network; and ii) common pool resources available to all networks. The division aims at combining the benefits of both approaches. Balance between the categories can be changed, even dynamically, which provides another degree of freedom to the configurability of the system. It may be needed, for example, to support differences in the used spectrum policies. Benefits and Drawbacks Significant advantages can be obtained when the spectrum is shared between RANs using the same RAT. Most importantly, flexible spectrum use enhances spectral scalability of the system. This allows deployment of multiple RANs at the launch of the system, even when spectrum is made available gradually according to increasing traffic demands. Such flexibility may turn out to be of particular importance for B3G systems requiring wide spectrum bands on frequencies suitable for efficient vehicular communications, that is, below 6 GHz [14]. Spectral scalability also facilitates versatile operation of networks, for example, with some operators providing more focused services and coverage than others. The function also allows that spectrum assignments are adapted to reflect the changes to the number of subscribers as well as to daily load patterns. Like the spectrum sharing between different RATs, the spectrum sharing gains within one RAT depend critically on RRM that controls the sharing. 9.2.2.3 Benefits and Drawbacks General benefits of this approach are the following: 1. The spectrum management infrastructure common to all systems can be particularly lightweight, since it does not participate actively in the frequency resource assignment process. 2. Spectrum management is not dependent on any radio system or radio access technology. Different systems can be adopted to the spectrum management scheme. 3. The add-on property of the spectrum manager allows easier implementation even on top of existing systems or infrastructure. In essence, different networks or systems can: ž ž ž ž

Be based on their capabilities. Be based on available spectrum resource information obtained from the spectrum registry. Be based on some predefined rules. Autonomously assign and utilize available frequency resources.

376

Technologies for the Wireless Future – Volume 3

It is important to note the system capability to utilize particular spectrum resources in this context. The spectrum management method, or the spectrum registry operation, does not in any way limit or set any requirements on the system capability to utilize certain frequencies. In particular, the systems can be anything from fixed waveform systems with fixed channel spacing to SDR technology-oriented ones with arbitrary waveforms. The spectrum registry informs about vacant spectrum, and it depends on the individual system’s capability to what extent it is able to utilize it. In this sense, technologically advanced systems would have an advantage over other systems due to their more flexible ability to access spectrum. As a drawback, this spectrum sharing approach is somewhat dependent on the cooperation between networks and systems. Additional policies, rules or cost mechanisms have to be in place in order to prevent systems unnecessarily assigning spectrum resources to themselves. Since the actual frequency assignment decision mechanism is distributed between networks and systems, updating the policies when a system is already in use is not straightforward. 9.2.3 Secondary Networks with Central Controller 9.2.3.1 Architecture Proposals have been made for secondary networks which operate on the unused channels of the incumbent (primary) networks without causing harmful interference to them. A recent proposal in this category is the use of vacant TV channels for broadband access. In the proposed secondary system the base station is the master, which commands the customer devices to undertake spectrum sensing and give the sensing results to the base station. The base station selects the channel to be used. 9.2.3.2 Proposal of the FCC In the Notice of Proposed Rule Making (NPRM) FCC 04–113 [15] the FCC proposed to allow unlicensed radio transmitters to operate in the broadcast television spectrum at locations where that spectrum is not being used. Licensed users to be protected from interference are analogue and digital TV reception, wireless microphones, and utility/commercial use in some metropolitan areas. If a licensed device starts operation on a previously vacant channel, the unlicensed device should cease operation on that channel. The unlicensed devices should use one or multiple 6 MHz TV channels. Two unlicensed systems are proposed: ž Fixed/access devices, Tx power max 1 W and max 4 W EIRP. ž Personal/portable devices, Tx power max 100 mW and max 400 mW EIRP. The FCC proposed the following techniques for determining TV channel availability. 1. Announcements The FCC proposed that the TV stations would send their coordinates and ranges over the radio medium to the secondary nodes. 2. Databases As an alternative or complementary mechanism to the announcements, the FCC proposed the creation of databases which contain lists of the coordinates and ranges of TV stations. The base stations of the secondary networks would read the databases to aid in decision making about the vacant TV channels. The drawbacks of the database approach, as indicated by IEEE

Methods for Spectrum Sharing

377

Primary network

Internet Base Station Secondary network

Terminals

Figure 9.5 Architecture of the secondary wireless access system of IEEE 802.22

802.18, include the potential inaccuracies in the databases and possibly the lack of models of mountains and other propagation obstacles. 3. GPS The FCC proposed the use of GPS techniques for the secondary nodes to determine their locations. After having received information on the locations and ranges of the TV stations and the secondary terminals, the base station would determine the vacant channels for the secondary network. 4. Spectrum Sensing The FCC proposed that the secondary nodes should use spectrum sensing to determine the vacant channels. In particular, it is necessary to sense those licensed devices that are not listed in databases and do not announce their location. The FCC recognized the existence of the hidden node problem, which arises when the secondary node does not sense a TV station due to propagation obstacles. As a result, the secondary node might transmit on a used TV channel, possibly causing harmful interference to a nearby TV receiver. The way the unlicensed devices share the spectrum between themselves could be based on future voluntary standards. The FCC’s role is not known. The FCC has received comments from various parties, including IEEE 802.18 Radio Regulatory TAG [16]. The view of 802.18 was that the database approach is insufficient since the database may not be up to date and geographical obstacles (mountains) may not be accurately modeled. 9.2.3.3 Wireless Regional Area Network (WRAN) Standardization in IEEE 802.22 IEEE 802.22 specifies the fixed/access system proposed by the FCC. It calls the system ‘wireless regional area network’. It prepared functional requirements for the 802.22 WRAN standard in September 2005. Some points from the requirement specification are described below.

378

Technologies for the Wireless Future – Volume 3

The system is meant for broadband access in rural areas (Figure 9.5). The system consists of a base station (BS) and customer premises equipments (CPEs) connected to the BS. With a BS antenna height of 75–150 m, the maximum distance between BS and CPE is typically 30–40 km, but in special cases it could be 100 km. The system should operate up to a distance of 100 km. The minimum bit rate for a CPE is 1.5 Mbps in the downlink and 384 kbps in the uplink direction. Television transmission uses horizontal polarization, and this has led to the following polarizations in the WRAN system. The CPE transmits and BS receives on vertical polarization to reduce interference to nearby TV receivers. In addition, the CPE uses directional antenna. In case of WRAN being an FDD system, the BS transmits and CPE receives on horizontal polarization in order to have isolation between the transmission directions. The detection of primary signals is mainly based on sensing. However, the channels used by TV stations should be configurable in the base station. The BS and CPE sense the spectrum using omnidirectional antennas with a gain of minimum 0 dBi in both polarizations. The primary signals will be recognized individually. Sensing the digital TV 8-VSB signal is based on the existence of a fixed pilot, which is sensed through a 10 kHz bandpass filter, resulting in the – 116 dBm sensing threshold (this is below 6 MHz noise level). Sensing the analogue TV signal is based on the video and audio carriers. The European DVB signal is sensed by identifying the missing OFDM subcarrier in the middle of the spectrum. For wireless microphones, 802.22 will specify a beacon signal to ease their sensing. The WRAN system is fully controlled by the base station, including the sensing instructions to the CPEs. There is a silent period during sensing. The base station collects all sensing information and makes the decision on the channels to be used. The appearance of a primary signal in the channel the WRAN system is using should be detected, and the WRAN system should move to a vacant channel. The base stations should be registered so that possible harmful interference to licensed devices can be easily resolved. To limit the interference to TV receivers, the maximum transmit power of the CPE is 4 W EIRP and the minimum distance between the antennas of the CPE and TV receiver is 10 m. To make efficient sensing possible, the CPE antenna should be mounted 10 m above the ground. The TDD/FDD question is open. If TDD or half-duplex FDD is used (CPE has a duplex switch) then the base station could also send on vertical polarization since polarization isolation at the CPE is not needed. The next phase in the schedule of WG 802.22 is the development of the standard. 9.2.4 Distributed Spectrum Sharing between Noncooperative Access Networks In this section, we focus on distributed identification and distributed allocation of the spectrum for two types of systems that are commonly in use: license-exempt systems and secondary systems. The other three types of spectrum sharing concepts have been described in the earlier sections of this paper. In a truly distributed system, both identification and allocation happen without any centralized coordination. A nonactive portion of a spectrum or a channel (or a set of channels) is identified by a user when needed, and it remains in use till it is released. Once released, the spectrum can be used by some other user. An important issue is avoiding collision when two or more users sense the same unused portion of the spectrum.

Methods for Spectrum Sharing

379

Distributed spectrum sharing systems can be further classified based on what types of system (license-exempt or secondary) and what types of RAT (single- or multiradio environment) they use. 1. License-exempt systems: 1. Spectrum sharing between different RATs. 2. Spectrum sharing inside one RAT. 2. Secondary systems: 1. Spectrum sharing between different RATs of one or more operators. 2. Spectrum sharing inside one RAT of one or more operators. 9.2.4.1 License-exempt Distributed Systems Systems that operate in the license-exempt band of the spectrum tend to have short range. Such systems can either be ad hoc peer-to-peer (P2P) systems or infrastructure- (access point-) based systems. (It should be noted that the current operators of the license-exempt systems who are deploying networks in the same geographical area are developing some guidelines/etiquettes to avoid interference). Typically, such systems use common access technology (not license-exempt heterogeneous technologies), i.e. one specific RAT, but consist of independent systems, such as multiple WLANs. At a more general level, nodes listen to the band and exchange information about the locally available frequencies. Thus, each node maintains a database of used frequencies along with other usage data in its neighbourhood, and first sends an ‘intent to use a certain part of the spectrum or one more channels’ notification signal to the neighbouring nodes. This is done over a common broadcast channel. The main issues in this approach are: 1. Collision avoidance when more than one node attempts to select the same channel. 2. Support for multicasting on the same frequency channel. 3. How to deal with varying propagation delays between the nodes when using the same control channel to share the channel usage data – require some sort of slot reservation and timing synchronization to manage access to the common channel. In the license-exempt distributed systems category, there are potentially many approaches. Here, we describe two such techniques that have been reported in the published literature. Techniques for License-exempt Distributed Systems Dynamic Selection of Frequencies with Low Occupancy In this approach [17], Figure 9.6, frequencies with low occupancy are dynamically selected such that their search is minimized and so is the risk of a large number of nodes choosing the same frequency. Each node maintains and updates in real time a table of the prior use and loading of all the frequencies and their threshold of use, which helps select one or more frequencies quickly. Each node maintains a table of frequency usage, when they were last used, along with which ones were heavily loaded, routing table, and the loading on each channel. All this information is used by the node to select a new channel or to switch to a new channel when the loading threshold of the channel currently in use has been exceeded. This

380

Technologies for the Wireless Future – Volume 3

Full mesh between nodes or broadcast sensed open spectrum Each node listens to every other node in its neighborhood

Figure 9.6 All nodes listen to the other nodes and maintain a database of all channels used, when used, by whom, and how much each is loaded. This information is used by the nodes to select a channel when there is a need to send, or to switch to a new channel when the current one is heavily loaded

technique can be used in ad hoc networks, wide-range cellular and license-exempt public systems. The algorithms could be implemented in software in the terminals, access points and/or base stations. The following improvements, among others, are needed in this approach: ž A solution to avoid collision when more than one node attempts to select the same channel. ž Support for multicasting. ž A way of dealing with varying propagation delays between the nodes when using the same control channel to share the channel usage data and when no protocol is provided for the same. Master-slave Configurations In master-slave P2P short-range systems, many configurations are possible. In one such implementation [18,19], Figure 9.7, the master node polls all the slave nodes. The slave nodes monitor the spectrum on a continuous basis, and the node which wishes to transmit responds to the polling with a request to transmit in open spectrum. If the master agrees to the spectrum request positively, the slave can begin transmission in the open spectrum. Here, the signaling is done over dedicated bi-directional signaling channels. Any node can be a master or a slave. However, this approach would need to address the interference problem that would be experienced in a multimode environment or when more than one system is operating. Also, the issue of collision for the same resource will need to be addressed. 9.2.4.2 Secondary Distributed Systems Secondary spectrum use systems operate simultaneously with the primary systems. ‘Primary system’ refers to a network system where an operator owns the spectrum for a certain service; in a secondary system, the unused portions of the same primary spectrum are open to be used by either another service offered by the same operator, a different operator or any other user who is not a subscriber of the primary operator, such that any interference with the primary

Methods for Spectrum Sharing

381

Slave

Master

Search for open spectrum

Master polls slave

Search for open spectrum

Reply to send data in unused part of spectrum Grants permission to send data in requested open spectrum

Transfer of actual data

Figure 9.7

One example of master-slave configuration

system is constrained to an acceptable level. In a secondary distributed system, nodes do the sensing independently to identify the leftover bands from the primary system and fill in those gaps in an opportunistic way, such that the transmitted power is constrained and the synthesized waveform is shaped so as to limit the interference with the primary system to an agreed level. An example of an implementation is the ‘interference temperature’ concept proposed by the FCC. Some major issues with this approach are as follows: 1. As users come and go, the noise floor and the interference will change with time; the issue is, how does the synthesized waveform adapt to these changes? This could be a major problem when working with the legacy systems, which can turn on whenever and wherever. 2. Collision avoidance when more than one node discovers the same empty space and tries to transmit at the same time using sometimes overlapping white spaces or holes. 3. Methods needed to authenticate the user and to employ proper charging mechanisms. Next, we describe published approaches for secondary distributed systems. Techniques for Secondary Distributed Systems The FCC’s Interference Temperature Concept In this approach, the availability of spectrum is increased by sharing existing allocated (and in-use) portions of the RF spectrum such that it minimizes the probability of interference with existing legacy users [20]. The maximum distance at which data can be received error-free is dictated by the acceptable S/N ratio at a particular location. The FCC’s proposal says that a device can make secondary use of the spectrum by transmitting in a shared spectrum, provided the noise or interference caused by it is below an acceptable threshold of S/N ratio; here signal power is attributed to the primary legacy users, and everything else (including secondary

382

Technologies for the Wireless Future – Volume 3

Prevent aggregation above interference temperature limit

Assumptions: Single narrowband frequency, one-dimensional space pictured

New opportunities for spectrum access

Licensed signal

Noise Floor

Current part 15 limits Distance from licensed transmitted antenna

Figure 9.8 The FCC interference temperature concept (source: Spectrum Policy Task Force)

users) is characterized by noise. Ultra wideband (UWB) is one primary example of secondary use of the spectrum. In another implementation, sensing function in a wireless device scans for the white spaces in the spectrum, and a waveform generator generates a waveform to fill one or more white spaces such that the interference to the legacy users is constrained as per the definition of the FCC’s interference temperature concept. Such a technique is also referred to as temperature-adaptive waveform generation, and uses a distributed underlay interference temperature method with use of white spaces. Spectrum Gap Prediction In another concept, similar to the one in the previous section, the sensing function in a device continuously senses holes in the spectrum and a companion function predicts the availability of the holes in the future [21]. Using the hole prediction data, one or more wireless communication channels are synthesized from the one or more predicted holes. When a request for bandwidth is made by an application, the predicted holes are used to assign those segments of the spectrum. The implementation is based on synthesizing communication channels. Other Approaches In a recent paper, Buddhikot et al. provide an excellent overview of new directions in dynamic spectrum access using coordinated spectrum access techniques [22], which offer coordinated spatially-aggregated spectrum access via a regional spectrum broker. Such an approach is more realistic for the immediate future than many other approaches, which have focused on open uncoordinated access to the spectrum. Three different types of coordinated access method have been introduced: 1) coordinated access band (CAB), (2) statistically multiplexed access (SMA), and (3) a dynamic intelligent management spectrum architecture that implements SMA in the coordinated access band.

Methods for Spectrum Sharing

383

Solutions have been proposed to many other problems and issues associated with dynamic spectrum allocation, e.g. interference measurement. One such method uses a proximity measure to compute interference [23]. In this method, when a mobile node is operating in the secondary mode in proximity to a primary communication system within the same frequency band, a proximity detector is collocated either at the primary site or in the mobile node that wanders within the interference zone of the primary system. When a proximity detector is located at the primary site it sends a warning signal at regular intervals. This signal is then received by the mobile terminals, its strength compared with the transmitted signal, and a determination is made whether the terminal might be creating interference for the primary system. If such is the case then interference may be eliminated, or in other words the interference zone may be altered, by ceasing the operation of the mobile node, reducing the power of the mobile node, or by other means. In another implementation, a proximity detector may be located with the mobile unit operating in the secondary mode and a warning signal receiver built into the fixed site of the primary system. When this receiver computes interference signal power to be above the threshold, it instructs the mobile node to modify its behaviour till such time as the interference is contained to an acceptable level.

9.3 Problems and Issues in Flexible Spectrum Use 9.3.1 Higher- (and Cross-) layer Issues To attain the envisioned gains in practice, the WINNER flexible spectrum use and spectrum sharing functionalities need to be efficient and reliable. This depends critically on RRM that controls the spectrum use. Although flexible spectrum use has drawn wide attention in the research community, only a few pragmatic solutions for the required RRM functionalities have been presented. In the proposed WINNER RRM architecture, the complex problem of flexible spectrum use and spectrum sharing is tackled by dividing it into smaller, tractable functions: inter-system sharing, long-term spectrum assignment between WINNER RANs, short-term assignment, and constraint processing. Functions also operate on clearly separated timescales, thus partially preventing unpredictable interactions between the functions. A radio resource management architecture considered for the WINNER system concept is presented briefly in [24,25]. The necessary function introduced to the RRM for controlling FSU and spectrum sharing is referred to as spectrum control, and it consists of: 1. Inter-system Spectrum Sharing – controlling sharing with systems using different RATs. This function is located in the logical network node access control server (ACS) [24] as presented in Figure 9.9. 2. Intra-system Spectrum Assignment – controlling spectrum sharing between multiple WINNER RANs. This function is also located in the ACS. 3. Constraint Processor – implementing the time-varying transmission restrictions imposed by the spectrum sharing, located on the base stations at the MAC control plane. The functions are run simultaneously in all WINNER networks, and the only central entity between networks is a central database (or spectrum manager), maintained by regulator or some other authorized party. The basic parameters in the database can be altered, say, monthly, thus allowing for higher-level control on spectrum assignments through spectrum priorities and fairness/cost metrics. The architecture is depicted in Figure 9.9.

384

Technologies for the Wireless Future – Volume 3

Spectrum Control

External IP Network

Central Database

• Inter-system sharing • Spectrum assignment

Negotiations ACS ACS WINNER RAN

BS

BS

WINNER RAN

Negotiations over the air • Constraint processing • Resource partitioning • Feedback for spectrum controls

Spectral measurements UT

Figure 9.9 Illustration of spectrum sharing RRM architecture

The WINNER radio interface superframe is divided into time-frequency chunks, providing the basic unit for subdividing the spectral resources. Spectral resources assigned to the WINNER networks are also divided into two categories, namely: resources assigned to a certain network with indisputable priority, to guarantee basic operation of the network; and common pool resources available to all networks. The division between the categories can be adjusted through the central database. The signaling between WINNER networks and the central database is carried over the IP-based core network. However, the WINNER mode based on time-division duplex (TDD), relying on synchronization over all networks, also allows for over-the-air signaling between networks. For this purpose, the random access channel (RAC) on the superframe preamble is used. 9.3.1.1 Spectrum Assignment (WINNER) Spectrum assignment adjusts the spectrum resources available to the network according to the predicted aggregate load on the cells. The update rate of the assignments is one of the crucial design parameters. Fast update rate allows for more accurate load predictions, as well as for faster reactions to the sudden changes on the load. On the other hand, increased update rate induces more inter-network signaling, computational complexity and complex, unpredictable interactions with other RRM functions. To solve this problem, splitting into long-term and local, short-term spectrum assignments is proposed. Long-term assignment provide very slowly varying assignments for large geographical areas, hence introducing fair, stable and flexible spectrum assignments between networks with acceptable inter-network signaling and computational complexity. To complement long-term assignment, short-term assignment supports faster and local spectral flexibility by introducing short-term, cell-wise variations to the large-scale solution. However, the timescale of the short-term assignment needs to be clearly longer than that of other RRM functions.

Methods for Spectrum Sharing

385

Long-term Spectrum Assignment (WINNER) The function coordinates and negotiates the spectrum assignments between multiple networks for large geographical areas with spatial granularity of several cells. Spectrum assignments are updated periodically and at slow rate, i.e. in a time frame of tens of minutes. The signaling between networks is carried out through the IP-based core network, although over-the-air signaling can also be employed. The function is composed of the following five subfunctions, carried out sequentially: resource request calculation, defining the requested spectral resources for the next assignment period and announcing voluntary resource releases and retrievals of prioritized resources to the other networks; resource negotiations between WINNER RANs; resource rearrangement calculation, aiming to improve the tentative spectral resource assignments; rearrangement negotiation between WINNER RANs; and resource update, which updates the spectral resource and transmission constraint information for the constraint processor, and sends an information update to the central database. Short-term Spectrum Assignment (WINNER) The function controls short-term and local variations to the large-scale spectrum assignments. The assignments may be performed in the timescale of several seconds. Hence it enables faster adaptation to the load variations and geographically more accurate spectrum assignments, thus complementing the long-term spectrum assignment and load sharing. The function resorts to simplified signaling and computations to keep the overall complexity acceptable. The function simply requests resources from other WINNER RANs after being triggered by the long-term spectrum assignment or load sharing. In the case of a resource request from other RANs, the functionality rejects, accepts or partially accepts the request based on load and MAC control feedback on the overlaying and neighbouring cells. Inter-network signaling is performed over the air or through the IP-based core network. 9.3.1.2 Inter-system Sharing (WINNER) The inter-system sharing function provides an option to use shared spectrum as capacity enhancement for the system, and defines a coordination point for the spectral measurements on the terminals, which allows for harvesting on the hidden sensor network character of the system. Different scenarios are possible for spectrum sharing between systems using different RATs, depending on the regulatory rules governing the shared bands. In WINNER, the focus is on so-called vertical sharing, where the WINNER system is in the secondary, and the other system in the primary position. In such a scenario, the WINNER system has to control its emissions to avoid interference with the primary system. The sharing scenario considered requires the introduction of a novel RRM function called inter-system sharing, likely located on the ACS logical node. The function itself will depend strongly on the characteristics of the involved legacy system. The challenge is to make best use of the white spaces in frequency, space and time, and to transmit in these white spaces only, without generating interference towards the primary system. To achieve this goal, the white spaces are identified and estimated, after which the transmission parameters are adapted accordingly, resulting in two fundamental stages: 1. Identification and estimation of white spaces and exclusion zones, presenting the core of the inter-system sharing function. The function i) coordinates measurements; ii) gathers

386

Technologies for the Wireless Future – Volume 3

and processes information, obtained through measurements and signaling, on radio environment; iii) defines transmission constraints for shared spectrum. 2. Adaptation of transmissions according to the constraints, implemented by the constraint processor and resource partitioning, located on the MAC layer. While the adaptation step is a complex but still well-defined optimization problem, the identification step is more difficult to handle and can consist of several components: 1. Spectral measurements in user terminals, measurement coordination and processing at the inter-system sharing function. 2. Regular downloads from the central database (spectrum manager), maintained by a regulator or another authorized entity, requiring that the spectrum use of the primary system is quasi-static. 3. Real-time beacon signal maintained by the primary system. The identification and adaptation steps can be seen to form part of the ‘cognitive cycle’ described earlier. In general, reliable detection of white spaces is a hard problem and is deemed to be infeasible at a reasonable complexity without knowledge of the properties of the primary system. However, solutions can be envisaged for simpler but more practical cases, e.g. when the primary system is a fixed system with low geographical density. Use of the network nature of WINNER systems in measurements can also provide much more accurate information on sharing opportunities than would be possible in single devices or short-range networks. However, it has to be noted that operation in shared spectrum is subject to a number of constraints which cannot be influenced. In order to meet the B3G system requirements, especially QoS guarantees, operation in inter-system shared spectrum is considered in WINNER only for capacity enhancements.

9.4 Conclusion In this chapter we provided a structured overview of the potential dynamic spectrum use concepts with some specific existing proposals found from literature. The work in this area is in an embryonic stage and a lot of work still remains to be done to prove and compare the various concepts. Only time will tell whether or not spectrum sharing is a viable technology, and if so, which techniques would actually work in the product environment.

9.5 Acknowledgements The following individuals contributed to the white paper on which this chapter is based: Juha O. Juntunen (Nokia), Kimmo Kalliola (Nokia), Jean-Philippe Kermoal (Nokia), Juha Pihlaja (Nokia) and Yi Wang (Huawei).

References [1] F. Berggren et al., “Dynamic spectrum access”, Wireless@KTH, 23 Sep. 2004. [2] Federal Communications Commission (http://www.fcc.gov/sptf/), “Promoting efficient use of spectrum through elimination of barriers to the development of secondary markets”, report WT document No. 00–230, Oct. 2002. [3] “Measurements from the Shared Spectrum Company”, http://www.sharedspectrum.com/.

Methods for Spectrum Sharing

387

[4] Y.Y. Wang and Y. Wang, “On the application of spectrum sharing”, WWRF 15 Meeting, Shanghai, Apr. 2006. [5] DARPA XG Program, http://www.darpa.mil/ato/programs/XG/index.htm. [6] S. Dixit, K. Kalliola, J.O. Juntunen and J. Pihlaja, “Review of methods for flexible spectrum use”, WWRF 14 Meeting, San Diego, CA. [7] IST-1999-12515 DRiVE (http://www.ist-drive.org/), “Dynamic radio for IP services in vehicular environments”. [8] IST-2003-507581 WINNER (http://www.ist-winner.org/), “Wireless world initiative new radio”. [9] IST-2001-35125 OverDRiVE (http://www.ist-overdrive.org/), “Spectrum efficient uni- and multicast services over dynamic multi-radio networks in vehicular environments”. [10] IST-2003-507995 End-to-End Reconfigurability (E2 R), http://e2r.motlabs.com/. [11] IST-2003-507581 WINNER, Deliverable D6. 1, “WINNER spectrum aspects: methods for efficient sharing, flexible spectrum use and coexistence”, Oct. 2004. [12] IST-2003-507581 WINNER, Deliverable D6. 3, “WINNER spectrum aspects: assessment report”, Dec. 2005. [13] N. Devroye, P. Mitran and V. Tarokh, “Achievable rates in cognitive radio channels”, 2005 Conference on Information Sciences and Systems (CISS), The Johns Hopkins University, 16–18 Mar. 2005. [14] ITU-R WP 8F, working document towards “Preliminary draft new report on radio aspects for IMT-2000 and systems beyond IMT-2000”, agreed at 14th meeting. [15] Federal Communications Commission (http://www.fcc.gov/sptf/), “Notice of proposed rule making, in the matter of unlicensed operation in the TV broadcast bands”, FCC 04–113, released 25 May 2004. [16] IEEE 802. 18 Regulatory TAG ( http://grouper.ieee.org/groups/802/18/), “Comments to TV band NPRM”, http://grouper.ieee.org/groups/802/18/Meeting_documents/2004_Nov/. [17] M.C. Chuah and K.R. Medapalli, “Method and apparatus for dynamic frequency selection in a wireless communications networks”, US patent application No. 2004/0266351A1, 30 Dec. 2004. [18] E. Callaway, “Wireless Sensor Networks: Architectures and Protocols”, CRC Press LLC, Boca Raton, FL, 2004. [19] S. Souissi and E. Callaway, “Method and apparatus for dynamic spectrum allocation”, US patent No. 6327300B1, 4 Dec. 2001. [20] FCC NPRM on “Interference Temperature Metric”, ET Docket No. 03–237. [21] R. Berezdivin, R. Breinig and A. Topp, “Next generation wireless communications concepts and technologies”, IEEE Communications Magazine, pp. 108–116, Mar. 2002. [22] M. Buddhikot, P. Kolodzy, S. Miler and J. Evans, “New directions in wireless networking using coordinated dynamic spectrum access”, WoWMoM, Catania, Italy, 2005. [23] D. Otten, “System for mobile communications in coexistence with communication systems having priority”, US patent No. 6449461B1, 10 Sep. 2002. [24] V. Sdralia and N. Johansson, “Towards radio resource management architecture for a B3G system”, WWRF14, San Diego, USA, 7–8 Jul. 2005 [25] IST-2003-507581 WINNER, Deliverable D7. 6, “WINNER system concept description”, Oct. 2005. [26] IEEE 802. 22 WG on WRANs (wireless regional area networks), http://grouper.ieee.org/groups/802/22/.

10 Ultra Broadband Home Area Network Edited by Djamal-Eddine Meddour (Orange Labs, France Telecom Group)

10.1 Introduction Recently, the home arena has captured the interest of a wide range of actors, from cable TV providers and electronic consumer manufacturers to the network operators. Digital home experience is every day becoming more and more facilitated by the growing speed of access links to the Internet. Notice that new high-data rate services such as TV programmes and video on demand over ADSL [1] are already a commercial success. Actually, with the evolution of networking technology, new ‘networking-ready’ mass storage devices are already in the market (Media Center, Media Render, set-top box, etc.) and promise to make the digital experience even more exciting, with for instance the upcoming HDTV. Indeed, the new generation of home devices is basically manageable and able to communicate through network interfaces. Moreover, end user devices are fitted with high-speed interfaces to easily transfer all types of multimedia support. On the other hand, the constant growth of end user demands for high-data rate connectivity (beyond the Gbps), either using wireless or wired support, and the popularity of real-time services (gaming, VOD) highlight the high stakes in this business, in particular for the network operator. Hence, providing a simple and efficient solution that facilitates the set up of convergent home network architecture is essential. These trends will certainly make the UBB-HAN (ultra broadband home arena network) a convergence arena in the future, where these devices and services will have to interwork at home and in continuity with the operator’s network services. In this study, we provide the reader with a deeper analysis of UBB-HAN systems and of the elements that we have to consider in order to meet the end user requirements in terms of seamless connectivity.

Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

390

Technologies for the Wireless Future – Volume 3

In the next section, we will give some examples of these devices, as well as their impacts. Then the challenges of this convergence platform will be tentatively described.

10.2 Applications Challenges Digital mass storage devices gain more success at the home every day. These devices, whose standardization is ongoing within, for example, DLNA [2] and HGI [3] fora, offer not only demodulation of digital broadcast programmes and access to remote services for operators’ networks, but high connectivity to end devices such as TVs, home cinemas and PCs. To enable the use of these devices, the trend is that it shall be possible to use them with high-data rate connectivity to transfer content either from remote servers or between end devices sparsely distributed everywhere at home. On top of that, end users are not always geeks, and most of the time will expect highly simplified use, installation and maintenance of these devices by means of user-friendly interfaces. The home network is used to connect multimedia terminals and appliances as well as ‘Internet things’ within and around the home. The main goal is to offer new user experiences in daily life, thanks to machine-to-machine and ambient intelligence, which free time and attention by delegating capture and information processing to machines, sensors or things. Users will expect that portable terminals and appliances will work equally in other environments away from home, such as cars, caravans, hotels and cafés. Home networks will be used, for instance, to view a TV programme recorded on a PVR in one room of the house on a screen in another room; to share the same music in different rooms; to transfer video from a camera in the garden to an editing station in the house. The network could also be used for home automation, for white goods/equipment connectivity, as well as for voice and video calls. It must be possible to install a home network with minimal alteration to existing homes. These requirements, along with the need to work outside the house, practically require the use of wireless or powerline networks (no new wire concept, including low-data rate and broadband technologies). It is obvious that each home will have its own network topology according to its architecture and it is clear that heterogeneous network configuration will be the most common situation.

10.2.1 UBB-HAN as a Convergence Platform Figure 10.1 depicts typical configuration of UBB-HAN; within this context are a number of new issues which need further consideration. Let us highlight some of them in the following points: 1. When moving from one room to another, an end device can be seen as either nomadic or mobile, which raises a set of issues for cellular network coverage everywhere at home. 2. Intrusive access or eavesdropping is therefore also to be considered. 3. As stated before, end devices such as HDTV, even hanging on walls, shall be connected by some means, raising the question of ‘UBB static coverage’ as well.

Ultra Broadband Home Area Network

Security ?

391

Qos ? Nomadism or Mobility ?

End devices Compatibility ?

Full VHDR Coverage?

wifi

ADSL RLL FTTH Analog Phone

Figure 10.1 HAN as a convergence platform

4. End devices from different manufacturers also have to interwork either directly or via the UBB-HAN. 5. When several services run simultaneously, some of them may tolerate transfer disruption or delay whereas others may not. Quality of service (QoS) is another dimension for consideration, to guarantee quality of simultaneous plays. The home gateway already gives some answers to these topics as a convergent box for access to the network and connectivity to end devices. It is now faced with increasing data rates as well as home coverage expectations. 10.2.2 Key Requirements Facing the above-mentioned challenges, the telecom operator has to define top-level requirements to make this convergence a success. First of all, high data rate needs definition. Thanks to a dynamic bandwidth assignment function, about 400 Mbps access speed to the telecom network can be envisaged, relayed by more than 1 Gbps at home. Ultra-small latency time, less than 1 ms round trip time, can be offered thanks to simple framing and coding of short packets. Further, heterogeneous QoS of simultaneous services has to be more classically classified in ‘real time’, ‘non-real time’ and ‘best effort’ categories. Major connectivity standards should ease the job, but the problem is, of course, to have them widely accepted throughout the industry. These standards should natively provide ‘beyond Gbps’ connectivity and be accepted worldwide by harmonized regulation. Among various other requirements, some attention has also to be paid to express coverage issues, not only in ‘geographic’ terms but also in terms of (semi-) static or dynamic scenarios.

392

Technologies for the Wireless Future – Volume 3

10.3 Connectivity Connectivity refers to the type of link used and its capacity. In UBB-HAN, several types of connectivity can co-exist, with different capabilities and characteristics, varying from wireless (802.11, 802.15, etc.) to wired (FSO, DSL, etc.). A brief description of the main connectivity technologies is presented in the following sections. 10.3.1 Wireless Several standards are working to provide solutions able to fulfil the services and application requirements. 10.3.1.1 IEEE 802.11n 802.11n [4] is considered the new generation of WLAN support; this is about to be standardized in 2008, with the following features: ž improvement of coverage ž very high data rate ž introduction of MIMO (multiple input multiple output) technique (up to 400 Mbps) Wireless networks using 802.11 standards are widely used to connect computers and other IT equipment. However, the 802.11 standards have been optimized for transfer of data files, an application that requires perfect accuracy but imposes no particular constraints on transfer time. They use repeat transmission of data that are not received correctly. Streaming audiovisual signals have quite different characteristics from file transfer: forward error correction can make them reasonably tolerant of transmission errors, but they must be transferred in real time. Tests have shown that transfer of audiovisual signals over existing wireless networks is extremely unreliable and inefficient. Notice that this concern captures the current interest of the IEEE 802.11 WG. This group, based on 802.11n principles, launched a new study group to meet real-time application requirements [5]. Further, it is worth mentioning that current wireless networks work in the license-free bands, which are now becoming very crowded. Issues of reliable transmission and interference could be mitigated by allocating a small amount of spectrum for home networks. This would require study of the absorption properties of building materials used in different countries. 10.3.1.2 UWB Ultra wideband (UWB) is a wireless technology intended to enable high-data rate application for personal area networks [6]. Its main characteristics can be summarized in these points: ž The frequency used is 3.1–10.6 GHz, possibly at 60 GHz. ž Power densities allowed in Europe are not yet settled. ž Coverage will be limited to 1–4 m at 480 Mbps. It is clear that a single AP located at the HGW will no longer be enough to have global coverage at home. But UWB is a good opportunity to overcome the problem of spectrum inefficiency and interference.

Ultra Broadband Home Area Network

393

10.3.1.3 FSO (Free Space Optics) FSO technology is a very useful alternative to the fixed line, particularly in overcoming the difficulties encountered when installing a physical wired connection between two communicating points or in situations where setting up a fibre is not economically viable. Two communication technologies are possible in this area, and their characteristics are described below: IrDAR (Infrared Data Association) Technology ž can reach 1 Gbps ž coverage limited to a single room ž feeding beacons with fibre optics. VLC (Visible Light Communications) Technology ž reuses the lighting infrastructure (LEDs) ž countermeasures are necessary to cope with the ambient noise. As the potential bandwidth is huge and few systems already operate in the wireless optics bands, free space optics offers a good way to solve the problem of spectrum shortage, or interference limitations. It is now under study in Japan. 10.3.2 Wired 10.3.2.1 Powerline Communications Technology Powerline technologies already exist, using many standards. HomePlug AV1 seems to be the most suitable solution at the moment. It is used to transport audio and video streaming services and will support all types of multimedia stream (TV, file downloads, VOD, VoIP, music transfer, etc.) with an acceptable quality of service from the user’s point of view. But the physical layer can still be improved in order to reach even higher data rates. So the study of next-generation HomePlug AV is inherently a topic for the next generation of home network. The main characteristics of the HomePlug AV are the following: ž 200 Mbps PHY, which is a first step towards future Gbps PHY. ž OFDM modulation and usage of turbo codes. ž Both TDMA and CSMA MAC layer are supported. Some improvements are envisaged for the After HomePlug AV, in particular: ž Using CDMA together with OFDM can increase bit rate by 25 %. ž MIMO techniques possibly offer a better robustness. Current powerline systems, both in the access network and for in-home networks, cause serious interference to wireless services in the MF and HF bands (radio broadcasts, air and maritime navigation services, RFID, etc.) which are guaranteed protection by the ITU Radio Regulations and by the EU’s EMC Directive. Further research is needed to ensure the co-existence of powerline systems and licensed uses of the radio spectrum. 1 HomePlug AV is concerned with PLC for high bandwidth communication with entertainment devices. It is sponsored

by the HomePlug Alliance, http://www.homeplug.org/home/.

394

Technologies for the Wireless Future – Volume 3

10.3.2.2 Optical Fibre Fibre is a very good candidate for the next generation of cabling. Technology has improved a lot, and one can now see very small-diameter fibre with very small curvature. As optical fibre is now reaching houses with FTTH with an ultra broadband capacity, optical fibre cabling seems to be a good solution for providing the house with ultra broadband capacity. One can find different types of fibre with different capabilities and features: Plastic Fibres: CYTOP Multimode Fibres ž ž ž ž

less than 25 dB/km loss (850–1300 nm) 1.2 Gbps/km core diameter 120 µm easy to connect.

Micro-structured Fibres: Hole-assisted Monomode Fibres ž no loss, even with 5 mm bending radius ž preconnected curl cord. 10.3.3 Hybrid Another topic is the possible combination of the core connectivity technologies (wire and no-wire technologies) to enable best global coverage. A mixture or combination of technologies could compensate for the weaknesses and strengths of respective core connectivity technologies. For example, the small-cell radius WPAN connected to a wire backbone in the home could provide best coverage, allowing wireless access when needed at high data rate. It would be a good way to solve the problem of frequency management. 10.3.3.1 Mesh Networks 802.11 and UWB could be utilized in an explore mesh network topology, or as an extension of a UBB backbone network (such as powerline and/or optical fibre). UWB access points could be implemented only where wireless was needed. Therefore, hybrid solutions combining wire and wireless technology are a good way to achieve global coverage of the home. 10.3.3.2 Radio-over-fibre RoF (Radio-over-fibre) was studied a couple of years ago, but components and applications were not yet ready for massive deployment. Now, with very high access network capacity and UBB-HAN, ROF is a very promising solution. The concept is to distribute radio signal on the optical signal directly to simplified access points. 10.3.3.3 802.21 IEEE 802.21 is only mentioned here as an IEEE working group [7] that needs to be considered in the scope of this home connectivity technologies convergence.

Ultra Broadband Home Area Network

395

In the context of standardization, the IEEE 802.21 working group is currently working on specifications of media-independent mechanisms that optimize handovers between heterogeneous media. The target cases are handovers between different IEEE 802 family technologies (.3, .11, .16) as well as between IEEE 802 media towards cellular access networks (3GPP, 3GPP2). These MIH mechanisms include: ž Event notifications (ES) from lower layers to upper layers. ž Command services (CS) from upper layers to lower layers, triggering procedures to be executed by access network layers. ž Information services (IS) to exchange relevant generic information for handover decision between protocol layers and between remote entities.

10.4 Access Challenges 10.4.1 The Concept of Access Network Continuity This section lies within the context of the deployment of access technologies beyond ADSL, such as Fibre to the Home, currently under investigation by France Telecom [8] and Home Gateway, branded under the name LiveBox. Several hundreds of Mbps are reasonably reachable in the near future, backed up by emerging standards such as GPON 2.4 G [9]. The deployments achieved until now in the home area network have led to configurations where the different flows between the end devices and the service nodes are normally segmented in separate tunnels. Each end device is directly connected to the related service node, or the service provided by the service node is terminated in the residential gateway and then locally handled by the gateway inside the home network; one given end device can only connect to one platform of service, and no interaction between the services is generally possible. More recent deployments have overcome this segmentation in specific cases, for instance allowing a mobile phone to connect itself either through the mobile network or through the broadband access network, according to the situation. With its numerous flows circulating from end device to end device and interacting between them, the ultra broadband home network therefore needs to be revisited from the point of view of continuity of access, especially the role of network termination; until now the home gateway has been the demarcation point between the private home domain and the operator’s network. This section proposes first a reference architecture model adopting an end-to-end point of view, and then elaborates the matter of the interaction between the UBB home network and the access network in the prospect of expected evolutions in the access. 10.4.2 Reference Architecture Model for the Interaction Between the UBB-HAN and the Access Network Figure 10.2 illustrates the proposed reference architecture representing the delivery of a set of multimedia services to a UBB residential user’s home, over a fixed infrastructure access network based on wireline (xDSL, fibre, etc.) or radio (WiFi, WiMAX, etc.) technologies. The representation of the UBB home is in accordance with the connectivity description in the former section; the home gateway (HGW) remains the interface point between the UBB-HAN and the access network but is no longer necessarily the central point of the user’s

396

Technologies for the Wireless Future – Volume 3

AAA + Policy Management TVNUm. VOD

Service Terminals

Gigabit Mesh Home Network

Service Nodes

Interconnection Device

Access Network (access + metro)

Interconnection Device

VolP Gateway

Aggregation VolP

Interconnection Device

Home

Access Node

Video Server

Node

Gateway ISP

Internet Home Network Segment

Access Link Segment

Access-metro Network Segment

Access Network Segment

SN Link Segment

Service Node Connectivity

Figure 10.2 Reference architecture of a multi-services access network

premises. Regarding topology, the UBB-HAN is no longer limited to a set of point-to-point links between terminals and the HGW, but may instead be based on a more or less complex mesh network between terminals, HGW and possible additional interconnection devices (e.g. an Ethernet switch), like in a classical LAN. On the access network side, the HGW implements the network interface to the access node (typically referred to as network termination (NT)). The functions supported depend directly on the access networking technologies (e.g. layers 2/3 technologies) and on the access link technology (e.g. DSL, optics, radio, etc.). The functions required are those necessary to set up, maintain and tear down connectivity with the SNs in the control, transfer and management planes. Depending on the access link technology, some specific additional functions may be required (such as in the case of a shared access link, like PON or WiMAX). The HGW could implement different models for interworking (IW) between the HN and the AMN depending on the level at which this IW is performed. Two basic models are possible: ž Termination on each side of the networking technology and the processing of native service payload for IW; simple re-encapsulation of service payload (almost transparent for service level) or processing of some signalization or media flows. ž The use of a common networking technology for IW (typically L2 or L3) with transparency regarding service level. As indicated earlier, it is assumed that the HGW includes a network termination function, which means that the HGW network identifier (network addressing) could be used to identify and authenticate the source of a traffic flow inside the network (and if necessary the associated user), particularly to implement a network access control function. This is possible if some specific points regarding HGW identification could be guaranteed, such as the uniqueness and trusted character of the identifier. Normally this identification is not applicable at service level, for which it is still the terminal that must be identified. This is however what is done in the case where the HGW is identified by the SN and it receives a public IP address for traffic forwarding and end point identification

Ultra Broadband Home Area Network

397

at service level (e.g. with the use of an NAT function in the HGW to share this address between several terminals). 10.4.3 The Evolutions Expected in the Access Network The access network reference architecture presented in Figure 10.2 could be seen as the architecture used for initial deployments of broadband access networks by most ILECs (incumbent local exchange carriers), as well as by most CLECs (competitive local exchange carriers) [10], [11]. Since these first deployments, a certain number of evolutions have appeared, such as: ž The generalization of IP-based multimedia services with an increasing use of video. ž The deployment of cellular mobile networks providing voice and multimedia services. ž The deployment of high-speed fixed radio access networks based on WIFI/WIMAX technologies. ž The technical and economical maturity of optical technologies, allowing cost-effective deployment of FTTx access networks, based on point-to-point or point-to-multipoint (PON) topologies. To satisfy the requirements imposed by these new applications, some evolutions of the initial access network architecture are necessary or at least desirable. These evolutions can be grouped into two broad classes: short-term evolutions and long-term evolutions. The first class typically aggregates all requirements which do not imply significant modifications of the initial architecture. These are mainly related to the increase of access transport capacity, the enlargement of service offers with associated requirements (e.g. enhanced models for QoS, security, etc.), the migration toward full IP-based services, etc. The deployment of FTTx architectures is a particular aspect of these evolutions as, if it does not allow new applications by itself, it does require the deployment of a new infrastructure which, associated with the higher speed and reach of optical technologies, offers the possibility of a reorganization of access networks. The second class of evolution mainly covers a key element regarding access networks: the convergence between fixed and mobile networks, referred to as FMC (fixed-mobile convergence). This convergence is a convergence at network level, not at commercial or service levels, which means the integration of different access networks in a single global architecture, capable of supporting in an optimized way all types of service, providing to customers a permanent and (almost) transparent access to their applications. The resulting network is often referred to as NGN (next-generation network). 10.4.4 Migration Scenarios Three different migration scenarios have been identified in the broadband access network which may impact the related ultra broadband home network: 1. A first approach is to get in line with a connection-oriented scenario based on an evolution of the existing access architecture presented in Figure 10.2, providing enhanced connectivity between users and SNs (service nodes), but still in a strictly controlled mode. In that scenario, all the packets of a given traffic flow are transported on the same connection, incidentally bringing the advantages related to the separation of traffic flows (support of signaling and CAC, facilities for traffic engineering and security).

398

Technologies for the Wireless Future – Volume 3

2. A second approach could be the continuity between home and access in a connectionless-oriented mode, based on packet processing like in a classical IP network, which allows the more open connectivity model in which all users can connect to all service nodes, as well as to other users. 3. A third option is to consider an evolution towards an architecture based on the integration of the fixed-access network in a next-generation network such as those studied by various standardization bodies (ETSI TISPAN, 3GPP, ITU-NGN, etc.), allowing the convergence between multiple broadband and QoS-oriented network technologies (mobile, fixed, etc.). These approaches are based on mature protocols at Ethernet or IP level for the connectivity establishment, the handling of multicast flows, the signaling operation, the management of QoS and other functionalities allowing the support of multiservice in the access. Therefore it appears quite permissible to take maximum advantage of the potential of these protocols in order to reach an end-to-end seamless continuity with the access. The assessment of their extension inside the gigabit home network will bring some significant progress to that concern, while keeping in mind the idea that the context of the gigabit home network will very likely require the investigation of other new specific mechanisms. Moreover, this approach should guarantee the coexistence of the home network with the different existing access systems.

10.5 Architecture As stated already, the upcome of new generations of devices with ‘network-ready’ capabilities at home, and the ever-growing demand for high data rate inside the home to fulfil the requirements of real-time services (gaming, VOD), highlight the need to design simple, robust and efficient architecture for UBB-HAN that facilitates the setup of convergent Gbps home networks. To meet the requirements mentioned in Sections 10.2 and 10.3, in addition to the home gateway, which provides the Internet connectivity to the home, novel physical network elements playing a central role in providing connectivity will be introduced within the home network. They must also embed several functions, including intelligent security, routing and scheduling capabilities. In the rest of this document we will call such an element a ‘bridge’. The interconnection of bridges and the HGW will form the backbone of the UBB-HAN. In the following we will discuss several architectural issues and considerations related to the design of the backbone of the UBB-HAN. A description of the entities that form the UBB-HAN will be provided as well. 10.5.1 QoS The home gateway must provide a set of mechanisms to enable end-to-end QoS, i.e. meet the traffic requirements coming from the network operator domain to the UBB-HAN and vice versa. Indeed, the HGW must support QoS policies both in the operator network and on the home network side when many simultaneous broadband services are used. The HGW must grant the needed QoS to in-home flows as well. A number of congestion points can occur, for instance when a significant difference between the physical rates of different technologies (PLC, wireless, etc.) is experienced, or when LAN-LAN traffic is aggregated with the downstream access network traffic and they compete for access to one of the LAN-side ports. Actually, the LAN-LAN traffic can burst at very high rates and so fill up the HGW buffers, thereby delaying notably lower-rate downstream traffic.

Ultra Broadband Home Area Network

399

Providing QoS in UBB-HAN is essential to meet the needs of real-time applications. This is a research area that captures a lot of interest. Actually, besides QoS routing proper to path computation, other solutions have been proposed, aiming at resource reservation and data prioritization. Therefore, different approaches can be considered. In the following we will detail the different schemes. 10.5.1.1 Home Gateway Initiative (HGI) Approach The HGI QoS approach has so far been concerned with managing QoS through the HGW itself. For reasons of simplicity and scalability, the HGI advocate a traffic class-based solution (`a la diffserv). The classifiers are combinations of the ingress packet header fields. LAN-side ingress QoS markings are usually untrusted, but can be used if the trust relationship is established by some other trusted means. The service classification is used to assign the packet to the appropriate queue, to set the L2 markings and finally to drop packets on the basis of classification. As for the admission control, the HGW does not support signaled admission control, but the overload protection mechanism can provide some features of admission control by limiting the number of service instances to a predefined value. 10.5.1.2 Universal Plug and Play (UPnP) Approach Since any proposed QoS solution needs to meet the requirements for multiple usage scenarios, it is expected that vendors may use any UPnP device as a container for the services defined in the UPnP QoS framework. Therefore, the QoS management elements must look for UPnP QoS services embedded in all UPnP device types. The UPnP QoS framework [12] is composed of three entities: QoSDevice service, QoSPolicyHolder service and QoSManager service. The QoSDevice will transmit the data stream and the QoSPolicyHolder service defines the appropriate policies for the network on which it resides. That is, the QoSPolicyHolder can be used to set the priority of a particular traffic stream by indicating the suitable traffic policies. Basically two values are in use: the TrafficImportanceNumber is used to assign the priority to a given traffic stream according to the traffic type, and the UserImportanceNumber to enforce traffic admission policies. 10.5.1.3 Admission Control Approach Admission control requires a decision to be made about resource availability before a session is established, and therefore it involves signaling. The admission control mechanism has to prevent new sessions if they would adversely impact the existing ones. 10.5.2 Topology We recall that the architecture of the ultra broadband convergent UBB-HAN consists of the interconnection of the following elements: ž HGW – this is the operator equipment that connects the UBB-HAN to the access network and consequently to the Internet. The interconnection with the access network can be either wireless (for instance WiMAX/3G) or wired (XDSL/optical).

400

Technologies for the Wireless Future – Volume 3

Bedroom

Home Office

Bridge

Backbone

Lounge WLAN and/or WPAN Gateway

Figure 10.3 Real topology of a UBB-HAN

ž Bridges – as stated before, bridges are the key element in extending the HGW coverage with smooth QoS support to the whole of the UBB-HAN. Bridges are operator-controlled equipment that aim to interconnect the different end devices. Bridges must be equipped and enable several connectivity technologies which were discussed in Section 10.3, such as WiFi, UWB, Ethernet and PLC. User end devices can either be connected directly to the HGW and/or the bridges, or via intermediate connecting devices such as switches and routers, which are controlled by the client in multihop fashion. Figure 10.3 depicts, from network perspectives, a real topology of a UBB-HAN. 10.5.2.1 Backbone Topology The interconnection of the bridges and the HGW forms the home backbone. Depending on the corresponding connectivity, two types can be envisaged: Point-to-point Backbone In the point-to-point configuration, the candidate technologies are: point-to-point optical (e.g. POF), Ethernet cabling, ROF and FSO; in other words, unshared media. In this case, bridges may be connected to the HGW through multiple-hops connectivity. Point-to-multipoint Backbone Optical broadcast and select, PLC bus, 802.11 a/b/g/n coverage are the related technologies. This implies that the several bridges and the HGW share the same medium. Nevertheless, the multihop connectivity is enabled in this case as well.

Ultra Broadband Home Area Network

401

From the deployment perspective, the first type of backbone is associated with upgrade (from node to node) of an existing installation of the client, while the second type could be better suited for new installations. Obviously, QoS criteria should also be taken into account for determining the appropriate topology, since unshared media offer better performance than shared ones. A hybrid backbone configuration where the HGW and the bridges are connected through shared and unshared media simultaneously is also possible. 10.5.2.2 Topology Management As we can see, the key elements characterizing the architecture are the operator-controlled equipments: the HGW and the bridges. Indeed, they could enable different features such as resource reservation and scheduling, routing, support for heterogeneous connectivity technologies, service discovery, inter terminals communication, load balancing, etc. Whereas the HGW principally handles network management operations and the interconnection with the external world, the bridges provide continuous access over the home and deal with the end-device heterogeneity. At the same time, we have to consider that some end devices are able to communicate directly without the network assistance. These factors combined point out clearly the need for a robust routing protocol to manage efficiently the network heterogeneity and to ensure end-to-end multihop connectivity. Furthermore, the network should support the introduction of new elements and the retrieval of existing ones automatically since simplicity is a key factor from the user point of view. For these purposes, we distinguish basically two kinds of routing approach: centralized and distributed. More details will be given below. 10.5.2.3 Centralized Approach In the centralized approach, the network can be organized in either a tree or a star structure at the HGW, which will act as the main home manager, while the bridges act as relays. The role of the bridges is twofold; first, to extend the coverage to the whole home, and second, to update the HGW with information on link status. Therefore, the network topology management is performed in a centralized fashion by the HGW. For the routing function, it will be coordinated by the HGW. As illustrated in Figure 10.4, the traffic is handled by disseminating the routing policies from the HGW to the network elements which will enforce them. Indeed, the HGW makes a centralized decision of a path, and then establishes the path by informing all the bridges along it of relevant path information. Note that in some specific scenarios all the traffic may go through the HGW to perform specific processing and/or adaptation (authentication, flux adaptation, etc.). As a matter of fact, this approach should be simpler than the distributed one, but it could raise points of failure and bottleneck issues in particular where the routing is gateway-centric. 10.5.2.4 Distributed Approach In contrast with the centralized approach, in the distributed approach the topology management is performed in a distributed manner using mesh network principles. Indeed, the HGW

402

Technologies for the Wireless Future – Volume 3

Routing policies

File transfer from PC1 to PC2 through HGW (4 hops) Streaming video from PC1 to TV2 through bridges (3 hops)

Figure 10.4 Centralized approach

and the bridges are interconnected without any predefined constraint, therefore forming a meshed network. Basically, the route selection and establishment decisions are distributed among the bridges and the HGW. This implies that either proactive or reactive ad hoc routing protocols can be applied at each bridge for route computation and establishment. For the deployment matter, the distributed model seems to be more upgradeable and more flexible. However, the distributed character of the solution, resulting in more reliability and scalability by avoiding the gateway-centric routing effects, generates some potential issues linked in particular to the QoS insurance, such as resource reservation, scheduling and routing decisions, which requires more effective policies. Figure 10.5 illustrates how the routing is performed in the case of the fully distributed approach. 10.5.3 Integration with Mobile Network The integration of the home network with the mobile network lies in the line of fixed-mobile convergence (FMC). From an end user point of view, FMC can be viewed as a single terminal that can be used both when at home (as a fixed-line telephone) and upon leaving home (as a cellular mobile phone). From an operator point of view, FMC can be viewed as an opportunity to offer commercial convergence (one customer point of contact, one bill), service convergence (uses mobile or fixed network transparently) or network convergence (provides seamless mobility user experience when moving across heterogeneous access networks). Hereafter an outline of the already existing FMC technologies is briefly presented, and the ongoing development and research directions are introduced. Architectural challenges regarding device and network convergence include multimode handsets, fixed-mobile network interconnection, optimization for circuit-switched/packetswitched (CS/PS) traffic, dimensioning, security, (seamless) mobility, access selection and

Ultra Broadband Home Area Network

403

File transfer from PC1 to PC2 Video Streaming from PC 1 to TV2

Figure 10.5 Distributed approach

charging differentiation. Several types of FMC are proposed to provide different kinds of service; they rely on different network architectures, with different levels of integration (access, core network, IMS and service platforms), requiring distinct solutions to the previous challenges, thus having different impacts on the network. Different combinations between these FMC solutions can be devised based on targeted services. 10.5.3.1 UMA/GAN One of the first FMC-deployed solutions is based on the generic access networks (GAN) standard. The GAN architecture was introduced to enable operators (fixed and mobile) to take advantage of emerging markets in relation to wireless VoIP. GAN is the 3GPP standard version of the unlicensed mobile access (UMA) architecture. GAN requires use of a new, GAN-enabled mobile terminal, supporting dual stack of e.g. WiFi (or Bluetooth) and GSM/GPRS technologies. It also requires introduction of a new node in the operator’s access network, the GAN controller (GANC). The GAN architecture has been designed to target mainly voice telephony (data connections are also accounted for in the architecture, but the solution is not optimized for packet data services). The currently available services are GSM/GPRS ones, thus the bit rates are considerably limited with respect to target of Gbps@home. It is worth mentioning that the GAN evolution to 3G is ongoing. 10.5.3.2 I-WLAN WLAN has been identified as an alternative to extend the existing radio access technologies (RATs) such as GERAN and UTRAN. This resulted in I-WLAN Release 6, an architecture introduced by 3GPP, for interconnection of 3GPP systems with WLANs. WLAN/3G Interworking may therefore allow user equipment, equipped with a USIM card, to access 3G services, Intranet and Internet through a WLAN hot spot. I-WLAN architecture necessitates the introduction of new network entities in the mobile core network.

404

Technologies for the Wireless Future – Volume 3

Concerning the evolution of 3GPP-WLAN Interworking Release 7, an evolved UMTS architecture is under definition, where it is expected that IP-based services will be provided through various access technologies. A mechanism to support seamless mobility between heterogeneous access networks is needed for future network evolution. I-WLAN is included to ensure a smooth migration path from the current I-WLAN work to a generic multi-access solution. On IMS part, another work item called ‘Voice Service Continuity’ (VCC) was created. The overall objective of this work was to define handover between CS and IMS. I-WLAN is an IP-capable access network and, as such, can be used in conjunction with IMS/VCC to hand off SIP-based VoIP call over WiFi to CS domain over Cellular. 10.5.3.3 SIP The session initiation protocol (SIP) [13], defined by IETF, can potentially bring new application capability to FMC when used over the FMC architectures considered. SIP is used in the control plane only, and allows establishment, modification and deletion of sessions of multimedia conversational services between two users or a group of users (conferencing). Multimedia means that real-time exchanges between end users concern voice and video, voice only, instant messaging, or any other application that may arise. 10.5.3.4 IMS IMS (IP multimedia subsystem) [14] is a network domain architecture that provides a control layer for multimedia services on top of IP connectivity layers. IMS is an SIP-based architecture, with some adaptations regarding SIP IETF specifications to consider some mobile networks requirements for IMS-based services delivery (e.g. user network location). The IMS was initially specified by the 3GPP, from Release 5, to provide IP multimedia services to the end users over a UMTS packet-switched domain. In addition to SIP-based solutions, the IMS proposes a set of enhanced procedures such as control of service invocation (integration of user service profile, operator policy), network resource reservation and enhanced charging correlating network-originated and IMS-originated CDRs (charging data record). IMS has also been standardized by TISPAN for the evolution of fixed networks; however, the two standards (3GPP and TISPAN IMS) are not exactly the same. 10.5.3.5 Femto Cells An indoor solution to allow mobile customers to access their services when they are inside buildings, at home or at work, consists of putting a mobile network antenna (together with additional functionality) inside the building, integrated into the HGW, which becomes a ‘HomeBaseStation’ or ‘FemtoCell’. In the new perspective of mobile network extension, the mobile indoor radio technology is the same as that used outside (2G (GSM/GPRS/EDGE), UMTS, 3G LTE, WiMAX, etc.). The objective is to enable the use of monomode terminals (and not dual stack terminals, as is the case with the UMA/GAN architecture). Recently (March 2006), related study items have been opened in 3GPP, both for UMTS and LTE.

Ultra Broadband Home Area Network

405

10.5.3.6 Convergence at Application Level Apart from the convergence at the network level, convergence can be offered via the application layers. The user can initiate a session under any access network. No mobility management is handled and no seamless mobility when changing access is provided, but the session/context is kept in an application server located in the network, which enables retrieval of user context when reconnection from another access or terminal occurs.

10.6 Conclusion In the near future, in order to fulfil user needs and address the requirement of high data rate in the domestic sphere, proposing cutting-edge solutions and innovative architectures will be essential. Throughout this chapter we have presented the state of the art and our vision relative to this matter, including a set of challenges that face the setup of UBB-HAN. We have emphasised the need for hybrid connectivity in the home with respect to the technology evolution, and the importance of ensuring the continuity of the service with the access network has been highlighted as well. We would like to further develop these ideas, and the WWRF is definitely a candidate which offers a good framework to elaborate perspectives, such as: ž The specification of the bridge functionalities. ž The definition of suitable QoS metrics. ž The proposal of a robust and efficient routing protocol which takes into account QoS requirements. ž The evaluation of connectivities hybridization in the home.

10.7 Acknowledgements The following individuals contributed to this chapter: Eftychia Alexandri, Martial Bellec, Pierre Jaffr´e, Abdesselem Kortebi, Riadh Kortebi and Philippe Niger (Orange Labs, France Telecom Group). The authors would like to express their deep appreciation and gratitude to Dr Bin XIA (Huawei Technologies) and Ben Worthington (Vodafone Group) for their valuable comments.

References [1] [2] [3] [4] [5]

MaLigneTV, “Digital TV on ADSL”, http://www.malignetv.fr/. Digital Living Network Alliance, http://www.dlna.org/home. Home Gateway Initiative, http://www.homegatewayinitiative.org/. “Status of the 802.11n standard”, http://grouper.ieee.org/groups/802/11/Reports/tgn_update.htm. Video Transport Stream (VTS) SG, https://mentor.ieee.org/802.11/public/07/11-07-1955-00-0vts-par-and5-criteria.doc. [6] I. Siaud and R. Legouable, “Turbo-coded MC-CDMA techniques applied to WPAN UWB/WB systems at 60 GHz (IST MAGNET Project)”, WWRF#15-WG5, 2005. [7] IEEE 802.21 working group, http://grouper.ieee.org/groups/802/21. [8] Very High Speed pilot program, “Fiber To The Home”, http://www.francetelecom.com/en/financials/journalists /press_releases/CP_old/cp060117.html, last checked 7 Aug. 2007.

406

Technologies for the Wireless Future – Volume 3

[9] GPON 2.4G, http://www.itu.int/newsarchive/press_releases/2003/04.html. [10] “Broadband service architecture for access to legacy data networks over ADSL”, TR-012, DSL forum, Jun. 1998. [11] “Core network architecture for access to legacy data networks over ADSL”, TR-025, DSL forum, Sep. 1999. [12] UPnP QoS Architecture: 1.0, http://www.upnp.org/standardizeddcps/documents/UPnP_QoS_Architecture1 .pdf. [13] Session Initiation Protocol (SIP), IETF RFC 3261, http://www.ietf.org/rfc/rfc3261.txt. [14] M. Poikselka, A. Niemi, H. Khartabil and G. Mayer, The IMS: IP Multimedia Concepts and Services, John Wiley & Sons, Ltd, 2006.

11 Combined View of Future Systems Edited by Mikko A. Uusitalo (Nokia Research Center, Finland)

11.1 Introduction The aim of this chapter is to combine material from all the work of the WWRF, including the previous chapters, in trying to formulate a combined view of what future systems will look like. This chapter presents the basic characteristics of future wireless communication systems. The WWRF concepts of future communication systems derive from the respective requirements collected by the members, which stand as a prerequisite in order to render the concept realistic. Two families of requirements were identified, stakeholder requirements and system requirements. This requirements work was ongoing and not published outside of the WWRF at the time of writing this, in June 2007. One important lesson from the development of wireless world systems was that potential future services and applications should take into account user needs from the very beginning. This approach is essential to enable the market penetration and the economic success of future systems. On the other hand, the mobile community is in a dilemma over how to develop technical requirements from user needs for new systems, catering for an unpredictable future of about 10 years. Therefore, future systems should enable sufficient flexibility to match the unpredictable operator and user expectations from a service and economic perspective. In response to this challenge, the WWRF vision starts from the user. Due to the long time perspective, until 2017, technology push will play a role, as it is too challenging to tie all technology development to user needs that cannot be fully anticipated. The goal is to specify an environment to realize mobile communication services that follow the vision of the future wireless world: Users are able to access, anytime and anywhere, services tailored to their preferences and environment. Context-aware applications aim to provide relevant information about the users in order to supply services and applications that best suit their preferences and environment.

In other words, future wireless systems are aimed at providing high-data rate transmissions and highly sophisticated services, comparable to those offered by wired networks and Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

408

Technologies for the Wireless Future – Volume 3

even going beyond them. This will be achieved by disposing a system architecture that is distributed, component-based and open. Such architecture will support the capability of context-awareness, personalization and adaptation, in order to be able to potentially encompass new-necessary modules and thus support various types of service and application, anywhere and anytime. Additionally, the communication substrate of future mobile communication systems is likely going to be based on the Internet protocol (IP) family. This holds for both provider- and infrastructure-based, ‘classical’ system types as well as for smaller or nonconventional systems like personal area networks, broadband area networks, home area networks, etc. As the IP family is characterized by a large degree of flexibility, it should easily be able to support these specific networking types as well. Nevertheless, as the somewhat cumbersome introduction of even only basic mobility support into IPv4 has shown, a wide perspective should be taken to guide the further development of the Internet architecture to make the support for such networks stronger, more efficient, and simpler at the same time. Specifically, for an appropriate IP-based communication subsystem, a need to address a number of issues – the support of end-user devices, the network control and management, the data transport – was identified. The WWRF in general supports the concept of ambient networking. In order to cater for the vision of ambient connectivity combined with seamless mobility, the concept of ambient networking aims to provide a domain-structured, edge-to-edge view of the network control. In this way, an ambient network is expected to embrace the heterogeneity arising from the different network control technologies so that it appears homogeneous to the potential users of the network services. The vision is to allow the agreement for cooperation between networks on demand, transparently and without the need for preconfiguration or offline negotiation between network operators. End users are increasingly not just owners of a terminal or a PC; they own and effectively operate a network of devices in their homes, offices and around the body. Consequently, they are included in this network of cooperation and are treated as operators of special, low-complexity networks. By making every device a network, the network is the primitive building block of our architecture, allowing all types of network to be integrated into a larger system. The focus of this vision is on the cooperation between current and future radio access technologies, including the support of common radio resource management functions and the realization of mobility across them. In addition, the WWRF supports the advent of cognitive (reconfigurable) networks. In general, future systems will be characterized by the convergence of mobile systems and IP networks, so as to jointly cooperate over a common access infrastructure, namely the B3G radio access infrastructure. In such a context, varying traffic demands impose reconsideration of the generic network management approaches. This approach demands cooperation among previously competing networks, assuming that diverse technologies such as cellular mobile networks, BRAN/WLAN, sensor networks and DVB can be components of a heterogeneous wireless-access infrastructure and cooperate in an optimal way, in order to provide high speed and reliable connectivity anywhere and anytime. Moving one step beyond cooperation, cognitive (adaptive, reconfigurable) networks are seen as a further major facilitator of the B3G vision. Cognitive networks are capable of exploiting knowledge from previous interactions with the environment (this typically involves machine learning) and continuously adapting to external conditions and/or user needs. Adaptation is mainly realized by means of self-organization (self-configuration and self-management), or in accordance with autonomic computing principles. Reconfiguration of the system’s own infrastructure may affect more than just the traditional networking layers of the protocol stack. For example, reconfiguration in the PHY/MAC layers includes the selection of the most appropriate technology for

Combined View of Future Systems

409

operation, MIMO transceivers that adapt to propagation and traffic conditions, radio resource management based on cross-layer optimization, as well as the flexible, dynamic and highly efficient management and use of spectrum. On the other hand, reconfiguration in the middleware and application layers includes the selection of the most appropriate policies to manage network elements and technology reconfiguration. It is envisaged that these reconfiguration types will enable operators to further reduce capital expenditure and the related operational costs. To benefit from these advanced capabilities, novel reconfigurable platforms should cater for the networks’ adaptation to external stimuli. In addition, management functionality should ease this adaptation by means of fast and efficient algorithmic processes. Last but not least, all adaptations to the ever-changing environment requisitions should be performed transparently and securely, in order to guarantee for the maximum possible levels of security, privacy and trust. This will help avoid numerous unwanted situations, so compliance with the constraints that will be posed by security aspects should be considered indispensable.

11.2 Applications and Services A key enabler for future services is personal networking that supports communication in different kinds of environment: in the personal domain, in an ad hoc community, in a digital home, and in infrastructure-supported networks. Sensors contribute the necessary information of local context to this vision, while actuators contribute the means to affect the real-world context situation. Mobile middleware will provide solutions to hide the complexity of software development for described environments from service developers. The focus is on both ad hoc communities (devices cooperating without external infrastructure) and on infrastructure-supported networks (devices connected to each other and the external world through communication infrastructure). The requirement is to find a lightweight solution that suits both environments. A mobile middleware enhanced device will be the core of the personal communication infrastructure, building up the most appropriate end user system which can be autoconfigured for a given context. The goal is to specify an environment to realize mobile communication services that follow the vision of the future wireless world: Users are able to access, anytime and anywhere, services tailored to their preferences and environment. Context-aware applications aim to provide relevant information about the users in order to supply services and applications that best suit their preferences and environment.

The concepts related to the applications and services should support: ž The user requirements from emerging applications and services. ž The requirements from service providers, as services and applications should be compatible with service providers’ expectations. ž The requirements from the converging digital industries; it is self-evident that every innovation in the applications and services field should comply with industrial trends. ž The requirements from future service platforms. 11.2.1 User Interface and Service Adaptation Today, a great variety of mobile terminal devices exist which users can employ to access services. The devices are heterogeneous with respect to their capabilities to handle input and present user interfaces or the media they support. Additionally, network capabilities are

410

Technologies for the Wireless Future – Volume 3

change when using mobile devices to access a service. In consequence, the user interface and service adaptation capability is needed to properly handle these discrepancies and provide the best user experience. It provides functionalities to allow service developers to make services available through multiple devices using multiple modalities. The user interface and service adaptation capability contains, for example, the following subcapabilities: ž Device and Modality – supports and enables mobile multimodal applications in environments where device configurations, various content, varying user preferences and the user’s situation may change over time due to the user’s mobility; provides a set of basic multimodality-supporting functionalities including components for fusion, fission and foremost dynamic device binding. ž Device Gateway – has the main task of actually discovering and connecting to available devices and modalities within a user’s vicinity. 11.2.2 Personalization The personalization capability provides profiles and preferences of users and groups to the services, applications and components, and supports learning of user and group interests as well as user and group preferences. To support user feedback for profile learning, it is directly related to the services and applications. The personalization capability consists of different subcapabilities, such as: ž Profile Manager – groups all personalization data in terms of the identified data model classes and allows querying of all personalization data that are needed in a specific interaction context. ž Recommender – supports learning of situational preferences and interests by means of application-independent learning mechanisms (content-based, collaborative or hybrid approaches) and provides recommendations based on this information. 11.2.3 Group Support The group support capability supports applications and services dedicated to groups by providing a description of the operational group context, managing groups and managing group preferences and information. It consists of two parts: ž Group Management – supports the managing of groups, i.e. creating, deleting, updating, managing lifetime, storing and dynamically updating templates for groups, providing mechanisms to ensure trustworthy and private communication within and towards the groups against threats. ž Group Evolution System – monitors group behaviour to learn new or modify existing group templates, which are used for the group creation. 11.2.4 Context Awareness The context awareness capability takes care of raw, interpreted and aggregated context data. This capability handles context data related to individual users and to groups of users. It supports the service developer by providing users’ and groups’ current context information

Combined View of Future Systems

411

through well-defined interfaces. New context information and changed context information can be notified to interested components and application services, and context information can be requested from the context awareness capability. The context awareness capability contains the following subcapabilities: ž Personal Context – receives raw data, user behaviour, service discovery data and application requests from the user interface and service adaptation capability and directly from services and applications. ž Group Context – produces interpretations of the environmental context of the group, the possible context the group is in and the possible contextual personalized control actions to be performed. ž Context Management – context framework specifications are needed to describe the representation, exchange, interpretation and prediction of context, as well as upper-layer context reasoning. 11.2.5 Privacy and Trust Services and applications deal with data related to the user, which raises the issues of trust and privacy of the personal user data. It is not just one user sharing personal user data, as is the case with traditional single user-application interaction; groups of users can be involved. The trust and privacy management of personal user data needs to be supported by a suitable policy system. The privacy and trust capability contains the following subcapabilities: ž Privacy Policy Storage and Management – deals with the management of user- and group-related privacy policies. ž Trust Engine – ensures that the system can be trusted to perform as is declared in the policies. 11.2.6 Service Usage and Provisioning The service usage capability covers all aspects related to the service usage; in particular it covers every step in the ‘timeframe’ between service discovery and service offering. The service provisioning capability holds a repository of services known to the system, their descriptions and properties, and offers functionalities for service discovery, proactive service provisioning and service composition. The service usage and service provisioning capabilities have a very strong relation to each other. Service usage contains the following subcapabilities: ž Service Discovery – the ability to discover all services already deployed in the service repository, according to user current context and user preferences. ž Service Composition – if a requested complex service is not available but can be composed from available elementary service parts, a service composition may take place in order to provide the requested service. Service provisioning contains the following subcapabilities: ž Service Catalogue Provisioning – stores service information in terms of description, properties, semantic information, etc.

412

Technologies for the Wireless Future – Volume 3

ž Proactive Service Provisioning – the functionality to provide services to users when they are in certain situations. The situations are derived through analyzing the current context of a user. This way, a certain location can trigger services, e.g. a bus stop triggers a service to buy tickets. This functionality also has to take care of selecting the services that best match the user’s needs in certain situations. 11.2.7 Operational Management The operational management capability supports the management of the whole lifecycle of services. To perform this, it needs access to information about services stored in the service provisioning capability, and access to applications and application services, and to the user agents running on users’ devices. To manage the privacy aspects, it also needs access to the privacy and trust capability. Operational management contains the following subcapabilities: ž Service Management – focuses on the knowledge of services (access, connectivity, content, etc.) and includes all functionalities necessary for the management and operation of communications and information services required by or proposed to customers. ž Resource Management – maintains knowledge of resources (application, computing and network infrastructures) and is responsible for managing all resources (e.g. networks, IT systems, servers, routers, etc.) utilized to deliver and support services required by or proposed to customers. It also includes all functionalities responsible for direct management of all such resources (network elements, computers, servers, etc.) utilized within the enterprise. ž Data Collection – all data collection processes interact with the resources to collect administrative, network and information technology events and performance information for distribution to other processes within the enterprise. Responsibilities also include processing the data trough activities such as filtering, aggregation, formatting and correlation of the collected information before presenting to other processes. 11.2.8 Charging and Billing Accounting is the process of collecting details of resource usage for the purpose of billing a customer. The resource usage consists of users’ activities while accessing network resources, like the amount of time spent in the network, the services accessed and their QoS while there, how often specific content was accessed, and the amount of data transferred during the session. After charging-related details have been collected, the accounting data is processed in order to charge the relevant business entities. With the research of new mobile services, the role of billing and charging is often forgotten, though they are vital for running a business successfully. Without billing/charging it is not possible to gather money from consumers and therefore possibilities for different business models becomes difficult. Also, money flows usually start from the customer who consumes the service. If the payment of the customer is missing, the compensation in value network is impossible or at least difficult. It is important to define systems such that they adapt to flexible charging methods and concepts; for the deployment of new services it should be easy to take an appropriate charging method in use. Some examples of the methods and concepts are: ž Offline Charging – the information of the transactions is collected and mediated, the services are priced and the charging from the accounts take place either after the service deployment or during it.

Combined View of Future Systems

413

ž Online Charging – the requests for the transactions are collected and mediated, the services are priced and the charging from the accounts take place before the service deployment (e.g. before the start of the video stream). ž Prepaid – the customer pays for the service before the service is delivered. ž Post-paid – the customer pays for the service after the service has been received or used. ž Hot Billing – the charging tickets are created during the usage of the service. OMA defines two charging interfaces: offline and online. Offline charging is a mechanism in which charging information does not affect the service rendered. The online interface allows real-time charging. Both charging interfaces can be categorized into two distinct charging submodels, namely the event-based charging model and the session-based charging model. In the event-based charging model, each service usage is reported with a single charging record or a resource usage authorization procedure. In the session-based charging model, the same service usage occurs within an end user session. The service usage is reported with several charging events and the creation of one or more charging records in offline charging, or performance of a credit control session in online charging. There are lots of challenges related to charging/billing arising from changes in both the technological and business environment, like how to produce reliable information on transactions of new services, used in heterogeneous environments, almost or fully in real time. Regarding networkcentric charging and billing, for example, operators can be seen as natural players in security and billing issues. One of the challenges is how to support electronic payment, especially micropayments in P2P environments. The networkcentric charging and billing enablers may not be available in this case, thus solutions residing on the terminal side come into the picture. These may include some already-mentioned solutions such as prepaying, but also credit card companies might act as billing or micropayment providers, with secured mechanisms installed on the terminal, for example. Another challenge is charging in service roaming. Assume that a user has subscribed to a service where personalization information and context information, for example, play important roles. The charging of the service usage should be dependent on what features, or information, are available each time a user is accessing the service. In addition, some of the features may be requested on a call-by-call basis, and the cost of the service depends on the used features. The more features are used, the more costly the service is. Typical information whose availability may change depending on the location (be it a network or a physical location) is the context information. Also, the price of using the context information may change from country to country, from service provider to service provider, etc. Currently the network operators compare the total amount of calls and costs taken by the visiting subscribers in their networks, and make the clearance actions according to the bilateral agreements. In the case of service roaming (other than voice telephony service) a player owning a customer may not necessarily be a network operator, and the commercial and contractual situations are more complicated. A bilateral agreement is needed between the service providers in order to agree on the charging and billing principles and mechanisms, and to make the context information available in the environment (network, services) where the user is visiting. The charging and billing in the service roaming has to be as simple as it is when it is a question of the voice telephony service in the GSM networks: the charging record will be created for each call and the clearance actions are based on the bilateral roaming agreements between the operators. As a totally new charging and billing system might not get wide acceptance, it should be investigated how the charging mechanisms implemented for the IMS system, for example, could be exploited when deploying new advanced services. In the IMS-based

414

Technologies for the Wireless Future – Volume 3

services the charging records are collected by the core network (SGSN/GGSN) and the CPS element of the IMS system. In addition, it is important to investigate the roles and relationships of different players in charging and billing when providing services. The challenge of charging in service roaming also includes the following issues: ž The implications for the business relationships between the different players in the service delivery chain should be worked out, e.g. SLAs between the service providers, use of the context information made available by a service provider to another service provider, compensation fees between the service providers, etc. ž Types of context information etc. should be universal, irrespective of the country and the network, in order to make their pricing easy. ž The service architecture should also consider a function for collecting charging data of a used service, allowing invoicing based on the used context information, and compensation fees for the context providers. Other important requirements for billing and charging include: ž Providing easy access/overview for the customer about their charging/billing situation. ž Being open and supporting new business models, like the ones introduced by Google, for example. ž Facilitating mobile payment and banking applications. 11.2.9 Service Layer Mobility Support The service layer mobility capability comprises the following five subcapabilities: ž Terminal Mobility – roaming of users with their devices between different access networks; usually supported by the network layer. ž Personal Mobility – reachability of a user on different devices; usually supported by overlay networks employing different addressing schemas than IP addresses. ž Service Mobility – availability of services in different access networks or in different network provider domains. ž Profile Mobility – access to a user’s profile information on different devices and in different access networks. ž Session Mobility – transfer of an active service session between different devices and possibly across different access networks. Whereas terminal and personal mobility are already available in today’s networks, the realization of service mobility, profile mobility and session mobility is still to come. This is a prerequisite for a next-generation service platform which can provide services in heterogeneous access networks and on diverse devices. 11.2.10 Peer-to-peer Services Support Peer-to-peer services are directly executed between two or more end users acting as service providers, without the need of the core network service infrastructure. Whereas a strict definition of such peer-to-peer services excludes the (centralized) service provider or operator from service provisioning, in some cases additional service provider or operator services are required to ensure service execution, for example negotiation, contract enforcement, etc.

Combined View of Future Systems

415

For a next-generation mobile service platform, it should be regarded as a general requirement to support peer-to-peer services in order to exploit ubiquitous communication networks such as ad hoc networks. 11.2.11 Negotiation Support Multidomain, multiprovider environments (ranging from BAN to Internet) will require advanced mechanisms to negotiate required QoS and respective service parameters, cost models, permissions, etc. Consequently, negotiation can be found on all layers as part of end-to-end reconfigurable systems.

11.3 IP-based Communication Subsystem The communication substrate of future mobile communication systems is likely going to be based on the Internet protocol family. This holds for both provider- and infrastructure-based, ‘classical’ system types, as well as for smaller or nonconventional systems like PANs, BANs, home area networks, etc. As the Internet protocol family is characterized by a large degree of flexibility, it should easily be able to support these specific networking types as well. Nevertheless, as the somewhat cumbersome introduction of even only basic mobility support into IPv4 has shown, a wide perspective should be taken to guide the further development of the Internet architecture to make the support for such networks stronger, more efficient, and simpler at the same time. Specifically, the need to address a number of issues for an appropriate IP-based communication subsystem has been identified: ž support of end user services ž network control and management ž data transport. 11.3.1 Service Support Layer It is required that future networks provide service support capabilities so that new value-added services can be efficiently developed and offered to end users. The Internet has taken a pragmatic approach here in keeping the network itself simple and putting any value-addition functionality at the network edge; this approach has allowed the rapid introduction of a vast array of new services. Nonetheless, these services have – so far – been strictly separated from the underlying network, isolating them from possibly relevant information. There is, therefore, a need for a more generic model of service creation support or service creation environments. From this standpoint, amongst others, the following major requirements have to be met in future networks: support for value-added services and support for service creation. 11.3.1.1 Support of Service Adaptation Networks should be capable of instantly adapting and maintaining services when the underlying network and terminal resources change. Two basic cases should be considered: 1) an ongoing service changes from an access system to another; 2) an ongoing service is transferred from one terminal to another. In both cases, it may be necessary to adapt the service attributes (e.g. presentation format, resolution, etc.) to the changes in the resource allocation and the capabilities of terminals.

416

Technologies for the Wireless Future – Volume 3

11.3.2 Provision of Flexible Environments for Applications and Services Networks should support applications that can utilize network and transport information to adapt to network conditions in a user-friendly manner. Networks should support provision of home-subscribed services and real-time provisioning of services (e.g. provision of subscription to local services advertised to inbound roamers). Networks should support common representation of user service profiles, that is, standardized data attributes for facilitating new service creation and improvement of convenience for subscribers. Service profile structure should be upgradeable. Networks should support capability for users to modify their service profiles within the limit of their subscriptions. Modification of the profiles should be possible through any access systems. 11.3.3 Network Control and Management Layer In its current form, the IP architecture is, compared to traditional telephony networks, relatively ‘thin’ in network control and management. For future networking applications, specifically for networks that integrate one or several wireless hops, this – in principle desirable – thinness might not suffice and additional control functionality could be necessary. While not all of this functionality would belong, strictly speaking, to the IP network as such, it will still have to be integrated into the Internet architecture at large. 11.3.4 Business Models and Functions Classification in Space Time Domains Business models refer usually to ‘economic’ issues. In the following, ‘resources’ refers to the set of radio resources (code, time slot, frequency) or HW/SW (RF part, backbone, wired segments, antennas, sites). The business models relate to the business transactions occurring when resources are shared. Two different sets of business models are considered: user-to-operator, and operator-to-operator. 11.3.4.1 User-to-operator This transaction occurs between one user and one or several BSs or APs, i.e. between one user and one or several cells. Each user deals directly with the APs or BSs. The BSs/APs are then in charge of granting the resources to the winning users. 11.3.4.2 Operator-to-operator This transaction occurs when several operators have to share resources dynamically. The two followed approaches are: operator-to-operator via meta-operator and operator-to-operator in peer-to-peer. Operator-to-operator via Meta-operator (MOP) The meta-operator is first in charge of collecting the different operators’ resources demands and offers their associated biddings. Then the MOP is also in charge of assigning the resources to the winning bidders. In this case, the ‘uncorrelated’, ‘partially uncorrelated’ and ‘correlated pool’ principles could be applied. For the time being, only the ‘correlated’ approach is considered.

Combined View of Future Systems

417

Operator-to-operator in Peer-to-peer The transactions occur directly between operators in peer-to-peer without the need of an MOP. The transactions also rely on a bidding strategy. The relevance and applicability of the business models, mapped to space domain, vary at certain levels, as explained below. User Level The advanced spectrum management (ASM) applies within a given cell encompassing several users. Each user interacts with the base station (BS) or access point (AP) to negotiate the data rate (i.e. amount of spectrum) with respect to the link to set up between the user and the BS or AP. Cell Level The flexible spectrum management (FSM) applies between a limited set of cells on a limited coverage area. Spectrum is shared on a cell-by-cell basis between these cells. Regional Level The FSM applies for a large area, i.e. for a large set of cells. Spectrum is shared on a spatial basis, implementing sharing between adjacent areas. 11.3.5 Quality of Service (QoS) Networks should support QoS mechanisms that are not tied to any specific link technologies, but which instead provide a common basis for QoS coordination across multiple access technologies, and standard-based interoperability across multiple network domains to provide an end-to-end QoS solution. Sufficient quality of service has to be provided to applications and services in the future. New applications will be developed with new requirements on service and network support. At the same time, new access technologies will be developed that have different characteristics. New terminals with different capabilities and requirements will be developed, and users will have different requirements and different needs to fulfil depending on where and what they are doing. Applications should request resources from the network at start-up time and continuously receive network capability feedback from the network. These negotiations should use standardized APIs to support simplicity and inter-access support. The specific link layer reservations are hidden from the application. Networks should support various QoS schemes supporting different classes of services and applications. That is, the network should support simple and scalable solutions such as differentiated services for flexible, widely used and less demanding applications. For applications with strict QoS requirements, a less scalable QoS scheme, which might require special or explicit QoS signaling, should also be supported. The provisioning of QoS resources need to be coupled with application signaling to allow efficient usage of QoS resources. Initiating and controlling the coordination between the QoS provisioning and the application signaling should be done by either the network or the user. Support should be given for adaptive QoS; for example, networks should be able to specify bandwidth ranges or multiple bit rates at which an application can operate, and associated control or prioritization values that will insure an agreed level of performance to enable resource allocation decisions. Capabilities for supporting interoperability with other Internet services should be supported.

418

Technologies for the Wireless Future – Volume 3

Networks should have the capability to support end-to-end QoS across multiple operators and a framework for QoS negotiation between operators, as well as between users/terminals and the network. Networks should support different QoS parameters in uplink and downlink for wireless links. Network resource (e.g. CPU processing, bandwidth) consumption for QoS control should be minimized. QoS mechanisms should scale to large numbers of flows and large networks. Networks should be able to control the QoS in the unit of packet, flow and session. The network function(s) should be allowed to report the status of QoS parameters or classes to the application server and/or end users involved in a session. 11.3.6 Duplexing The spectrum allocation impacts the duplexing choice, which significantly impacts the air interface design and places limitations on the choice of both the air interface features and the access control and resource management mechanisms. Duplex method selection is influenced by: ž Choice of TDD/FDD – TDD is intended for local area coverage, whereas FDD is for wide area coverage. Hybrid schemes, such as hybrid division duplexing, can be considered for flexible coverage of both scenarios. ž Spectrum Allocation – if unpaired spectrum is allocated, FDD cannot be used, whereas both FDD and TDD options are possible when paired spectrum is allocated. ž Traffic Symmetry/Asymmetry – TDD enables asymmetric allocation of degrees of freedom between uplink and downlink. ž Need for Link Reciprocity – to support channel estimation at transmitter, TDD or hybrid schemes such as band switching support channel reciprocity, however differences in the transmitter/receiver RF chain may limit link reciprocity and should be carefully considered. ž Synchronization and Link Continuity Requirements. ž Distributed Control – e.g. terminal-to-terminal. 11.3.7 Inter-cell Coordination In integrating mobility into a network, additional issues that pertain to the integration of the radio access system into an Internet architecture have to be taken into account as well. One example is the coordination of resources among multiple cells, which can increase efficiency but requires support from the backhaul network to provide necessary information exchange. ž Interference Management – coordinated coherent transmission is a powerful tool, but puts additional backhaul requirements on the network. ž Techniques to Address Inter-cell Interference Coordination: ž interference avoidance ž interference averaging ž interference smoothing. ž Alternatives to Detailed Interference Management: ž fast cell switching ž partial handoff with MIMO systems.

Combined View of Future Systems

419

ž Paging, System Discovery – difference from earlier wireless systems: relaying considerations. ž Self-organization – probe the environment and adjust accordingly a number of parameters, such as antenna configurations, transmit power, etc. 11.3.8 Integrated OA&M (Operation, Administration and Management) ž Unified Network Management – common management should be supported for all nodes in the network with a standardized protocol. ž Easy, Low-cost and Scalable OA&M – easy and low-cost network management and operation should be facilitated for all service networks. Integrated OA&M should enable easy deployment of new services or expansion of current services. ž Distributed Architecture – network management functions should be decentralized in easily upgradeable logic modules located separately through the network. 11.3.9 Transport and Higher Layers in Internet-based Networks 11.3.9.1 Connectivity Session Management Connectivity session management is about setting up, maintaining and terminating connectivity as requested by service session management. The connectivity session management is the capability (or a set of functions) to determine the specifications of the connectivity sessions based on the requirements of the service sessions; to distribute the information of the connectivity sessions to the relevant network entities and terminals for setting up, maintaining and terminating the connectivity, if required; and to feed the capability information of the network entities and terminals back towards the service sessions. The specification parameters of a connectivity session include network addresses and QoS parameters. ž Suitable interfacing between service and connectivity session managements is required, each at appropriate abstraction level and independent of technology, to support unbundling the innovation cycles of applications and networks. ž The network should be able to respond to connectivity requirements of service sessions, independently of the content type. ž Network should provide notifications and statistics to service sessions, so that a service session can react to network conditions in order to, in turn, change connectivity if required, or account for the use of resources based on information originated in the network. ž Connectivity session management should support different connectivity and flow models, for example peer-to-peer, multiparty, etc., and this should be easily expressible in the service session management. ž Connectivity session management should be amenable to many different types of end links, whether they are wired or wireless, including mixtures, so that, for instance, mobility support is typically available, i.e. it is not the exception but the regular case. ž Connectivity session management should be a scalable and distributable control of each service session. ž Connectivity session management should have the capability to determine and maintain identification for each connectivity session uniquely. ž Connectivity session management should give priority to the establishment of a prioritized connectivity session, such as an emergency call or prioritized call.

420

Technologies for the Wireless Future – Volume 3

Data Transport Transportation of data should be performed as efficiently as possible with regards to application and service requirements on one hand, and network capability on the other. With the introduction of new access technologies, applications and services, as well as new terminal capabilities and user preferences, the dynamicity increases. In order to support as efficient a transportation of data as possible, the different networks, terminals and applications must be able to adapt to new environments. All users’ data traffic and signaling should be on IP transport. ž Interfaces between next-generation IP networks and legacy transport networks are necessary. ž Transport network architecture and operation should be independent of the utilized transport technology. ž Transport networks could support broadcast and multicast. ž Access technologies should be separate from transport technologies; the access technologies utilized should be transparent to the common and standardized transport infrastructure. Connectivity and Routing Multihop connectivity and routing must be supported both in the proximity short range and in WANs at the IP and other layers, which is easy to support, is independent of the underlying MAC and PHY technologies, and is future-proof. Routing/Switching The IP-based communication system should adhere to the routing and switching solution defined in the IETF RFCs. Advances made to classical routing and switching, such as QoS routing, MPLS, multicasting, naming and addressing, and IPv6, should also constitute part of the requirements. Naming and Addressing Networks should have the capability to separate the network address from the subscriber name or number. Addresses could support any addressing schemes. Networks should allow for an identification that can be assigned to a user/subscriber as long as they allow it, while being independent of the addressing scheme given by the network technologies. Networks should support static and dynamic addresses for mobile terminal interface(s). Networks should support associating public and/or private addresses for mobile terminal interface(s). Networks should support mapping between subscriber number and addresses/URLs currently used within the network. The number or identifier of the destination and/or source should not change, at least during the session, even when the node moves to another access point or access technology network. 11.3.9.2 Mobility Management Mobility management should be customizable to take into account various factors and policies, enabling different network systems architectures, business models and/or user values

Combined View of Future Systems

421

Mobility management should accommodate dynamic changes of factors and policies, enabling adaptation to changes in user and network environments. Mobility management should be on a per-session basis to modularize network operation (mobility management and session management) and facilitate user interaction with the network; the network should support session management independently of the mobility management. This means that the session management will be an independent operation and it should be able to be operated over multiple access network technologies independently of administrative domains (e.g. roaming). Handover Handover between networks should assure optimal network selection and service continuity. Handover should support seamless mobility with minimum delay and data loss, without compromising on security, to minimize any negative impacts on user perception of service quality. Different but coherent handover schemes could be deployed to best adapt to different parts of the systems architecture in order to support seamless mobility. Requirements for Mobility Support The mobility management should provide means to support terminal, user, session and service mobility. Access Security Security must be implemented to prevent unauthorized access as well as to ensure privacy. Access restrictions and limitations in the access rights of a user must be possible. A mobile terminal should be authenticated when it attempts registration and access. The terminal should also be able to authenticate and authorize the network(s) offered to it. Location Confidentiality The mobility management should hide the user’s actual location if the user wishes this. Trusted entities (e.g. user, network) should be able to get user location information for route optimization and/or location information service. The location information of particular users should be concealed from nontrusted entities. Route Optimization The path for users’ data flow should be optimized in terms of length, load and/or cost. Optimum paths should minimize latency and allow for better resource management. Resource Optimization Resources are to be used as efficiently as possible in order for as many users as possible to simultaneously be connected and communicate using the shared resources. Support of a Moving Network Consisting of Several Nodes By controlling all nodes in a moving network as one group, the operator can efficiently manage their mobility, since it can reduce the network resource consumption for signaling. Personal area networks and mobile terminals in a moving vehicle can be controlled as a moving network, for example. Also, user data traffic sent to and from nodes within the moving network can be concatenated, thus saving network resources.

422

Technologies for the Wireless Future – Volume 3

Support of Mobility between Heterogeneous Access Technologies A mobile terminal or mobile router should be able to address many access technologies – one at a time or several simultaneously. A mobile terminal can also be part of a mobile network. Examples of access technologies are W-CDMA, cdma2000, IEEE 802.11 and ETSI HiperLAN/2. Support of Roaming between Heterogeneous Network Management Systems Taking the current and past development network technologies as an indicator, we can expect the networks to vary not only in the used access technology but also in the used Authentication, Authorisation and Accounting (AAA) and management architectures. To allow for ubiquitous communication in such heterogeneous environments, the user terminals need to support not just different access technologies but also different access and authentication mechanisms. To allow a user to roam into a foreign provider’s network without having to establish contractual relations with this provider, both home and foreign providers need to support mechanisms for mapping their AAA and management mechanisms among each other. Seamless Handover between Networks The mobility management should be fast providing seamless handover between networks. This may include negotiation between the networks where crucial information (e.g. QoS or AAA information) is exchanged between them before the handover and where mobility-related signaling may be delegated to another interface or node. Support of Mobility between Different Administrative Domains Mobility management should support mobility between different administrative domains. For example, a user logs on to an operator’s network, and then moves to another operator’s network while being connected to or being active in the network. The change of domain can be triggered by, for example, the user changing access technology and hence operator, where the user can have multiple subscriptions with different operators, or the user moving from one domain area (e.g. country) to another domain area (e.g. country) with the same operator. Modularity To meet different requirements for mobility, such as regional difference of mobile node density and required QoS difference for each session, mobility management should be modularized. For example, one module for urban area management, one for rural area, one for real-time service and one for best-effort service should be prepared. This modularization allows operators to select and adopt modules which they want. Keeping Required QoS Depending on the user’s preferences, the QoS of all connections after the handover should be kept at least at the same level as that negotiated prior to the handover, whenever is possible.

11.4 Access Network 11.4.1 Flat Full-IP Access Architecture Future wireless systems are recognized as those that can achieve high-data rate transmissions and provide adequate capacity, cost efficiency and highly sophisticated services, comparable to those offered by wired networks, for a variety of applications, such as interactive multimedia, VoIP, network games or videoconference. Moreover, future systems are expected to

Combined View of Future Systems

423

be based on IP technology yielding into a common, agile and seamless all-IP architecture design, supporting scalability and mobility. As long as none of the currently known RATs is mature enough to satisfy the aforementioned criteria, the idea of diverse RATs being optimally combined and coordinated under a global infrastructure stands as a basic prerequisite for the consolidation of future systems. 11.4.2 Wired Technologies Access networks will make use of the plethora of existing and future wired technologies to connect fixed components. Wired technologies in use will range from simple, cheap technologies like Ethernet over CAT-V copper cables used in home environments, to high-end, high-performance technologies like optical fibre forming the backbones of large networks. All these wired technologies will have to support IP-type protocols on top of them. This requirement could raise issues for operators of traditional telecommunication networks, but since even here a transition to IP support is well under way or has already taken place, this should not place any significant additional burden. 11.4.3 Wireless Ubiquitous Coverage The vision of the WWRF is based on the assumption of ubiquitously available communication means. Despite today’s widespread availability of wireless and wireline communication systems, the end user is not yet able to take full advantage of the deployed infrastructure, as it is to a large extent organized as vertically separated systems. The objective of WWRF is to tear down the boundaries existing between the individual communication networks and systems and eventually accomplish the vision of ambient connectivity characterized by the unrestricted cooperation between these networks. The main challenge communication research faces in realizing the vision of ambient connectivity lies in the control layer of the communication systems. Plain connectivity between networks is today easily achievable, as the transportation of IP packets has become the common denominator for the user-plane of most networks. There is, however, an increasing divergence in the network control layer. Different control mechanisms are deployed to facilitate services like VPNs, security, mobility, QoS, NAT, multicast, etc. This lack of a common control layer for enabling cooperation between multiple networks represents a crucial challenge both technically and from a user perspective. Usage scenarios that should be realizable in the mid-term future include utilization of multiple devices, multiple networks and multiple access technologies in an integrated fashion. This is not easily controllable or manageable with today’s technologies. WWRF supports the concept of ambient networking, which aims to provide a domain-structured, edge-to-edge view of the network control. In this way, an ambient network is expected to embrace the heterogeneity arising from the different network control technologies so that it appears homogeneous to the potential users of the network services. The vision is to allow the agreement of cooperation between networks on demand, transparently and without the need for preconfiguration or offline negotiation between network operators. End users are increasingly not just owners of a terminal or a PC; they own and effectively operate a network of devices in their homes, offices and around the body. Consequently, they are included in this network of cooperation and are treated as operators of special,

424

Technologies for the Wireless Future – Volume 3

low-complexity networks. This approach generalizes to different kinds of network that are currently appearing, such as inter-vehicle networks, body area networks and sensor networks. By making every device a network, the network is the primitive building block of our architecture, allowing all types of network to be integrated into a larger system. 11.4.3.1 Addressed Requirements The ambient networks concept addresses the requirements for networks identified and agreed in the WWRF. From the list of general system requirements, especially the support of different business models, the coexistence of legacy and new networks, the possibility for evolutionary deployment and the existence of value-added interfaces are reflected in the three open reference points included in the ambient network concept. The architectural principles are best reflected in the modular structure of the ambient control space, allowing the existence of independent functional blocks and enabling the inclusion of innovative solutions for security and network control and maintenance. As the ambient networks concept is meant to embrace both wired and wireless networks, it is also influenced by the requirements listed for access networks. The focus of the ambient networks vision is on the cooperation between current and future radio access technologies, including the support of common radio resource management functions and the realization of mobility across them. The ambient control space contains functional blocks taking care of these tasks. 11.4.3.2 Ambient Networks Features Ambient networks are characterized by the presence of a commonly shared and harmonized network control layer, referred to as ambient control space. This control space extends current mobile networks by several innovations, such as network composition (beyond simple Internetworking), enhanced mobility, embedded support for service providers and effective handling of heterogeneity. Network Composition One basic mechanism of ambient networks is the dynamic, instant composition of networks. Composability as required for ambient networking goes beyond what the Internet and mobile networks provide today. Internetworking shall not only happen at the level of basic addressing and routing; additional functions for incorporating higher-layer support (such as content distribution or service control functions) are required. Ambient networks will deploy a universal framework for network composition, as an approach for building a unified communication environment out of the resources of individual networks, which can be specialized for particular types of access technology or business model. Network composition in ambient networks has to function across operator and technology boundaries, provide a security framework and be executable without user involvement. In addition, the execution of the composition process has to be rapid in order to follow fast topology changes, as expected for example for mobile personal area networks (PAN). Another example where instant network composition is relevant is the joining of operator networks, which today are based on explicit, human-negotiated and -executed agreements, and are therefore too slow and cumbersome to set up, e.g. for rapid service roaming. Rather, access, interconnectivity and service level will be associated on the fly between two networks.

Combined View of Future Systems

425

Mobility Existing IP-based mobility solutions target either intra-domain mobility within a static network architecture or roaming solutions across domain boundaries. Ambient networks focus on integrated mobility concepts not only for both of the above scenarios, but also for localized communications, e.g. in PANs, and device-device interactions. In dynamically composed network architectures, mobility of user group clusters can support effective local communication. Furthermore, mobility mechanisms must interact efficiently with the control interfaces needed to enable quality of service and optimal routing and rerouting of individual multimedia flows. An ambient networks mobility solution will have to work well across business and administrative boundaries, which requires solutions for the roaming and security issues between different operators. Service Support Ambient networks seek to harmonize functionality and ease service deployment in a dynamic mobile networking world. Service provisioning today often lacks feedback from the radio network environment, as well as the capability to demand treatment of flows beyond simple QoS. While these advanced functions are offered optionally, they are deeply embedded in the ambient networks concept. Feedback and registration of triggers is handled by a context information base which can provide notifications to applications or actively influence the routing in case of changes in the networking situation. On the other hand, media routing will enable applications to specify the wanted behaviour of media flows in the delivery chain to the user. The ambient network layer should take full control of routing and adaptation of these flows, locally optimized to match the network and terminal conditions. Heterogeneity Ambient networks will be based on a federation of multiple networks of different operators and technologies. On the one hand, this leads to increased affordability of ubiquitous communication, as the user has full freedom to select technology and service offering and the investment needs for new networks are reduced. On the other hand, networks will have to integrate the capabilities of different technologies into an end-to-end, seamless and secure solution for the user. Ambient networks take a new approach to embrace heterogeneity visible on different levels, such as link technologies, IP versions, media formats and user contexts. Diversity of access links, especially of links provided by mobile networks, is supported by a generic link layer concept, which will efficiently enable the use of multiple existing and new air interfaces. Ambient networks also consider the implications of heterogeneous wireless systems on the overall network, especially the impact on end-to-end QoS and multimedia delivery. In particular, the novel concept of network composition will include the negotiation between different networks regarding their capabilities, e.g. QoS. Ambient networks provide an integrated framework for enhanced support of multimedia delivery in heterogeneous environments by embedding novel media flow routing and transport functionalities into the overall ambient network architecture. 11.4.3.3 Architecture and Components of the Ambient Control Space An ambient network is supposed to provide well-defined reference points to other ambient networks and to service platforms or applications, to provide (at least a subset of) the ambient

426

Technologies for the Wireless Future – Volume 3

Ambient Service Interface

Ambient Control Space

Ambient Network Interface

Trigger & Context Mgmt. Composition Mgmt.

INQA & SLA Establishment Network Mgmt.

Bearer & Overlay Mgmt.

Ambient Connectivity

Flow Mgmt. & MRRM

Ambient Resource Interface Connectivity & Generic Link Mgmt.

Ambient Network Interface

Mobility Mgmt.

Security domain Mgmt.

Figure 11.1 Illustration of the ambient control space and its three external reference points

control space functions and to be able to dynamically compose with several other ambient networks to form a new ambient network. Cooperating ambient networks could potentially belong to separate administrative or economic entities. Hence, ambient networks provide network services in a cooperative as well as competitive way. When ambient networks and their control functions are composed, care must be taken that each individual function controls the same resources as before. Figure 11.1 illustrates the logical organization of the control space internals. It consists of a collection of control functions that cooperate to implement specific control functionality. These control functions exist within the overall control space framework. Besides offering common, required functionality, the framework structure also enables modularization of the control space. This modularization enables operators to adapt their networks’ control functionality to their specific needs, while maintaining global interoperability, even with other networks that do not implement the same subsets of control space functionality. A second advantage of a modularized control space is the dynamic, plug-and-play integration of new control space functionality during the lifetime of a network. Figure 11.1 also illustrates how the common, distributed control space encapsulates both legacy and future Internetworking infrastructures and employs generic reference points that are independent of specific network architectures. Network entities interact with the new control space through three different reference points. Higher-layer applications and services use the ambient service interface (ASI) to access the control space functionality. Connectivity resources interact with the control space through the ambient resource interface (ARI). Finally, the ambient network interface (ANI) facilitates communication between the control spaces of different networks, creating the shared, common control space that enables the advanced Internetworking capabilities the ambient network project aims to achieve. 11.4.3.4 Connectivity Abstractions The control functionality gathered in the ambient control space requires the ability to interact with the user plane resources. An abstraction framework has been developed to realize this

Combined View of Future Systems

427

interaction and also keep the ambient control space independent of the network technologies deployed in the connectivity plane. The connectivity abstractions – depicted in Figure 11.2 – define different views of the underlying connectivity. Almost all types of network, regardless of whether they use circuits, packets or other data transport mechanisms, can be described as a number of nodes and links. This generic view of the network resources forms the basis for the abstraction model. The control space functions, however, do not operate on this detailed view of links and nodes. The abstraction mechanism only exposes a subset of these entities to the ACS. This subset forms a common representation of the underlying network that is visible though the ARI and can also be influenced through the ARI. The resulting connectivity graph forms the basis for all connectivity-related operations. The ACS representation of an actual transfer of data over these connectivity resources is referred to as ‘flow’. The applications or services operating on top of the ACS are, however, not aware of details like the flows that are used to transport the data. They operate on a higher-level view of the connectivity referred to as bearer. The bearer is an end-to-end communication mechanism that is offered at the ASI. It hides the implementation of the connectivity and provides the end-to-end transport service the application requires. The control space functions manage the mapping of bearers to flows, as well as potential updates of this mapping, e.g. in the case of a mobility event. In a simple case, a bearer provides little more than what a flow provides, which is often sufficient for basic services like a best-effort file transfer. For other applications, a bearer can also be more sophisticated. In such a case, additional functionality provided by the ACS functions can be dynamically included to enhance the bearer, e.g. by providing QoS reservation, mobility or customized media manipulation and routing capabilities. For the application however, this process remains largely transparent. 11.4.3.5 New Air Interface Next-generation wireless systems will support much higher bit rates than those in third-generation systems. Originating from ITU-R, widely-stated goals for wide-area

Application

Ambient Network B Session

Application

Ap pl ic

at

io ns

Ambient Network A

B Ab ear st er ra ct io

n

Bearer Endpoint

Bearer Endpoint

Network Boundary Intermediary

Bearer

Flow Endpoint

Flow Endpoint

Ab F st low ra ct io n

Flow Transit

ity

Flow

C

on

ne

ct

iv

Node

Path

Figure 11.2 The ambient network connectivity abstractions

Link

428

Technologies for the Wireless Future – Volume 3

deployments include a peak aggregate bit rate of about 100 Mbps at mobile speeds up to 250 kmph. For local hot spots, with low mobility, up to 1 Gbps peak aggregate bit rate requirements are foreseen. A flexible air interface concept should allow burst transmissions, on demand, to and from individual user terminals at up to a substantial fraction of these bit rates. It should be reconfigurable, to adjust to varying user requirements and radio environments. It should also be scalable in cost, performance and power consumption, economically accommodating user terminals, such as sensor devices, which communicate at very low bit rates (e.g. a few hundred bps) and consume very low power. New air interface design should allow for backward compatibility with at least some 3G and earlier air interfaces. Full compatibility will likely be impossible, since 3G and earlier interfaces are not all compatible with one another. One stepping stone to backward compatibility would be requiring symbol rates to be simple multiples of some earlier systems’ symbol rates. However, backward compatibility in radio interface terms is not an obvious notion. It may refer to any of the following possibilities: ž Legacy devices still being used on a new air interface. ž The new air interface not interfering with legacy ones. ž The maximization of commonalities between legacy and new air interfaces in order to build dual-mode terminals with maximum HW/SW reuse (and minimization of cost and development time). The new air interface must also accommodate a much greater range of user applications, bit rates, and maximum packet loss and delay requirements, than do 3G systems. It must accomplish this in the same types of radio propagation environment as those encountered by 3G systems. It is expected to place very high spectral efficiency requirements on the new air interface – up to 10 bps/Hz. However, the attainable spectral efficiency experienced by individual users will depend on characteristics of their radio propagation environment, and on their wireless terminal capabilities and antenna configurations. ž Transmitted signal features: ž Frequency domain-based techniques, which provide excellent performance/complexity trade-off for expected large multipath spreads. Variants include: ž OFDM(A) (orthogonal frequency division multiplexing (access)), SC-FDE (serial modulation with frequency domain reception, with or without spreading), MC-CDMA (multicarrier code division multiple access). ž Different variants may be employed in different circumstances: e.g. in wide area cellular systems, OFDM(A) or MC-CDMA may be used in downlink, and SC-FDE in uplink for maximum efficiency of user terminal power amplifier. ž Advanced detection and error control techniques such as iterative (turbo) coding and processing, HARQ (hybrid ARQ), space-time coding and MIMO (multiple input multiple output). ž Adaptivity to traffic and channel variations, through: ž Code rate, modulation. ž Reconfigurable space-time processing. ž Spectrum, occupied bandwidth, power. ž User terminal capability. ž FFT (fast Fourier transform) size, training overhead. ž Scalable (with respect to data rate and cost and spectrum flexibility) air interface, supported by a reconfigurable signal processing architecture, for example.

Combined View of Future Systems

429

ž Multiple access features: ž Packet based, with packets partitioned and assigned to time division/frequency division/space division ‘chunks’. Assignment takes user terminal bandwidth, bit rate and signal processing capabilities into account. ž Multiple bandwidth frequency domain-based transmission for opportunistic cognitive radio. ž Broadcast and multicast must be implemented efficiently. ž High granularity, to support very low- and very high-data rate users at the same time (e.g. WSN and video streaming in one system). ž Low latency, to support real-time applications (e.g. VoIP). ž Efficient and flexible duplexing: ž Beyond the traditional paradigm where only pure FDD and TDD are evaluated, the aim is to bring alternative options into consideration. Hybrid schemes can be utilized that combine the advantages of both FDD and TDD and allow for flexible use of their features. ž Radio resource management based on cross-layer optimization: ž Fast/hybrid ARQ. ž Packet aggregation. ž PHY-aware scheduling and routing. ž Distributed scheduling with service differentiation. ž QoS-aware error control coding. 11.4.3.6 Relay-based Multihop Ubiquitous coverage is one of the prime user requirements of next-generation wireless systems. This requirement leads to the requirement for adequate received power in all locations, even at distances of a kilometre or more from the originating transmitter. This places a severe challenge on the link budget for bit rates of tens or hundreds of Mbps. In order for the transmitted power to be kept within reasonable limits, it is necessary to shorten the effective transmission distances by employing relay-based multihop transmission techniques. Relaying can increase coverage, reduce required terminal power and enhance capacity, but not necessarily all simultaneously. There are wide-area scenarios in which relaying can enable very high bit rate coverage, even near cell edge. Relay-based deployment modes include: ž Mobile relays, e.g. for sensor networks: ubiquity, but issues include privacy, power consumption, signaling overhead, uneven coverage. ž Or, alternatively, mobile infrastructure relays: controlled coverage, privacy and power consumption, but additional service provider cost. ž Digital, decoded relaying and analogue (amplify and forward) relaying approaches are available. Most work has so far concentrated on digital decoded relaying concepts. ž Smart antenna enhancements. ž Cooperative relaying (virtual antenna arrays): ž Exploitation of spatial diversity, without large numbers of antenna elements at user terminals. ž Introduces issue of channel resource sharing and additional interference. ž Routing issues: possible approach is overlay wireless network concept: cellular-based multihop (CBM).

430

Technologies for the Wireless Future – Volume 3

Deployment scenarios for relaying address: ž Mesh networking, which has been traditionally considered for short-range communications, can also be considered as part of a radio network deployment concept for infrastructure-based networks using fixed relay nodes. In order to react to topology changes, self-organization will play an important role and will make mesh networks robust against node failure, allowing for a high degree of resilience. Mesh networks must achieve synchronization over several hops without central coordination. ž Wireless backhaul as a means to enhance performance. This deployment option is particularly attractive, especially for those areas which are currently not covered with wireless infrastructure. The wireless option may be faster and more cost efficient than a wired backbone. Relays can be either user terminals themselves, or infrastructure devices deployed by the service provider. Physical and MAC layer design of the air interface must allow for the deployment of relays. If user terminals can act as relays for other terminals, power consumption and billing policies must be put in place to compensate the terminals’ users. The security aspects of such relaying terminals need to be strictly monitored and controlled. If infrastructure-based relays are used, the security and cost issues are simpler, but they still need to have reliable power sources. Furthermore, the problem of provisioning (numbers, locations and relay-to-access point links) is to be addressed. The system MAC layer design must include a provision for control and signaling among relays, access points and user terminals. 11.4.3.7 Channel Modeling and Propagation Next-generation systems will transmit over much wider bandwidths and in different (likely higher) frequency bands than those of 3G and earlier wireless systems. It is therefore very important to understand and accurately model the resulting types of radio channel, in order to design, analyze and simulate future systems. Since many future systems will be based on MIMO (multiple input multiple output) techniques, it is especially important to model the propagation phenomena among multiple transmitting and receiving antennas for the 100 MHz–1 GHz bandwidths and the frequency bands envisaged for 4G systems, for both short and long range. It is also critical to develop realistic spatiotemporal models for the characterization of the MIMO channel in a highly variable environment, and representative interference models for adequate evaluation of base-band signal processing techniques in multiple-antenna, multi-user, multi-service, multi-technology radio networks. MIMO channel modeling based on statistical formulation of the scatterers’ distribution should be supported by real-time MIMO measurement campaigns and link-level simulations for a variety of antenna array configurations, air interfaces, user mobility patterns and service profiles. Realistic interference modeling should be based on system-level simulations, taking into consideration the intra- and inter-cell impact of smart antenna techniques, nonuniformity of traffic (e.g. hot spots), mixed services scenarios and interoperability between different air interfaces. Realistic performance evaluations will rely on accurate channel and interference modeling and efficient/accurate interface between link- and system-level results.

Combined View of Future Systems

431

11.4.3.8 Smart Antennas – MIMO Smart antennas and MIMO transmission/reception configurations are key elements in satisfying the high spectral efficiency requirements for future wireless systems. MIMO transceivers designed to maximize spectral efficiency typically fall into three categories, corresponding to the maximization of diversity, data rate and signalto-interference-and-noise ratio (SINR): ž In the case of maximization of diversity, joint encoding – space-time coding – is applied and thereby the level of redundancy between transmit antennas is increased, as each antenna transmits a differently encoded fully redundant version of the same signal. ž The maximization of data rate is achieved by performing spatial multiplexing, i.e. by sending independent data streams over different transmit antennas. ž The maximization of SINR is achieved through focusing energy into the desired directions and minimizing energy towards all other directions. Beamforming allows spatial access to the radio channel by means of different approaches, e.g. based on directional parameters or by exploiting the second-order spatial statistics of the radio channel. Reconfigurable MIMO transceivers are adaptive techniques designed to adjust their structures and parameters to achieve optimal performance in a variety of scenarios associated with propagation, traffic, interference, mobility, antenna configurations, radio access technologies and channel state information (CSI) reliability. Multi-user MIMO processing has the potential to combine the high throughput achievable with MIMO processing with the benefits of space division multiple access (SDMA). It involves the ability to coordinate the transmission from all the antennas and exploit the CSI available at the transmitter to allow these users to share the same channel and mitigate or ideally completely eliminate multi-user interference by linear precoding or by the use of ‘dirty-paper’ codes. The successful integration of multiple-antenna techniques in future wireless systems will rely on: ž Advanced, reconfigurable MIMO processing algorithms with adaptivity to varying propagation conditions and robustness against network impairments. ž Innovative strategies for optimization of radio resources in a cross-layer fashion and interference management. ž Realistic performance evaluation of the proposed techniques based on accurate modeling, suitable performance metrics and simulation methodologies. Multiple antennas at a terminal or access point occupy a larger area and require multiple RF processing elements (amplifiers, mixers, etc.). Keeping cost, power consumption and physical size within reasonable limits will rely on continuing advances in DSP, RF devices and nano-technologies. Reconfigurability and adaptability are also important requirements for multiple-antenna systems; in response to the radio environment and current user needs, their main functions may have to shift between diversity protection against fading and system capacity enhancement. A transmitting or receiving antenna array may also have to adjust its operation to the capability and antenna configuration of the terminal at the other end of the link.

432

Technologies for the Wireless Future – Volume 3

Approaches to low-cost, simple implementation: ž Taking into account differing array implementation constraints at base stations and user terminals. ž Distributed (virtual) antenna arrays – cooperation among separate user terminals. ž Possible new implementation approaches include MEMS (micro-electro-mechanical systems), RF (radio frequency) combining, and efficient space-time-frequency DSP (digital signal processing) architectures. Combination of MIMO and OFDM The increasing demand for higher data rates translates into two main challenges for the physical layer development: to increase both spectral efficiency and bandwidth, and to simultaneously reduce the costs per bit. A combination of smart antennas (MIMO), the frequency domain-based physical layer (OFDM) and the exploitation of channel state information at the transmitter (TDD) appears to be suitable to meet these requirements. The combination of MIMO and OFDM allows for a substantial reduction in the complexity of the spatiotemporal processing since OFDM pre- and post-processing transforms the multipath channel into multiple flat fading channels, for which well-known MIMO equalization strategies can be used. 11.4.4 Sensor Networks A sensor network is a spatially distributed network consisting of a potentially large number of autonomous, heterogeneous sensor nodes that can be wirelessly connected to each other. Many sensor network topologies feature one or more base stations that receive reports from or query the sensor nodes, though fully decentralized sensor networks are also an important option. Unlike the nodes, base stations have in principle a considerable amount of energy available. In general, every node carries out a sensing function by which it monitors one or more physical, environmental or situational conditions around it. Wireless sensor networks were originally developed for military applications, but civil applications have rapidly emerged and practically taken over. Applications of wireless sensor networks already exist in a wide range of fields which rapidly grows and diversifies. 11.4.4.1 Applications The most common applications of wireless sensor networks include area monitoring (weather, seismic, agriculture, disaster, etc.), home automation and control, health care (patient monitoring, etc.), industrial and commercial (stock control, product manufacturing and management, logistics, etc.), military (control, surveillance, etc.) and others (traffic control, etc.). In a wireless environment, these can provide valuable information for ‘context awareness’. 11.4.4.2 Key Characteristics The most important characteristics of a wireless sensor network are a) large number of low-cost nodes that potentially interact, b) limited resources of nodes (processing power, data rate support, energy capacity, information storage), c) dynamic network topology (nodes move, join, disconnect), and d) prone to failure (nodes, links).

Combined View of Future Systems

433

11.4.4.3 Key and Emerging Technologies Many technologies are involved and combined in a wireless sensor network. At the physical layer, conventional and UWB-based radios are the basic radio technologies applied for implementing the air interface. In general but not necessary always, short-range communication links are assumed between the nodes of the network. As the physical area of the network may well reach beyond the communication range of the nodes, not all nodes are able to communicate with each other. Of utmost importance in wireless sensor networks is the fact that nodes have severe energy limitations and hence power-efficient or energy-aware techniques need to be developed. The choice of the medium access control (MAC) protocol heavily influences power efficiency and the maximum lifetime of the sensor network. As a large number of nodes participate, network topology is in principle more complex than that of ad hoc networks. Routing is carried out with the aims of minimizing the amount of energy required for packet delivery and maximizing the time until the energy resources of a node are depleted. Routing algorithms and protocols have received a lot of attention, though extensive research continues in this field. Network self-organization is one important technology needed to cope with the dynamics of the network topology. A wireless sensor network by nature supports cooperative interaction among the nodes. This cooperative effort is in fact essential to attaining a better utilization of the resources while providing connectivity. Positioning (location estimation) in wireless sensor networks is currently an important research topic, with promising results already available. The routing layer may use positioning information to select the best routes. 11.4.4.4 Challenges Sensor networks create a unique set of challenges compared to traditional networking technologies. The following challenges and promising technologies have been identified. Challenges for Sensor Nodes To achieve unattended multi-month and multi-year lifetimes of nodes, a combination of low-power design and efficient power management is required. Energy scavenging techniques play an important role in refilling the battery resources or reducing the amount of energy drained from the node’s battery. The long unattended lifetime and the envisaged applications of sensor networks in possibly safety-critical areas demand high robustness and reliability of the software and hardware. Lightweight network protocols, such as nano-IP, and ultra low-power MAC and PHY implementations are likewise important. The emergence of smart sensors that possess local intelligence helps to reduce the amount of data transmitted in the network as nodes become capable of preprocessing measurements locally. Last but not least, low-complexity designs that can be produced at very low cost have to be achieved. Challenges for Wireless Sensor Network Systems In the future, a massive deployment of RFID tags and networked sensors can be foreseen – they will begin to inhabit every physical object. Hence, protocols and algorithms that are scalable to networks of thousands and many millions of nodes will be designed. The dramatic increase in traffic volume caused by an ‘Internet of things’ has a significant impact on the architecture of the global IT infrastructure. A shift of processing functions, such as data aggregation and filtering, to the network edge will happen. Data gateways that provide the connection from the traditional Internet and telecommunication networks to the specialized sensor networks will be deployed.

434

Technologies for the Wireless Future – Volume 3

Challenges in the Physical Layer of Wireless Sensor Networks Energy spent on wireless communication can make up a major part of the total energy consumed by the sensor node. Therefore, it is crucial to reduce the power consumption of the wireless modem during data transmission and reception, but also when it is inactive. When aiming at minimizing the total energy consumed for data communication, it is not always beneficial to reduce the data rate as much as possible; instead an optimum data rate that reduces the transmission time and has a reasonably low power consumption must be found. Robustness against interferers and efficient spectrum management in a densely populated world of wireless devices will be incorporated in future wireless sensor networks. Challenges in the MAC of Wireless Sensor Networks Spectrum management and coexistence functions are going to be implemented in future sensor network MAC protocols. As it is the MAC protocol that determines the channel access regime and therefore decides when and for how long the radio transceiver – which contributes largely to the overall energy consumption – can be put in ultra low-power mode or send/receive mode, very power-efficient channel access mechanisms and power-saving modes will be designed. In a dynamic network environment, the MAC layer will provide mechanisms for dynamic topology management and mobility support, as well as self-organization and multihop communication. Especially for industrial applications, but not limited to this area, quality-of-service (QoS) guarantees become important, and new MAC protocols that offer this feature are needed. Challenges in the Networking Layer of Wireless Sensor Networks Energy-efficient routing techniques are and will continue to be one major research topic within wireless sensor networks, as the election of the propagation path through the network of nodes has a fundamental impact on the overall energy expenditure. Interaction and cooperation with other networks is also an area that deserves further exploration, as wireless sensor networks are usually studied in isolation from other wireless networks. In practice, wireless sensor networks will become an integral part of future highly heterogeneous wireless networks (4G), and hence investigating ways to dynamically interact and cooperate with other networks exploiting possible synergies between them is a challenging task. The problem of delivering information from one node to another, both part of a wireless sensor network, can be approached and modeled from the data information, node and location standpoints. The election of a particular approach and its associated model depends on the specific scenario and application, and the problem to solve in general. Deeper understanding of these approaches would allow the designer or wireless sensor networks to develop the most appropriate solutions for each particular type of problem. Challenges in Applications of Wireless Sensor Networks On the application layer, research work has been and continues to be focused on sensor data coding techniques, with the objective of reducing the amount of data to be transferred in the network. This can be achieved by data fusion algorithms, for instance aggregation, abstraction and filtering of measurements. A related field of continuous research is node addressing and scoping. In contrast to traditional networks, it becomes less important in sensor networks which particular node responds to a query so long as it belongs to a certain set of nodes, say

Combined View of Future Systems

435

all nodes located in the same room. Data security as well as resistance against tampering with nodes gains importance when sensor networks start becoming ubiquitous in the physical environment and companies rely on them in their production processes or value chains. Security algorithms and protocols are faced with the processing and energy limitations of nodes and have to provide secure communications in a dynamic, peer-to-peer environment. When sensor networks are used by service providers to offer services to their clients, reasonable solutions for accounting and billing of services must be introduced.

11.5 Development of Reconfigurability and Cognitive Wireless Networks 11.5.1 Introduction Future systems will be characterized by the convergence of mobile systems and IP networks, so as to jointly operate over a common access infrastructure, namely the B3G radio access infrastructure. In such a context, varying traffic demands impose reconsideration of the generic network management approaches. Major facilitators of such an approach are the cooperative networks concept, as well as the advent of adaptive (reconfigurable/cognitive) networks. The cooperative networks concept assumes that diverse technologies such as cellular 2G/2.5G/3G mobile networks and their evolutions (GSM/GPRS/UMTS/HSDPA), wireless local/metropolitan area networks (WLANs/WMANs), wireless personal area networks (WPANs) and short-range communications, as well as digital video/audio broadcasting (DVB/DAB) can be components of a heterogeneous wireless-access infrastructure and cooperate in an optimal way, in order to provide high-speed and reliable connectivity anywhere and anytime. The cooperation is materialized through the agreement to exchange traffic or sharing spectrum among the cooperative NPs and/or the joint configuration of network segments, providing assistance to each other in handling new traffic conditions or service management requests and maximizing the offered QoS levels. Yet, the realization of the B3G concept based only on network cooperation may not be viable or efficient. First, there could be objections to a business model that requires extensive inter-NP cooperation. Additionally, the cooperative networks concept implies that the whole set of the alternative RATs should be deployed (installed and configured) a priori in both network segments and terminals, which would require constant, potentially risky, investments in software and hardware whenever new technologies are introduced. Obviously, this would be inefficient, considering that all the technologies are not suitable for all the conditions. Adaptability (reconfiguration) is seen as a means to overcome the shortcomings identified above. Reconfigurable networks have the ability to dynamically adapt their behaviour (configuration) to the various conditions (e.g. hot-spot situations, traffic demand alterations, etc.) at different time zones and spatial regions, by exploiting deployments with many fewer pre-installed components. This process, in general, imposes (re)configuration actions which may affect all layers of the protocol stack. Such actions indicatively include RAT selection, spectrum allocation, algorithms selection and parameter configuration (at the PHY/MAC layer), TCP adaptation, IP QoS configuration, etc. (at the network/transport layer) or adaptation to appropriate QoS levels (at the middleware/application layer). The advent of this new era implies benefits for the involved business-level entities.

436

Technologies for the Wireless Future – Volume 3

11.5.1.1 Users Users will be provided seamless and ubiquitous connectivity, through simple and mostly invisible service provision and adaptation. Roaming capabilities will also be extended, through the dynamic adaptation to regional contexts. Similarly, the ability to download and dynamically install software on terminals will lead to enhanced personalization features and more advanced services. Users are offered the ‘best-suited’ air interface for each service, as well as its associated nonfunctional aspects, including quality of service (QoS) and cost. 11.5.1.2 Network Operators (NOs) NOs will have more options for providing the required QoS and capacity levels. This will be possible through reconfiguration of their own infrastructure, enabling them to introduce value-added services more easily. Such reconfiguration will decrease CAPEX with reasonable OPEX, as the same equipment will be able to operate multiple air interfaces to supply a variety of services, thus decreasing the need to deploy parallel multiple systems. In the long run, reconfigurable equipment will simplify the use of additional frequency bands. 11.5.1.3 Manufacturers Reconfigurable technology will increase the flexibility, since it introduces the option to dynamically upgrade equipment, thus accelerating the time to market. Moreover, it allows simplified product evolution, reduced development cost, economy of scale and increased reliability. 11.5.1.4 Service Providers Such technology can also be exploited at the application level; it extends the current ability to download and install applications in user terminals. So they will have the possibility to introduce and deploy new services, without the need to provide separate service implementations at design time. 11.5.2 Advent of Cognitive Networks Two prospective contributors towards this convergence are the concept of cooperation among networks and the advent of adaptive systems (which might be merely reconfigurable, or enhanced by cognition capabilities). The cooperation among previously competing networks assumes that diverse technologies such as cellular 2G/2.5G/3G mobile networks (GSM/GPRS/UMTS), BRAN/WLAN and DVB can be components of a heterogeneous wireless-access infrastructure and cooperate in an optimal way, in order to provide high-speed and reliable connectivity anywhere and anytime. In this context, traditional competition among NOs is replaced by NPs’ cooperation towards the common purpose of ubiquitous and qualitative service provision. The cooperation is materialized through the agreement to exchange traffic between one network and another, providing assistance in handling new traffic conditions or service management requests and maximizing the offered QoS levels. Thus, cooperative NOs should properly manage their resources, so as to have capacity available, which is necessary for cooperating with other

Combined View of Future Systems

437

networks in the global infrastructure. Advanced management functionality is required for supporting the cooperative networks concept. Relevant research attempts have been made in the recent past, concentrating in particular on the allocation of traffic to the different RATs and networks, as well as on the allocation of users to QoS levels. The cooperation and complementary use of the numerous available radio access technology (RAT) standards is a basic feature of the so-called Beyond the 3rd Generation (B3G) vision [1,2]. The target of this vision is the provision and activation of revolutionary services, at extremely high bit rates and in a cost-efficient manner. It is believed that the fulfilment of these requirements through one RAT will require increased capital expenditure (CAPEX), as it does not exploit the versatile capabilities offered by the complementary use of different RATs. In general, cooperative networks assume that an NO can select the best among a set of alternative networks in order to offer the best possible services to its customers. However simple this may sound, the overall goal of providing seamless mobility and connectivity may still be difficult to achieve. The main reason is that mere cooperation, as described, implies that the entire set of RATs should be a priori deployed by the NO. This is not the most efficient way to reduce CAPEX. In support of the above, reconfigurable networks [2,3] are seen as a major facilitator of the B3G vision. Cognitive networks are capable of continuously adapting to changing environmental conditions and/or user needs. Adaptation is mainly realized by means of self-management, or in other words in accordance with autonomic computing principles [4], and typically involves machine learning. Reconfiguration of the system’s own infrastructure may affect more than just the traditional networking layers of the protocol stack, i.e. the middleware, presentation and application layers, in addition to the PHY (physical), MAC (medium access control), LLC (logical link control), network and transport layers. For example, reconfiguration in the PHY/MAC layers includes the selection of the most appropriate RAT(s) and spectrum for operation, whereas reconfiguration in the middleware and application layers includes the selection of the most appropriate policies to control network element and technology reconfiguration. It is envisaged that these reconfiguration types will enable NOs to further reduce CAPEX and the operational costs (OPEX). Moving one step further, if reconfigurations are enhanced and supported by a means to store information from previous interactions with the environment and perform future adaptations based on this information, then one can talk about ‘cognition’. Cognitive networks dispose mechanisms that retain data from previous interactions with the environment and plan their actions accordingly. Moreover, cognitive networks do not presume the fixed deployment of technologies in terminals and network segments; rather, they have embedded intelligence that enables them to learn, and, based on this learning, adapt their functionality according to external stimuli. This is depicted in Figure 11.3, where each segment is made up of cognitive elements. Each element and terminal is reconfigurable (can operate with alternative configurations) and has the intelligence to select the best configuration, in order to adapt to environmental conditions. In this context, reconfiguration at the PHY/MAC layers provides the ability to dynamically select the set of the most appropriate RATs and spectrum, in order to better handle business-, service-, resource-, location- and/or time-variant requirements. The alternative configurations that should be utilized are known by the cognitive elements, enabling context-aware selection. Configurations change in time and space. Reconfigurations are software-defined. Therefore, a reconfiguration is carried out by activating the appropriate software, which implements the selected RAT.

438

Technologies for the Wireless Future – Volume 3

Cognitive Network State − n + 1

Cognitive Network State - n B3G Infrastructure

B3G Infrastructure

Services and Applications

Services and Applications

IP core network

IP core network

Cognitive Network Segment

Cognitive Network Segment

Cognitive Element

Cognitive Element

RATj, fj

RATj, fj RATi, fk

New Service Area and Demand Requirements

RATj, fm

Cognitive Network Segment

Cognitive Network Segment

Cognitive Element

Cognitive Element

RATj, fj

RATj, fk RATk, fn

RATi, fm

Figure 11.3 Example of cognitive wireless network. Elements may change RAT, frequency, or both, when new conditions are identified

11.5.3 Management Architecture and Operation 11.5.3.1 Design Choices Since a cognitive network consists of numerous elements and terminals of highly heterogeneous natures, located in different places, a centralized management approach becomes prohibitively complex and inappropriate. Hence, distributed management approaches, relying on pertinent technologies, e.g. autonomic computing, are currently being focused upon. This approach can offer scalability and modularity (which provides low complexity). 11.5.3.2 Components and Functionality Figure 11.4 depicts the overall management architecture of a B3G infrastructure. Its entities are organized in a hierarchical manner that consists of three tiers. At tier 1, each entity controls a whole network segment (subset of the network). Tier 1 is made up of mechanisms whose primary purpose is to coordinate with the backbone network, as well as the decisions of tier 2 management entities. The entities in the second tier manage a particular reconfigurable network element (access point). The entities in the third tier are targeted at terminals. Tier 2 and tier 3 entities have the internal structure shown in Figure 11.4. The ‘Monitoring, Discovery’ component is set to continuously sense the environment, so as to monitor the demand for and discover the capabilities of alternative configurations. Network elements (and also terminals) should be constantly monitoring the environment. This procedure is foreseen as an essential entity for a networkcentric as well as for a terminalcentric scenario, where the need for reconfiguration is imposed. This need would become apparent when a terminal ‘came to realize’ (either individually or through its communication

Combined View of Future Systems

439

B3G Infrastructure Services and Applications

Tier 1

IP core network

Segment Management Functionality

Cognitive Network Segment

Cognitive Network Segment

Segment Management Functionality

Tier 2 Element Management Functionality

Element Management Functionality

Reconfigurable element

Cognitive Element

Reconfigurable element

Cognitive Element

Tier 3 Terminal Management Functionality

Terminal Management Functionality

(a)

Cooperation with other elements/NOs

Profiles, Policies, Goals

Negotiation, Selection and Reconfiguration Implementation

Monitoring, Discovery

Learning Capabilities (b)

Figure 11.4 (a) Overall management architecture; (b) management functionality for individual element/reconfigurable terminal

with the network) that its operation parameters could be better if it used an alternative RAT. So the purpose of the ‘Monitoring, Discovery’ procedure is the identification (within the monitoring range of the terminal) of an alternative RAT (other than the one operating so far), with better offers, in terms of better circumstances of coverage, QoS, etc.

440

Technologies for the Wireless Future – Volume 3

Therefore, a functional entity that implements this procedure should be included. This entity should be capable of interacting with entities external to the whole management module, such as the network support functions, through the appropriate interfaces. Furthermore, it should also be capable of communicating and exchanging monitoring with the rest of the internal entities. The ‘Cooperation with other elements/NOs’ component can communicate with other elements/NOs so as to acquire their requests, offers, etc. The ‘Profiles, Policies, Goals’ component provides user, application and element requirements and characteristics, as well as policies and business goals of the NO. The ‘Negotiation, Selection and Reconfiguration Implementation’ component decides upon and implements the changes to be made on the reconfigurable element, based on policies, profiles and the integrated learning capabilities. In particular, after the discovery of alternative RAT choices, network elements and terminals should be capable of interacting with these RATs, in order to negotiate in terms of quality and cost factors. The negotiation phase is of course followed by the selection phase, during which the terminal selects the best reconfiguration pattern. The above necessitate the existence of an associated functional entity, in order to realize this procedure. This entity should interact with the network support functions in order to exchange the necessary negotiation information, as well as with the rest of the entities internal to the equipment. Other issues related to the download and installation that might be important during the selection of an alternative RAT for operation include: ž Cost commensurate with function: ž Scalability: peak and average power, bandwidth, bit rate, hardware requirements. ž Hardware impairments and/or limitations: power amplification, phase noise, power consumption, heat dissipation. ž Performance metrics. ž Other wireless media: satellites, high-altitude platforms. ž Some open issues, for example: ž Backward compatibility with 3G. ž Future-proofing air interfaces. ž Evolution path. Cognitive radio architectural approaches should provide for the maximum possible individual as well as group operation, so as to decrease the system’s complexity and support its scalability. As already introduced, it is anticipated that such a fully-distributed approach can be provided by the use of autonomic computing. Autonomic computing is derived from the human nervous system – just as the human nervous system performs involuntary actions (such as pumping blood) to free the human brain to address other tasks, autonomic computing systems perform tasks that previously required intensive manual operation (such as (re)configuring a device) to enable the autonomic system to perform more strategic tasks (such as optimization and planning). Local optimization is achieved by sending to entities the appropriate policies, which direct self-management towards a global network operational goal. Ideally, the distributed decision entities have full knowledge of the context, thanks to cognitive support functionality.

Combined View of Future Systems

441

11.5.4 Management of Network Segments/Elements NOs have traditionally designed and deployed the radio access networks to cover the traffic demand of the planned services in a static approach, considering the busy hour traffic in each geographical zone. Future wireless systems will, though, continuously experience transformations, according to time and space variant demand. Therefore, no exact separation between planning and management will exist, but a combined scheme will have to be applied to future contexts. Consequently, novel network planning mechanisms have to be established. Such methods can initially serve as a proof of concept for the aforementioned benefits, but can also be utilized in real-time adaptive networks. In future systems, an NO can be access provider, backbone provider or service provider. For instance, an IP-based radio access network is large (several thousand or more routers), geographically widespread, centrally managed (from at most a few management centres), and management is often in-band (no dedicated management network) to reduce cost. Because of the lack of homogeneity, the current network management is complex and tedious. Future network management architecture shall include dynamic mechanisms and entities that allow network operators to configure, monitor and manage the operation of individual entities as well as the entire system. Some of these functions shall be automated and potentially distributed (e.g. autoconfiguration, path restoration); others may be made available from appropriate management tools and interfaces. It is probably too difficult to have one network management solution that fits every network and every operator. But there is a great need to reduce the complexity and to get more standards into the area. Much of the management that is done manually today should be automated in the future. 11.5.4.1 End-to-end Management and Control – Functional Architecture Adaptive (cognitive/reconfigurable) systems are aware of their environment as well as their own internal structure. The cognitive radio is, for example, aware of the number of multipaths it sees and can adjust the equalization algorithm accordingly. It can even take hints from the information being exchanged by the user to adapt its detecting strategy and power consumption according to the user’s behaviour. Such an intelligent radio system has two primary objectives: 1) highly reliable communication whenever and wherever needed, and 2) efficient utilization of the radio spectrum. Translating the vision of cognitive radio into reality is done by investigating and introducing concepts in advanced spectrum management (ASM), advanced radio resource management (ARRM) and dynamic network planning and management (DNPM). The ASM will optimize the spectrum allocation adaptively. This includes the optimization of guard bands between the radio access technologies (RATs). The ARRM should handle the optimization of traffic through the available RATs. One of the main concerns of ARRM is the vertical handover between RATs. The DNPM algorithms deal with the dynamic radio cell behaviour through power allocation and antenna techniques. The ASM, ARRM and DNPM will take the evolution of mobile communication systems one step further towards cognitive radio. The functionalities of DNPM, ASM and ARRM are closely interlocked and coupled (see Figure 11.5). Nevertheless, the interworking of these three concepts can be considered as three interlocked loops. Each loop reacts based on the output parameters of the adjacent ones. The further inside a loop is located, the faster is its reaction time. Therefore the entities of the middle and inner loops should be locally decentralized in order to combat delay through the

442

Technologies for the Wireless Future – Volume 3

DNPM GSAM LSEM

Inner-loop

IORM Traffic Estimator

IOEM

Operator #j IORM

IOEM

DNPM GSAM LSEM

LSAM

LSAM

ARRM /JRRM

ARRM /JRRM

Iom: interfaceforoperator/ metaoperator

RRM-terminal

Traffic Estimator

Operator #i

Ioo. interface for operators

Meta Operator

Int: interface for network and terminal

ASM Agent

Middle-loop

Reconfi. Agent

Outer-loop

Figure 11.5 Overview of functional blocks

route to a central entity. The function of the outer loop can be executed in a central entity at a central place, e.g. for GSM in the core network. Within the outer loop the network will be planned and the DNPM will give recommendations to the operator about the needed spectrum in time and space. The operator’s entity inter-operator economic manager (IOEM) will decide how to trade spectrum based on the advice of the inter-operator resource management (IORM). The IOEM can offer demand for spectrum depending on expected traffic. The DNPM plans the network based on the spectrum trading results, and the global spectrum allocation management (GSAM) computes the best opportunity for spectrum division to the operator’s RATs. This happens over the long term, based on the reaction of the middle loop. DNPM is specified for the O&M system irrespective of its two phases, i.e. the planning phase and the management phase. Referring to the 3GPP, documents on inter-operator telecommunication management network interfaces and functions for registering, monitoring and controlling the operating frequencies are included in the O&M subsystem. The interface between operators, i.e. the interface supporting inter-PLMN/inter-organization operations, supports the inter-PLMN mobile service provisioning, e.g. for the roaming users, and can be extended as spectrum trading. High-level policy agreement, e.g. access to meta-operator, and certain levels of service agreement between operators can be transferred through this interface. On the other hand, some fast interactions concerning the spectrum reallocation might be directly dedicated to the control plane of the radio subsystem. Therefore, in Figure 11.6, DNPM is conceptually allowed to partially cover the functions of IORM, IOEM and GSAM.

Combined View of Future Systems

443

Context (Traffic and mobility information, system aspects)

Policies (Nos policies, agreements)

Optimization Functionality (Algorithms, Objective Function)

Reconfiguration (transceivers’ configuration, traffic distribution, quality levels assignment)

Profiles (user classes, terminals, applications, elements capabilities)

Figure 11.6 Dynamic network planning and management – problem description

In the middle loop a local spectrum economic manager (LSEM) trades the spectrum of each base station to the users. Based on the trading results, the local spectrum allocation management (LSAM) assigns the RATs operated/used by each spectrum user the gained radio resources as a number of generic resource elementary credits (GRECs), which are the elementary resource units offered per RAT. The ARRM reacts fastest and therefore represents the inner loop. Its task is to trigger and manage the vertical handover and optimize spectrum usage, using traffic splitting over different RATs. If a user does not need the whole spectrum gained by negotiation thanks to ARRM, the unused spectrum can be reused for other users. In this case the ARRM triggers the LSAM in the middle loop to rearrange the spectrum. The general idea in a reconfigurable infrastructure is that many RATs are available and candidate for operating in each network element, and one of them can be selected each time for operation, as part of the overall reconfiguration decision. That is, only the appropriate technologies are selected, activated and used, based on context. In such a flexible environment, the decision and final selection of the optimum reconfigurations per network element appear to have significantly high importance. In legacy design systems, a classical problem is the following: ‘Given the traffic that must be served by specific networks/RATs, find the best configuration in terms of resource assignment (spectrum, power etc.).’ A variation of this well-known problem is the following: ‘Given the traffic that must be served by specific networks/RATs, check whether a specific configuration may be acceptable or not.’ However, in the case of reconfigurable networks, a preliminary step needs to be made before handling the previously-stated, well-defined problems. More specifically, the problem that must be solved first is the following: ‘Given the traffic to be served, find the best partitioning to the various candidate RATs.’ The input to the dynamic network planning and management problem is classified into three main categories: i) context; ii) profiles; iii) policies. Context information specifies the demand per user class and service, as well as the capabilities of alternative configurations currently not in use by the element. Profiles provide information on user classes, terminals and applications. User class profiles describe the permissible QoS levels per application, the importance or ‘utility’ of each QoS level for the user class, and the maximum tolerable cost. Terminal and element profiles provide

444

Technologies for the Wireless Future – Volume 3

the capabilities in terms of permissible configurations. Application profiles also provide the permissible QoS levels. Policies are related to the objectives of the NO and the strategies to achieve them. The offers of cooperating networks can provide the cost per QoS level, and the traffic volume associated with affiliated NOs. The output of the problem is classified into the following categories: i) allocation of RATs and spectrum to transceivers of the segment; ii) allocation of QoS levels to the users; iii) allocation of demand to RATs. Moreover, separate solutions can cater for the efficient management of resources when the allocation of RATs and spectrum to transceivers comprises 3G technologies, as well as when WMAN are selected for operation in the segment. Currently, we are in a position to find the best configuration based on monitoring (demand), discovery, profiles and policy information, for individual elements or network segments. Main extension will be the tight integration with learning functionality. This will enable the fast (automatic) identification of the best configurations, based on experience and the current situation. 11.5.4.2 Cognitive Pilot Channel (Cognition Enabling Radio Channel) in a Multi-RAT Environment In the context of a heterogeneous radio network environment, it is necessary for reconfigurable radio terminals to be able to initiate a new user session, in order to be connected to the most suitable access point of the most appropriate radio access technology (RAT). In particular, after ‘power on’ the mobile does not know which RAT may be the most appropriate or in which frequency bands potential RAT(s) are operating. This last point will be even more critical in the long term, when new regulatory approaches to spectrum usage will allow the implementation of dynamic spectrum allocation (DSA) and flexible spectrum management (FSM) (which includes spectrum pooling). In this case, the mobile terminal will have to initiate a communication in a spectrum context which is completely unknown due to dynamic reallocation mechanisms. Without any information about the location of RATs within the frequency range reachable from the mobile terminal (e.g. 500 MHz–>6 GHz), it is necessary to scan the whole frequency range in order to discover the spectrum constellation. In this context, a cognitive pilot channel (CPC) should provide relevant information (such as frequency bands, available RATs, services, load situation, etc.) to a mobile terminal so that it can initiate a communication session in an optimal way regarding time, situation and location. This would allow a number of meaningful advantages from several stakeholder viewpoints: ž It would simplify the selection procedure, avoiding a large band scanning. ž The gain for the user would be lower battery consumption. ž It would be an appropriate solution for the implementation of DSA/FSM, hence the advantages for operators and spectrum management regulators, in a dynamically changing radio environment. The selection procedure using the CPC would consist of the following steps: ž At ‘switch on’, the mobile listens first to the out-band CPC. ž Getting the list of all existing operators and preferred RATs, the mobile selects the most suitable one to camp on.

Combined View of Future Systems

445

The information update can be dynamically processed in line with each operator strategy. 11.5.5 Terminal Management The idea of diverse RATs cooperating over a common access infrastructure imposes novel management mechanisms for agile equipment. The proposed framework should consist of the following two main modules, responsible for its basic functionality: ž A functional entity within the equipment (configuration management module – CMM) (terminal, base station/access point or network) that manages the reconfiguration processes according to specified semantic protocols and configuration data models (which may be stored in a distributed configuration database system). From the equipment perspective, the various CMMs also interact among themselves, as well as with supporting equipment entities within the network, through an external (transparent) interface. ž One additional entity, the configuration control module (CCM), which is a supporting entity responsible for the control and supervision of the reconfiguration execution. This is done using specific commands/triggers and functions of a given layer or a given execution environment. Three main layers are considered here: application, protocol stack (L2–L4) and modem (L1). Other entities closely related to equipment management are the execution environment and the reconfigurable protocol stack framework (or reconfigurable functional layers). The execution environment is the means for providing the basic mechanisms required for dynamic reliable and secure change of equipment operation. The execution environment aims to offer a consistent interface to the equipment reconfiguration manager in order to apply the needed reconfiguration actions. For reconfigurable equipment, reconfigurable components need to be used. Such components are programmable processors, reconfigurable logic and parameterized ASICs (offering software control on their parameters). The execution environment sits on top on this hardware platform and offers basic mechanisms enabling the exploitation of the reconfigurable hardware components. The reconfigurable protocol stack framework is an open protocol stack framework which can be used to support several RATs with diversified protocols and protocol functions. This implies an architecture that supports dynamic insertion and configuration of different protocol modules in a common manner, taking into account the resources and capabilities of the target devices. The functional entities of the equipment management architecture and other internal and external entities will be described in detail in the following paragraphs. 11.5.6 Reconfigurable Equipment (HW Aspects) 11.5.6.1 Reconfigurable Elements Reconfigurable elements [5] (terminals or network elements) can operate with diverse alternative configurations, especially RAT and spectrum at the PHY/MAC layers. Reconfigurable elements are a prerequisite for cognitive wireless networks. Figure 11.7 depicts the trends regarding the evolution of reconfigurable elements. According to Figure 11.7(a), a fully reconfigurable element should consist of a fully flexible radio frequency (RF) front end, capable of operating at various frequencies, as well as a very power-efficient base-band (BB) signal processing unit. This is the most challenging approach.

446

Technologies for the Wireless Future – Volume 3

Programmable highpower BB signal processing

Fully Flexible and Adaptive Front-End

(a)

Memory for reconfigurations

RF

RF

RF BB Signal Processing

RF

(b)

RF

mode 1

BB Signal Processing

RF

mode 2

BB Signal Processing

RF

mode 3

BB Signal Processing

RF

mode 4

BB Signal Processing

(c)

Figure 11.7 Reconfigurable elements approaches: (a) fully reconfigurable equipment; (b) multimode equipment – software-defined signal processing; (c) multimode equipment

Accordingly, Figure 11.7(b) is the second most innovative approach. Here, the RF entity of the element is not fully flexible. This is compensated by the ability to choose among a number of potential modes. However, the BB signal processing is fully flexible, requiring memory for storing the algorithms and parameters of potential configurations. Finally, Figure 11.7(c)

Combined View of Future Systems

447

represents the case where an element is just multimode, consisting of multiple parallel RF and BB chains. The choice of a configuration activates the appropriate chain. In general, the potentials that come out of full reconfiguration constitute indisputably the most preferable choice, allowing all elements to be flexible, in the sense that they can operate with most current (or future) RATs. However, this approach is associated with difficulties, such as i) the realization of an RF module capable of operating in a wide range of frequencies, ii) increased power consumption. 11.5.6.2 Tolerance to HW Imperfections It is advisable to push most of the dynamic range and selectivity requirements into the digital domain. This will boost the reconfiguration capabilities of the whole RF part and will result in a future-proof architecture, since semiconductor technology developments in the field of digital signal processing normally outperform the development speed of purely analogue-driven technologies. Furthermore, making use of digital signal processing capabilities in an advanced CMOS technology has advantages over a ‘pure analogue’ RF in terms of stability, reconfigurability, power consumption and cost. The most relevant topics to be considered are: ž lower supply voltages, rising carrier frequencies, lower cost. ž ‘dirty RF’ becomes an issue: ž phase noise ž nonlinear HPA ž I/Q imbalance ž ambiguity/jitter ž use of digital compensation techniques. ž micro-electronics technological roadmaps: ž take into account deep sub-micron effects in system studies. Hardware imperfections include power amplifier nonlinearity, phase noise and frequency offset, DC offset and I/Q imbalance. Judicious design choices must be made between compensating these kinds of imperfection (e.g. via adaptive linearization of the power amplifier, use of digital mixers, etc.) and making air interfaces robust to them (e.g. by choosing modulation schemes with low peak-to-average power ratio).

11.6 Other End-to-end Aspects 11.6.1 Self-organization Wireless technologies continue to penetrate rapidly, connecting not only mobile phones and computers, but myriads of small devices, sensors and everyday items. This trend creates new applications but, in turn, adds to the spatial-temporal complexity and dynamics of the network. It thus increases the burden on network administrators and users. An important question is how this complexity can be reasonably managed without requiring users to become technical experts in the field, and the network owners to spend time and resources in managing networks. One promising approach is to increase the degree of self-organization, i.e. to design and develop networks that minimize human intervention and organize themselves as far as possible.

448

Technologies for the Wireless Future – Volume 3

We can sort the various network scenarios into two categories: Internetcentric and ad hoc & sensor networks. Today’s communication and computer networks already have several functions that contribute to a higher level of self-organization. In particular, the Internet and its applications give us many examples where user-friendly self-configuration and distributed operation can be used to reduce operational costs, enhance usability and create completely new services. It is strongly recommended that IPv6 be required, since it offers several mechanisms for autoconfiguration of hosts; in particular, these mechanisms allow hosts to obtain an IP address and other information that enables them to access the network and search for more advanced and more specific configuration data. In addition, router autoconfiguration, ZeroConf and the Rendezvous protocol, if implemented correctly, can also enable a great degree of self-organization. Peer-to-peer networking is another scenario and communication service in which each participant has the same capabilities and every member can initiate a connection to another member. Figure 11.8 provides a summary of the characteristics of different kinds of peer-to-peer networks and compares them to classical client-server networks. It is evident that some P2P networks are more amenable to self-organization than others. For example, the client-server and centralized P2P architectures do not lend themselves to self-configuration. It is recommended that pure P2P, hybrid P2P or structured P2P models be adopted to have a reasonable degree of self-organization, although they come with their own strengths and weaknesses. It is quite common to associate self-organization with dynamic topology control. The more distributed and dynamic the topology control model is, the more it is self-organized; therefore, this is a desirable requirement to have on any network design project. However, the sensitivity to self-organization must be carefully tuned so as not the render the network unstable, resulting in loss of data. In ad hoc and sensor networks, topology control can Client-Server Server is the central entity and single provider of content and service. → Network managed by the server Server as the higher performance system Client as the lower performance system Example: WWW

Peer-to-peer

Self organization / Shared resources / Servent concept Unstructured P2P

Structured P2P

Overlay structure is not determined by the protocol

Any peer can be removed without loss of functionality → No central entities P2P network structure is determined by the protocol

Centralized P2P

Pure P2P

Hybrid P2P

Central entity is necessary to provide the service

Any peer can be removed without loss of functionality → No central entities

Any peer can be removed without loss of functionality → dynamic central entities (elected by criteria like bandwidth etc.)

→ Single point of failure Central entity is some kind of index database Example: Napster

Self-organization achieved by flooding → High signaling traffic Examples: Freenet, Gnutella 0.4

Examples: JXTA, Gnutella 0.6

Figure 11.8 Peer-to-peer networking

Deterministic connections in the overlay network Examples: Chord, CAN

Combined View of Future Systems

449

Topology Control

Transmit Power

Optimized Topology (energy and/or capacity)

Figure 11.9 Topology control as a self-organizing paradigm

Locationbased Directionbased Neighbourbased

Local info

Worst-case conn

1-hop 1-hop 1-hop

yes yes no

Additional hardware GPS dir. antenna no

Optimize energy

Optimize capacity

yes yes yes

no no yes

Figure 11.10 Main features of the various approaches to topology control

be regarded as the art of coordinating the transmit powers of the nodes such that certain network-wide properties (e.g. maintaining connectivity, reducing node energy consumption) are achieved. Given the fully distributed nature of ad hoc networks, topology control should be implemented without the intervention of any centralized infrastructure. Figure 11.9 shows the self-organizing nature of topology control: it is a mechanism which, by acting on the transmit power levels (local choice at the nodes), returns a network with certain desired topology features (global property). Out of the three main current approaches (location-based, direction-based, and neighbour-based), neighbour-based approaches are the best suited for mobile wireless networks, followed by direction-based, and lastly location-based. Therefore, the recommendations follow in the same order for the concept and the system requirements development. See Figure 11.10 for a comparison of the various topology control approaches. The automatic configuration of addresses to nodes in ad hoc networks needs additional functionality compared to the approaches used in traditional networks. Although this work is still in the standardization phase, the goal is to standardize solutions for IPv4 and IPv6 address autoconfiguration in two scopes: a local scope (i.e. addresses that are only valid in a particular ad hoc network) and global scope (i.e. addresses that are routable in the global Internet). Therefore, this is another requirement on the future wireless world system. From the perspective of self-organization in network management, distributed approaches are the most desirable, and they can fall in any of the following three approaches: policy-based, pattern-based or knowledge plane-based. Although quite some research is needed, it is recommendable to adopt the knowledge plane-based approach since it offers the most autonomous and robust self-organization. Finally, there are a number of graph theoretic and biologically-driven approaches currently in the study phase, but they are not sufficiently developed to qualify for inclusion in the WWRF system concept and requirements.

11.7 Summary and Conclusion In the last two decades, cellular mobile communication has experienced an unprecedented market success. This is clearly shown by the fact that the mobile industry reached the 3 billion customer milestone in 2007.

450

Technologies for the Wireless Future – Volume 3

The WWRF has developed the vision of: ‘7 trillion wireless devices serving 7 billion people by 20170 [6]. According to this vision, the 3 billion is just one point on the way to serving all the people on the planet. To realize this vision, a wide range of technologies starting from the physical layer up to the service platform is required. Also, the simplest forms of wireless devices have to be affordable and easily operable. The goal is to specify an environment to realize mobile communication services that follow the vision of the future wireless world [7]: Users are able to access, at anytime and anywhere, services that best tailor their preferences and environment. Context-aware applications aim to provide relevant information of the users to provide services and applications that best suit their preferences and environment.

That is, the mobile terminal becomes the interface between the user and the ‘digital world’, offering information and communication services far beyond services like voice, SMS, email or Internet connectivity. Important market trends and developments the industry is facing include [8]: ž ž ž ž

Significant growth of mobile customers, as mentioned above. Broadband everywhere. Ever-increasing spread and importance of the Internet and its services. Convergence of digital industries, like FMC (fixed mobile convergence) and digital music, digital cameras (cameras in handhelds). ž Deregulation and globalization. ž Services and applications are the key. Key principles in the vision of the WWRF are: ž Users and their requirements are the starting point [9]. This so-called I-centric approach and the overall reference cake model [10] is shown in Figure 11.11. ž Users are in control through intuitive interactions with applications, services and devices where services and applications are personalized, ambient-aware, and adaptive (individual-centric, I-centric) – ubiquitous from the point of view of the user. ž Seamless services to users, groups of users, communities and machines (autonomously communicating devices), irrespective of location and network connection, with agreed quality of service. ž Users, application developers, service and content providers, network operators and manufacturers can efficiently and flexibly create new services and business models based on the component-based open architecture of the wireless world. As can already be seen today, the wireless networks are characterized by quite some heterogeneity of networks, combined with a trend towards always-higher bit rates: ž For the cellular networks, starting from GSM, B3G, HSPA up to the latest requirements of NGMN [11], bit rates of more than 50 Mbps for the uplink and more than 100 Mbps in the downlink are planned. In local hot spots and with low mobility, peak bit rates could be up to 1 Gbps [7]. The requirement for the spectral efficiency is: up to 10 bps/Hz. Another requirement is low delay for services.

Combined View of Future Systems

451

Personalisation

Adaptation Conflict Resolution

Service Deployment

Environment Monitoring

Service Creation

Service Discovery

Service Control

Service Bundling

Business Model

Ambient Awareness

Communication Space (Contexts & Objects) Service Semantic

Appl. Scenarios

User Model &

Generic Service Elements for all layers

Application Support Layer

Service Platform

Service Execution Layer Service Support Layer

IP based Communication Subsystem

Network Control & Management Layer IP Transport Layer Networks

Wired or wireless Networks

Terminals

Devices and Communication End Systems

Figure 11.11 Reference I-cake [10]

ž The different standards of WLAN for more localized hot spots, with bit rates of 1 Gbps and more. An ultra high-speed version could achieve 10 Gbps and more [12]. ž Short-range communication over distances of metres, where current standards are e.g. Bluetooth, NFC, with technologies such as UWB, wibree, or infrared. ž Wireless body area networks. ž Sensor networks, especially for acquisition of context data and area monitoring. ž Broadcast networks such as DVB-H and DVB-T. ž Main examples of the technologies, most of them still requiring quite some R&D effort, are smart antennas, UWB, variants of OFDM, advanced detection and error control techniques such as hybrid ARQ or MIMO, optical technologies, and relay-based multihop. So far, the WWRF vision still assumes that the future networking technology will be IP based. Nevertheless, further requirements, which should be addressed in an evolved IP network, include: ž ž ž ž

support of end user services network control and management data transport mobility support.

A main challenge that the communication research faces is the realization of the concept of ambient networking. This vision will allow for cooperation of networks on demand, transparently and with autoconfiguration, i.e. a federation of multiple networks of different operators and/or operators. Another key technology is cognitive radio. This is seen as a central enabler of future networks, because it allows for continuous adaptation to changing environmental conditions

452

Technologies for the Wireless Future – Volume 3

and/or user needs. This is achieved via mechanisms of autonomic computing, involving machine learning, and by adapting the middleware, the physical and other layers, and parameters of the protocol stack. Adaptation is mainly realized by means of self-organization, or in accordance with autonomic computing principles. Last but not least, all adaptations to the ever-changing environment requisitions should be performed transparently and securely, in order to guarantee the maximum possible levels of security, privacy and trust. This will help avoid numerous unwanted situations, so compliance with the constraints that will be imposed by security aspects should be considered indispensable.

11.8 Acknowledgements This document is based on the WWRF Vision Committee document ‘Visions of System Concepts for Wireless World’. That document was co-edited by Mikko A. Uusitalo (Nokia Research Center, Finland), Jean-Claude Sapanel (France Telecom, France) and George Dimitrakopoulos (University of Piraeus, Greece). The other members of the editorial team were Andrew Aftelak (Motorola), Angeliki Alexiou (Alcatel-Lucent, UK), Stefan Arbanowski (Fraunhofer Fokus, Germany), Brigitte Cardina¨el (FT, France), Klaus David (ComTec, University of Kassel, Germany), Panagiotis Demestichas (University of Piraeus, Greece), Sudhir Dixit (Nokia Siemens Networks, USA), Bernard Hunt (Philips, UK), Nigel Jefferies (Vodafone Group R&D, UK), Wolfgang Kellerer (DOCOMO Euro-Labs, Germany), Vinod Kumar (Alcatel, France), Werner Mohr (Nokia Siemens Networks, Germany), Angela Sasse (UCL, UK), Knud Erik Skouby (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark), Lene Sorensen (Center for Communication, Media and Information Technology/Copenhagen Institute of Technology, Aalborg University, Denmark) and Hu Wang (Huawei, China). Contributions to charging are acknowledged from Mika Klemettinen (Nokia, Finland), Jukka T Salo (NSN, Finland), Pertti H¨oltt¨a (Elisa, Finland) and Olavi Karasti (Elisa, Finland). The team also received material via the WWRF WGs and SIGs.

References [1] A. Jamalipour, T. Wada and T. Yamazato, “A tutorial on multiple access technologies for beyond 3G mobile networks”, IEEE Commun. Mag., 43(2), pp. 110– 117, Feb. 2005. [2] Mitola III and G. Maguire, Jr., “Cognitive radio: making software radios more personal”, IEEE Personal Commun., 6(4), pp. 13– 18, Aug. 1999. [3] R.W. Thomas, L.A. DaSilva and A.B. MacKenzie, “Cognitive networks”, IEEE DySPAN 2005, First Symposium on Dynamic Spectrum Access Networks, USA, Nov. 2005. [4] J. Kephart and D. Chess, “The vision of autonomic computing”, IEEE Computer, 36(1), pp. 41–50, Jan. 2003. [5] J. Palicot and C. Roland, “A new concept for wireless reconfigurable receivers”, IEEE Commun. Mag., 41(7), pp. 124– 133, Jul. 2003. [6] http://www.wireless-world-research.org/, last checked 1 Aug. 2007. [7] http://www.wireless-world-research.org/fileadmin/sites/default/files/publications/Other%20Pub/WWRF_ System_concepts. pdf, last checked 7 Aug. 2007. [8] R. Tafazolli (ed.), “Technologies for the Wireless Future Volume 2: Wireless World Research Forum (WWRF)”, John Wiley & Sons, Ltd, 2006. [9] J. R. B. de Marca, R. Tafazolli and M. A. Uusitalo, “WWRF Visions and Research Challenges for Future Wireless World”, IEEE Commun. Mag., pp. 54–55, Sep. 2004.

Combined View of Future Systems

453

[10] S. Arbanowski, P. Ballon, K. David, O. Droegehorn, H. Eertink, W. Kellerer et al., “I-centric communications: personalization, ambient awareness, and adaptability for future mobile services”, IEEE Commun. Mag., pp. 63– 69, Sep. 2004. [11] http://www.ngmn.org/, last checked 7 Aug. 2007. [12] “Ultra wideband: technology and future perspectives”, V3. 0, white paper, Mar. 2005, available at http://www.wireless-world-research.org/, last checked 1 Aug. 2007.

Appendix: Glossary 2G 3D 3G 3GPP 3GPP2 3GPP-LTE 4G AAA ACID ACF ACS ADL ADSL AJAX AMR AN ANI AoA AoD AP API App APP ARI ARQ ARRM AS ASI ASIC ASM AWGN B2B B3G BAN BB BCH BCJR

2nd Generation Three Dimensional 3rd Generation 3rd Generation Partnership Project 3rd Generation Partnership Project 2 3GPP Long-term Evolution 4th Generation Authentication, Authorisation and Accounting Atomic, Consistent, Isolated and Durable Autonomic Communications Forum Ambient Control Space Architecture Driven Languages Asymmetric Digital Subscriber Line Asynchronous JavaScript and XML Adaptive Multi-rate codec Application Node Ambient Network Interface Angle of Arrival Angle of Departure Access Point Application Programming Interface End-user Application A Posteriori Probability Ambient Resource Interface Automatic ReQuest for repetition Advanced Radio Resource Management Application Server or Angle Spread Ambient Service Interface Application Specific Integrated Circuit Advanced Spectrum Management Additive White Gaussian Noise Business-to-business Beyond 3rd Generation Broadband Access Network Baseband Bose Chaudhuri Hocquenghem Bahl Cocke Jelinek Raviv

Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

456

BER BGCF BICM BLER BP BPEL4WS BPSK BRAN BS C2C CAC CAD CapEx CAT-V CBM CC/PP CCM CCSDS CCSRC CDL CDMA Cdma2000 CFU CLEC CMM CN CNB COST CPC CPE CPT CPU CS CSCF CSI CSMA CTC CW DAB DAML DAML-S DAS DB TC DCS DECT DEEP DIY

Technologies for the Wireless Future – Volume 3

Bit Error Rate Breakout Gateway Control Function Bit Interleaved Coded Modulation BLock Error Rate Belief Propagation Business Process Execution Language for Web Services Binary Phase Shift Keying Broadband Radio Access Network (ETSI Technical Committee) Base Station Car-to-Car Connection Admission Control Computer-Aided Design Capital Expenditure Cable TV Cellular Based Multihop Composite Capability/Preference Profiles Configuration Control Module Consultative Committee for Space Data Systems Cellular Cooperative Short-Range Communication Clustered Delay Line Code Division Multiple Access Code Division Multiple Access for 3rd Generation Check nodes Functional Unit Competitive Local Exchange Carrier Configuration Management Module Check Node Check Node Block European Cooperation in the field of Scientific and Technical Research Cognitive Pilot Channel Customer Premises Equipment Conditional Probability Table Central Processing Unit Central Station Call Session Control Function Channel State Information Carrier Sense Multiple Access Convolutional Turbo-Code Codeword Digital Audio Broadcasting DARPA Agent Markup Language DAML Services Distributed Antenna System Duo-Binary Turbo Code Distributed Communication Sphere Digital Cordless Telecommunication Standard Destination Endpoint Exploring Protocol Do-It-Yourself

Appendix: Glossary

DLNA Dmn DNPM DRiVE DRM DSA DSDP DSL DSSA DS-UWB DVB/DAB DVB-H DVB-RCS DVB-RCT DVB-SH DVB-T E2 R EAM ECC ECMA EJB EMC EMEA EMS ESA ETA eTOM ETSI EU EVDO EXIT FCC FDD FER FFT FIFO FLOWS FMC FOAF FOCTC FP6 FSM FSO FTTH FU FWA

457

Digital Living Network Alliance Domain Ontology development Dynamic Network Planning and Management Dynamic Radio for IP services in Vehicular Environments Digital Rights Management Dynamic Spectrum Allocation Device Software Development Platform Digital Subscriber Line Domain-Specific Software Architecture Direct Sequence Ultra Wideband Digital Video/Audio Broadcasting Digital Video Broadcasting – Handheld Digital Video Broadcasting – Return Channel for Satellite distribution systems Digital Video Broadcasting – Return Channel for Terrestrial distribution systems Digital Video Broadcasting – Satellite services for Handhelds Digital Video Broadcasting – Terrestrial End-to-End Reconfigurability Electro-Absorption Modulator Error Correcting Code European Computer Manufacturers Association Enterprise Java Beans Electromagnetic Compatibility Europe Middle East Africa Extended Min-Sum European Space Agency Estimated Time of Arrival enhanced Telecom Operations Map European Telecommunications Standards Institute European Union Evolution Data Optimized EXtrinsic Information Transfer Federal Communications Commission Frequency Division Duplex Frame Error Rate Fast Fourier Transform First In First Out Ontology of service concepts used in SWSF Fixed-Mobile Convergence Friend of a Friend Frame-Oriented Convolutional Turbo Code Framework Programme 6 Flexible Spectrum Management Free Space Optics Fibre to the Home Functional Units Fixed Wireless Access

458

GAN GBA GCM GF GPON GPRS GPS GREC GSAM GSM GUI HAL HAN HARQ HDMI HDR HDTV HF HFC HGI HGW HiperLAN/2 HMI HSDPA HSS HSUPA HTTP HW ICT ID ID ID-FF ID-SIS ID-WSF IEEE IETF IF ILEC IMS IMT IN IOEM IORM IP IPR IPv4 IRDA

Technologies for the Wireless Future – Volume 3

Generic Access Network Generic Bootstrapping Architecture Generative Communication Model Galois Field Gigabit Optical Passive Network General Packet Radio Service Global Positioning System or General Positioning System Generic Resource Elementary Credits Global Spectrum Allocation Management Global System for Mobile Communications Graphical User Interface Hardware Abstraction Layer Home Area Network Hybrid ARQ High-Definition Multimedia Interface High Data Rate High Definition TV High Frequency Hybrid Fibre Coax Home Gateway Initiative Home Gateway HIgh PERformance LAN/2 Human Machine Interface High-Speed Downlink Packet Access Home Subscriber Server High-Speed Uplink Packet Access Hypertext Transfer Protocol Hardware Information and Communication Technology Identity Identifier Identity Federation Framework Identity Service Interface Specifications Identity Web Services Framework The Institute of Electrical and Electronics Engineers, Inc. Internet Engineering Task Force Intermediate Frequency Incumbent Local Exchange Carrier IP Multimedia Subsystem International Mobile Telecommunications Information Nodes or Intelligent Network Inter-Operator Economic Manager Inter-Operator Resource Management Internet Protocol Intellectual Property Rights Internet Protocol Version 4 Infrared Data Association

Appendix: Glossary

IRS IR-UWB ISI ISIM IST IT ITS ITU ITU-D

459

Internet Reasoning Service Impulse Radio Ultra Wideband Inter-Symbol Interference IP Multimedia Services Identity Module Information Society Technologies Information Technology Intelligent Transportation System International Telecommunications Union International Telecommunication Union – Telecommunication Development sector ITU-R International Telecommunication Union – Radiocommunication sector ITU-T International Telecommunication Union – Telecommunication standardization sector I-WLAN Intelligent Wireless Area Network J2EE Java 2 Platform, Enterprise Edition J2ME Java 2 Platform, Micro Edition J2SE Java 2 Platform, Standard Edition JAIN SLEE Java standard for SLEE JBI Java Business Integration JDK Java Development Kit JMX Java Management Extensions JOaNS Java Open Application Server JVM Java Virtual Machine KC Knowledge Consumer KMF Knowledge Management Framework KN Knowledge Name KS Knowledge Source LAN Local Area Network LBS Location-based Service LDPC Low Density Parity Check (Code) LDR Log-Density Ratio LiMo Linux Mobile Foundation LLC Logical Link Control LLR Log-Likelihood Ratio LNA Low Noise Amplifier LOS Line Of Sight LSAM Local Spectrum Allocation Management LSEM Local Spectrum Economic Manager LTE Long-term Evolution LUT Look-Up Table MAC Medium Access Control MAP Maximum A Posteriori MBWA Mobile Broadband Wireless Access MB-OFDM Multiband OFDM MC-CDMA Multi-Carrier CDMA MCF Maximum Contention Free MC-SS-MA Multi-Carrier Spread-Spectrum Multiple Access

460

MDA MDCS MDS MF MIH MIMO mITF MLD MNO MPC MPLS MPO MPSoC MRF MS MSA MUPE NASA NAT NB-LDPC NBI NFC NGMC NGMN NGN NLOS NO NP NSN O&M OAM OCMC OFDM OFDMA OMA OMG OpenXDF OpEx OS OSS OTA OWL OWL-S PAN PAS PC

Technologies for the Wireless Future – Volume 3

Model-Driven Architecture Multimodal Delivery and Control System Maximum Distance Separable Medium Frequency Media-Independent Handover Multiple Input Multiple Output Mobile IT Forum Maximum Likelihood Decoder Mobile Network Operator Multipath Component Multi-Protocol Label Switching Meta-Operator Multiprocessor System on Chip Media Resource Function Mobile Station Min-Sum Algorithm Multi-User Publishing Environment National Aeronautics and Space Administration Network Access Termination or Network Address Translation Non-Binary LDPC Narrowband Interference Near Field Communication Next-Generation Mobile Communications (Korean Forum) Next-Generation Mobile Networks Next-Generation Network Non-Line of Sight Network Operator Network Platform Nokia Siemens Networks Operation and Maintenance Operation, Administration and Management or Operations and Maintenance Optically Controlled Microstrip Converter Optical Frequency Division Multiplexing or Orthogonal Frequency Division Multiplexing Orthogonal Frequency Division Multiple Access Open Mobile Alliance Object Management Group Open eXchange Data Format Operational Expenditure Operating System Operation Support System; Open Source Software Over the Air Web Ontology Language Semantic Markup for Web Services Personal Area Network Power-Azimuth Spectrum Personal Computer

Appendix: Glossary

PCB PCCC PEP PD PDA PDF PDP PEG PHY PIM PLC PLMN PN PNO POF POI PVR QAM QC QoS QPP QPSK R&D RA RAM RAN RAT RDF RF RFID RIM RLL RM RNC RoF RRM RS RSC RSS RX SATO SAWSDL SC-FDE SCCC SCE SCM SCME

Printed Circuit Board Parallel Concatenated Convolutional Code Policy Enforcement Point Photodetector Personal Digital Assistant Probability Density Function Power-Delay Profile or Policy Decision Point Progressive Edge-Growth Physical Layer Personal Information Management Power Line Communications Public Land Mobile Network Parity Nodes or Pseudo-Noise Public Network Operator Plastic Optical Fibre Point of Interest Personal Video Recorder Quadrature Amplitude Modulation Quasi-Cyclic Quality of Service Quadratic Permutation Polynomial Quaternary Phase Shift Keying Research & Development Repeat Accumulate Random Access Memory Radio Access Network Radio Access Technology Resource Description Framework Radio Frequency Radio Frequency Identification Research in Motion Radio in Local Loop Reed-Muller Radio Network Controller Radio over Fibre Radio Resource Management Reed-Solomon Recursive Systematic Code Really Simple Syndication Receiver Service Aware Transport Overlay Semantic Annotations for WSDL Serial modulation with Frequency Domain reception Serial Concatenation Convolutional Code Service Creation Environment Spatial Channel Model SCM Extension

461

462

SCW SDD SDR SDMA SDSA SED SEE SemSDP SFN SGSN SIB SID SIG SINAD SIP SISO SLA SLEE SMS SN SNR SOA SOAP SONET SOVA SP59 SPI SSO Svc SW SWS SWSF SWSL SWSO TC TCP TDD TDL TDMA TMF TPC TX UAProf UBB UDDI UI UMA

Technologies for the Wireless Future – Volume 3

Service Creation Workbench Soft Decision Decoding Software-Defined Radio Space Division Multiple Access Semantics-Driven Software Architecture Squared Euclidian Distance Service Execution Environment Semantic Service Development Platform Single Frequency Network Serving GPRS Support Node Service Independent Building Blocks Service Identifier Special Interest Group Signal to Noise and Distortion Session Initiation Protocol Soft-In/Soft-Out Service Level Agreement Service Logic Execution Environment Short Message Service Service Node Signal-to-Noise Ratio Service Oriented Architecture Simple Object Access Protocol Synchronous Optical Network Soft-Output Viterbi Algorithm Sphere Packing bound of Shannon 1959 Service Programming Interface Single Sign-On Semantic Service Creation Semantic Web or Software Semantic Web Service Semantic Web Services Framework Semantic Web Services Language Semantic Web Services Ontology Turbo Code Transmission Control Protocol Time Division Duplex Tapped Delay Line Time Division Multiplexing Access TeleManagement Forum Turbo Product Codes Transmitter User Agent Profile Ultra Broadband Universal Description, Discovery and Integration User Interface Unlicensed Mobile Access

Appendix: Glossary

UML UMTS URL UROOF USIM UWB VAS VCSEL VFU VHDR VLC VL-DAS VLSI VM VMM VN VNB VOD VoIP VPN W3C WAN WBAN W-CDMA WER WG WiFi WiMAX WINNER WLAN WPAN WRC WS WSDL WSDL-S WShRNs WSMF WSML WSMO WSMX WSN WWAN WWI WWRF WWW XML

Unified Modeling Language Universal Mobile Telecommunications System Uniform Resource Locator UWB Radio over Optical Fibre Universal Subscriber Identity Module Ultra Wideband Value Added Service Vertical Surface Emitting Laser Variable nodes Functional Unit Very High Data Rate Visible Light Communications Very Low-cost Distributed Antenna System Very Large-Scale Integration Virtual Machine Virtual Machine Monitor Variable Node Variable Node Block Video on Demand Voice over IP Virtual Private Network The World Wide Web Consortium Wide Area Network Wireless Body Area Network Wide Band CDMA Word Error Rate Working Group Wireless Fidelity Worldwide Interoperability of Microwave Access Wireless world INitiative New Radio Wireless Local Area Network Wireless Personal Area Networks World Radio Conference Web Service Web Service Description Language Web Service Semantics Wireless Short-range Networks Web Service Modeling Framework Web Service Modeling Language Web Service Modeling Ontology Web Service Modeling eXecution environment Wireless Sensor Networks or Web Services Notification Wireless Wide Area Networks Wireless World Initiative Wireless World Research Forum World Wide Web eXtensible Markup Language

463

Index 3GPP LTE 3GPP SCM 3GPP 395, 396, 403, 404 60 GHz communication 252, 253 Access Network 393–395, 397–404, 414, 421–4, 441 Access Point 394 access provider requirements 23 accessibility 11, 14 Activity Theory 83, 84, 117 ad-hoc 109, 124, 129, 132, 143, 147, 156, 159, 160 Adaptation 58–159, 124 ADSL 389, 395 Africa 2, 4 Alcatel-Lucent 2 Ambient Control Space 166–167 ambient intelligence 157, 158 Ambient Network Interface 167 Ambient Network Service Interface 147 Ambient Control Space (ACS) 147, 147–151 Ambient Network Interface (ANI) 147, 148, 149 Ambient Resource Interface (ARI) 147, 148 Ambient Service Interface (ASI) 147, 147–151 Ambient Networks 165, 425–7 Ambient Resource Interface 166–167 Ambient Service Interface 166–167 Ambient 450–1 AMR voice codec 3 Apple iPhone 4 application layer 166 Application Programming Interface (API) 68, 82, 95, 98, 104–106, 117, 122, 123, 136–159 Application 57–159 Architectural Principles 166 Architecture 57–67, 72, 73–82, 85, 86–116, 118, 119, 120–132, 135, 137, 138, 139, 140, 142, 143–155, 156, 157–160

scalability 79, 96, 120 augmented environments 18 Automatic Composition 167 Autonomic computing 408, 437–8, 440, 452 autonomic systems 124 B3G/4G 222 backbone provider requirements 22 beamforming, 198, 234 downlink 235 uplink 235 belonging 11, 12 Beyond the 3rd generation: 349 Blackberry 4 block diagonalization (BD) 235 business evaluation 113 business issues 114 in Service Domain 114 in Technology Domain 114 in Organisational Domain 114 in Financial Domain 114 Business models 412, 414, 416–7, 424, 450 Business model 30, 39, 50, 60 business model 80–88, 106, 107, 109, 110–118, 132, 160 Customer Centric Business Model 109, 110 Device Manufacturer Centric Business Model 110 Service Provider Centric Business Model 110 Network Operator Centric Business Model 112 business roles 17, 20 CAC (Connection Admission Control) 397 capability plane 11, 15 CAPEX 436–7 capital expenditure (CapEx) 22, 23 cellular networks 252–257, 265, 271, 273 Centralized spectrum sharing 370 channel state information 199, 225, 234 charging and billing 60, 112, 120, 127, 153–160

Technologies for the Wireless Future – Volume 3 Edited by Klaus David  2008 Wireless World Research Forum (WWRF)

466

China Mobile 2 China 4 chirp 276, 277, 291, 327, 340 clustered delay line (CDL) 226 Codes on graphs 205 coexistence 314, 324, 331, 335, 336–338 Cognitive Pilot Channel 444 Cognitive Radio 191, 335–337, 368 Cognitive Wireless Networks: 350 Cognitive 408, 429, 435–447, 451 community 61, 75, 83–123, 160 Competitive Local Exchange Carrier 397 composed service 58, 82, 83, 118, 119, 130–137 Context Awareness 9, 10, 11, 13, 16, 17, 19, 20, 168, 180, 183, 408, 410–1, 432 Context Broker 184–185 Context Client 184 Context Information Base 184, 187, 192 context information 181–182, 187 Context Management 183, 186 Context Provider 182–186 Context Provisioning 180 Context Representation 184 Context Source 181, 183, 192 context 58–159, 75 aware service 64, 72, 75, 80, 120, 159 awareness 67–79,125, 159 broker 120 control 11, 12, 13, 14, 15, 21, 23, 24 convergence 58, 64, 80, 106–122, 151–153 cooperation 252–254, 257–259, 261–265, 270, 272 cost-effective 18, 19, 20 Creative workshop 32, 33 Creativity 30, 31, 32 Cultural probes 30 DAMLS-S 68 DAML 65 decision feedback equalization 237 Decoder architectures 213 Demand 29–54 Device Context 181 Device 58–160, 110, 137 Aspects 137 service architecture 137, 138, 139 modularization 137–139 platformization 137–139 Digital Rights Management (DRM) 60, 102 dirty paper code 235

Index

Distributed Communication Sphere (DCS) 66, 128, 129, 130, 144 Distributed Spectrum Sharing 373, 378, 379 diversity 198, 199, 218, 235 DLNA (Digital Living Network Alliance) 390 DRiVE project 370, 371 Driving forces 35, 37, 38 DS-UWB 252, 274, 275 dynamic channel allocation 369 E2 R project 372 E2R 165 Edholm’s law 341 Electro-absorption transceiver (EAT) 277, 278, 288, 292, 293, 300–310 enabling technology 118 End-to-End reconfiguration 166, 193 end-user perspective 94 equipment manufacturer requirements 24 Ericsson 2 Error Control Coding 199 eTOM 128 Expert workshop 33 External cognitive aids 30, 32 FaceBook 4 fading (see also radio channel models) 231 FCC 376, 377, 381, 382 Fiber To The home 395 Fixed Mobile Convergence 397, 402 Fixed-Mobile Convergence (FMC) 58, 80, 121, 151–154 Flow Context 181 FOAF 67 France Telecom 2 Free Space Optics 393 Generic Access Network 403 Generic Bootstrapping Architecture (GBA) 135, 136, 144, 155, 160 Generic Service Element 60 geometry-based stochastic approach 228 Google Android 4 GPRS 403, 404 GPS 63, 66 Graphical User Interface (GUI) 101, 102, 109 GSM 403, 404 handover process 178–179 Hardware Abstraction Layer (HAL) 138, 145, 146

Index

HDTV (High Definition TV) 389, 390 Heterogeneous Radio Resource Management 169 Home Area Network 389, 395 Home Gateway Initiative 399 Home Gateway 391, 395, 389, 399 HSDPA 3, 5, 7, 57 HSS 104, 122, 135, 136 HSUPA 3, 5, 7 Huawei 2 human capability augmentation 11, 13 Human Machine Interface (HMI) 63, 157, 158 Human User/Group Context 181 I-centric 20, 22 I-WLAN 403, 404 IDEA MAGNET 45 Identity Federation Framework (I-FF) 135 Identity Service Interface Specifications (ID-SIS) 135, 136 Identity 60, 64, 72–76, 120–133, 134–136, 153–160 Handling 72, 76 Management 60, 64, 75, 121, 129, 133, 160 Framework 133 IEEE802.15.3c 251, 252, 273, 316 IETF 404 ILEC 397 imaging 258, 323, 324, 332, 338–342 IMS 403, 404 India 4 Intelligent transportation systems 279, 286 Intel 2 inter-system coordination 369 inter-RAT handover 178–179 International Telecommunications Union 5 intra-system coordination 369 IP Multimedia Subsystem (IMS) 103–105, 120–123, 129, 134–136, 143, 151, 153–155, 159, 160 IR-UWB 274, 275, 324, 325, 327, 328, 330, 333, 334 IrDA 393 Iterative decoding 202 ITU 393, 398 Java 73, 79, 95–101, 106, 118, 122, 123 J2EE 73, 79, 97, 98 J2ME 99, 100, 101 J2SE 73, 79, 99, 101 JAIN SLEE 97, 98, 99 JBI 73, 74, 79

467

JDK 98 JOnAS 97, 98 Knowledge Management 191 knowledge management 130, 131, 132, 143 LDPC 201 legal and regulatory requirements 26 LG Electronics 2 License exempt systems 378, 379 LiMo 4 Linux 4 localization 253, 273, 285, 323–325, 331–334, 338–342 Location Based Service (LBS) 62, 63, 76 loosely coupled 129, 130, 155 Low-fi paper prototyping 47 LTE 404 MAGNET 31, 43 Management 192–194 market requirements 17 market 57–64, 80, 87–93, 102, 106–108–110, 114–119, 152–157 MB-OFDM 274, 275, 299, 324, 330, 333 Microsoft Windows Mobile 4 MIMO 198, 253, 283–285, 334, 341 Mobile probing kit 44 Mobile TV 4 mobile 57–160 device/phone 58, 72–80, 97–113, 119, 124, 134–138, 156, 157 services 58–66–76, 79, 80, 91, 100–136, 144, 159, 160 MobiLife 165 Mobility management 177–180 multimodal 60, 144 multiuser MIMO 234 MUPE 97, 99, 100, 101 MySpace 4 Narrative stories 31 Navigation 4 Navteq 4 NEC 2 needs plane 11 Network Context 181–184 network layer 182 New air interface 425, 427–9 Next Generation Networks (NGN) 121, 153, 154 Next generation wireless systems 197

468

NFC 64, 76 NGMN 5 NGN 397, 398 Nokia Siemens Networks 2 Nokia 2 Nomadicity 29, 43, 47, 53 Nortel 2 of hardware 139, 140 OMA 67, 102–105, 117–121, 127, 155 Ontologies 182–183 Ontology 65, 66, 67, 68, 70–73, 75–79, 124, 128–131, 144 Mediation 66, 71, 72, 73, 79 operational expenditure (OpEx) 22, 23 operational management 126 Operational 409–10, 412, 437, 440, 448 OPEX 436–7 Opportunistic spectrum use 369, 381 optical communication 287, 289, 314 optical fibre 252, 272–274, 276, 277, 279, 280, 282, 287, 288, 289, 313, 314, 394 OverDRiVE project 370, 371 OWL-S 67, 68–70, 78 OWL 65–71, 144 Participatory Design 30, 31 peer-to-peer networks 252–254, 267, 270, 272 peer-to-peer 58, 64, 80, 109, 123, 124, 127 Personal network 31, 44 personalisation 63, 64, 67. 72, 75, 78, 90, 106, 114, 119, 120, 125, 159 Personalization 408, 410, 413, 436 personalization 11, 15, 16, 18 platform, 58–81, 82, 86, 88, 90, 96, 97–101, 109, 110, 114, 116, 118–120, 123–133, 136–138, 139, 142–145, 146, 147, 152, 153, 155, 157–160 device 137, 144 service interfaces, 145, 146 low-level 146 high-level 145 Policy Enforcement 190 Policy 75, 80, 121, 126, 132–134 Decision Point (PDP) 133 Enforcement Point (PEP) 121, 133 precoding, 234 linear 234 non-linear 237 Tomlinson-Harashima 237 Presence 121, 132, 135, 143, 155

Index

pricing 19 Privacy 11, 14, 25, 26, 27, 59, 60, 64, 72, 75, 76, 79, 80, 105, 106, 119, 125, 126, 127, 129, 130, 132, 132, 133, 134, 144, 160, 187 and trust 125 management 129 framework 132 Product codes 200 profile 61, 63, 68, 106, 108, 112, 114, 125, 127, 130–136, 143, 152 programmable network 140, 141, 142 propagation (see also radio channel models) Quality of Service (QoS) 60, 73, 80, 103, 108, 112, 114, 119, 127, 132, 154, 160 Quality of Service: 350 Quality of Service 417, 425, 434, 436, 450 radar 277, 287, 324, 325, 337–340 radio channel measurements 226 radio channel models 224 3GPP LTE 231 3GPP SCM 230 COST273 226 COST2100 226 DVB-H 230 IEEE 802.11n 232 IMT-Advanced 232 WiMAX 232 WINNER 226 radio resource management 372, 374, 383 RDF 65, 67, 75, 144 Reasoning 67, 71, 74, 125, 131, 132 Recommendation 125, 132, 143, 144 reconfigurability layer 182 Reconfigurability 431, 435, 447 Reconfigurable 408–9, 415, 428, 431, 435–452 reconfiguration plane 166 reconfiguration 191–192 reference architecture 86, 121 regularized block diagonalization (RBD) 236 regulation 252, 279, 315, 316, 324, 335, 336–338, 341 Relay-based multihop 429, 451 reliability 11, 15, 24 Research in Motion 2 revenue 13, 19, 21, 22, 25 RFID 64 Roaming 177 RoF (Radio over Fiber) 394, 400

Index

rule-based algorithm 59, 75 safety 10, 11, 12, 13, 25 Samsung 2 SAWSDL 67, 68, 70, 71, 78 Scalabilit`a 79, 98, 120 Scenario landscape 32 scenario, 57, 58, 60, 61, 62, 63, 76, 89, 90, 119, 133, 156, 159, 160 scenarios for service creation 89 scenarios for vehicular communications 119, 156, 159, 160 Scenarios 30 SDMA 199, 234 security 10, 11, 12, 13, 15, 18, 19, 20, 24, 25 Security 187 security 59, 60, 63, 72, 75, 76, 79, 80, 93, 105, 107, 108, 110–114, 119, 121, 123, 130, 134, 135, 137, 138, 142, 144, 145, 147, 149, 150, 159 self actualization 13, 14 Self-Organization 408, 419, 430, 433–4, 447–9, 452 Self-ware Reconfiguration Management Plane (S-RMP) 191–193 Semantic Web 59, 65, 67, 69, 73, 74, 75, 76, 78, 87 Services Framework (SWSF) 67, 68, 70, 71, 74, 78 Services Language (SWSL) 70, 73, 78 Services Ontology (SWSO) 70, 78 semantic 58, 59–61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77–80, 81, 85, 86, 87, 88, 91, 94, 117, 118, 123, 129, 130, 143, 158 services 58, 59, 63, 67, 68, 69, 70, 71, 72, 74, 75, 77, 79, 80, 81 service provisioning 60, 72, 78 Web 59, 65, 67, 69, 73–78, 87 Web Services 59, 63, 67, 70, 71, 74–80 sensing 267, 323, 324, 332, 337, 338, 341, 342 sensor networks 252, 253, 256, 270, 272, 323, 331, 335, 340 Sensor Networks 408, 424, 429, 432–5, 448, 451 sensor 60, 64, 73, 123, 132, 156, 157 Data 132, 156 networks 60, 64 Service Context 181 Service Delivery Framework 119, 142 Service Development 46

469

service discovery 261 Service Level Agreement (SLA) 59, 76, 111, 112, 133, 141, 143, 144, 160 Service Oriented Architecture (SOA) 58, 67, 72, 74–79, 98, 101, 123, 145, 147 Service Platform Architecture 59, 60, 123, 159 service provider requirements 21 Service Value Network 109 service 57–161 adaptation 58, 74,124, 125 architecture 57–61, 64, 72, 75, 79, 87, 118–121, 124, 131, 132, 137, 138, 139, 145, 157, 158, 159, 160 broker 145 capability 123 component 65, 72, 78, 91, 93, 129, 130, 145 composition 72, 78, 94, 95, 106, 126, 129–131, 143, 145 creation 58, 60, 61, 67, 76, 77, 80, 81, 86, 89, 90, 93, 94, 96, 101, 102, 105, 106, 107, 108, 109, 113, 116–119, 129, 143, 154, 160 creation environment 93, 94, 101, 105, 106, 119, 143, 154 Eclipse 96, 101 Service Creation Workbench 96, 101, 102 discovery 61, 62, 68, 71, 75, 125, 126 domain 64, 82, 114, 115 enabler 67, 82, 97, 102–105, 121, 129, 131, 136, 155 execution environment 97, 102, 119, 133, 142, 143 infrastructure 57–161 interworking 72 layer 127, 142, 143, 144 lifecycle 60, 72, 78, 126, 127, 128, 131 ontology 68 provisioning 59, 60, 63, 72, 74, 76, 78, 79, 109, 119–121, 124, 126, 157, 159 requirement 63, 107 roaming 136, 137 usage 58, 60, 107, 118, 126, 127, 136, 160 Short-range wireless communications 251 Single Sign-On (SSO) 121, 160 SIP 404 Smart antennas 431–2, 451 SMMSE precoding 238 SOAP 67, 69, 136, 145, 148 social networking 252, 264 spatial channels (see also radio channel models) 224

470

Spectrum Sharing 367, 368, 369, 370, 372, 383, 384, 385 SPICE 165 stakeholder requirements 9, 10, 11 subsistence 11 successive optimization 236 sum-rate capacity 237 Supply 50, 51 sustainable development 18, 20 Symbian 4 System Architecture 407 system requirements 9, 10 Telemanagement Forum (TMF) 128 triggering mechanism 179 trust 11, 13, 14, 15, 25, 27 Trust 59, 60, 64, 73, 76, 79, 107, 112–114, 125–127, 134, 156–160 Engine 126 Turbo codes 201 Turbo-principle 218 UBB 389–390–395–405 ubiquitous communication 13, 18 ubiquitous information access 11, 13, 17 Ubiquitous 59, 60, 64, 81, 89, 124, 131 computing 64, 89 service 59, 81 UDDI 67, 70, 102 Ultra high speed wireless communication 334 ultra wideband (UWB) communication 272, 273 Ultra Wideband Radio over Optical Fibre 252, 272, 287 UMA 403. 404 UMTS 404 Usability 29, 47, 53 User centred design methods 29, 53 User centricity 29, 37 User focused workshop 31 user requirements 10, 11, 21 User requirements 43, 41 User scenario 31, 33, 37 User-centred design 2 user-centric approach 59 user-generated products/services 94 User 58–160 centricity 59, 74, 75, 86, 92, 107 context 58–61, 74, 75, 99, 130 preference 58–62, 74–76, 79, 108, 123, 132, 152 USIM 403

Index

utility 11, 15 UWB antennas 326, 332 UWB indoor channel 312 UWB 369, 382, 392, 394, 400 V-BLAST 237 Value Added Service (VAS) 61, 64, 111, 112, 142, 143 value plane 11, 14 VCSEL 276, 288, 289–291, 299, 300–304, 306 Virtualization 120, 139, 140, 141, 159 Visionpool 44, 45 Vision 407–11, 437, 441, 450–2 Vodafone 2 VOD 389, 393, 396, 398 VoIP 57, 80, 154, 393, 403, 404 Web Service 58, 59, 63, 67, 68, 69, 70, 71, 72, 74, 75, 77, 78, 79, 80, 81, 102, 122, 145 Description Language (WSDL) 67, 69, 70, 71, 78, 102, 122, 145 Semantics (WSDL-S) 68, 69, 71, 78 Modelling Framework (WSMF) 70, 78 Modelling Language (WSML) 65, 70, 74 Modelling Ontology (WSMO) 67–69, 70, 71, 74, 78 Modelling eXecution environment (WSMX) 74 Western Europe 4 Wiki 3 WiMAX 395–399 WINNER project 374, 375, 383, 384, 385 wireless grid 252, 254, 257, 258, 260, 264, 265, 272 Wireless HDMI 318 Wireless Regional Area Network 377 Wireless ubiquitous coverage 423 WLAN 392, 403, 404 Workshop 31, 32, 44 WWI 2, 6 WWRF articles of association 3 WWRF chairman 2 WWRF chairs 2 WWRF executives 2 WWRF General Assembly 2 WWRF Reference Scenarios 30, 34 WWRF Steering Board 2 WWRF system concept 1 WWRF treasurer 2 WWRF vice-chairs 2 WWRF vision 407, 450–2 WWRF working groups 2