ADVANCED WIRELESS NETWORKS Cognitive, Cooperative and
Opportunistic 4G Technology Second Edition Savo Glisic Beatriz
Lorenzo University of Oulu, Finland A John Wiley and Sons, Ltd.,
Publication
ADVANCED WIRELESS NETWORKS
ADVANCED WIRELESS NETWORKS Cognitive, Cooperative and
Opportunistic 4G Technology Second Edition Savo Glisic Beatriz
Lorenzo University of Oulu, Finland A John Wiley and Sons, Ltd.,
Publication
This edition rst published 2009 C 2009 John Wiley & Sons
Ltd., Registered ofce John Wiley & Sons Ltd, The Atrium,
Southern Gate, Chichester, West Sussex, PO19 8SQ, United Kingdom
For details of our global editorial ofces, 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 identied 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. Library of Congress Cataloging-in-Publication
Data Glisic, Savo G. Advanced wireless networks : 4G technologies /
Savo Glisic, Beatriz Lorenzo Veiga. 2nd ed. p. cm. Includes
bibliographical references and index. ISBN 978-0-470-74250-1
(cloth) 1. Wireless communication systems. I. Veiga, Beatriz
Lorenzo. II. Title. TK5103.2.G553 2009 621.384dc22 2009001817 A
catalogue record for this book is available from the British
Library. ISBN 978-0-470-74250-1 (H/B) Typeset in 9/11 Times by
Laserwords Private Limited, Chennai, India Printed in Singapore by
Markono Print Media Pte Ltd
To our families
Contents Preface to the Second Edition xix 1 Fundamentals 1.1
4G Networks and Composite Radio Environment 1.2 Protocol Boosters
1.2.1 One-element error detection booster for UDP 1.2.2 One-element
ACK compression booster for TCP 1.2.3 One-element congestion
control booster for TCP 1.2.4 One-element ARQ booster for TCP 1.2.5
A forward erasure correction booster for IP or TCP 1.2.6
Two-element jitter control booster for IP 1.2.7 Two-element
selective ARQ booster for IP or TCP 1.3 Green Wireless Networks
References 1 1 7 9 9 9 9 10 10 11 11 11 2 Opportunistic
Communications 2.1 Multiuser Diversity 2.2 Proportional Fair
Scheduling 2.3 Opportunistic Beamforming 2.4 Opportunistic Nulling
in Cellular Systems 2.5 Network Cooperation and Opportunistic
Communications 2.5.1 Performance example 2.6 Multiuser Diversity in
Wireless Ad Hoc Networks 2.6.1 Multiple-output and multiple-input
link diversity 2.6.2 Localized opportunistic transmission 2.6.3
Multiuser diversity-driven clustering 2.6.4 Opportunistic MAC with
timeshare fairness 2.6.5 CDF-based K-ary opportunistic splitting
algorithm 2.6.6 Throughput 2.6.7 Optimal opportunistic MAC 15 15 16
19 20 22 25 27 29 30 31 34 34 37 37
viii CONTENTS 2.7 2.8 2.6.8 Contention resolution between
clusters 2.6.9 Performance examples Mobility-Assisted Opportunistic
Scheduling (MAOS) 2.7.1 Mobility models 2.7.2 Optimal MAOS
algorithm 2.7.3 Suboptimum MAOS algorithm 2.7.4 Mobility estimation
and prediction 2.7.5 Estimation of Lagrange multipliers 2.7.6
Performance examples Opportunistic and Cooperative Cognitive
Wireless Networks 2.8.1 The system model 2.8.2 The outage
probability 2.8.3 Cellular trafc shaping 2.8.4 User mobility
modeling 2.8.5 Absorbing Markov chain system model 2.8.6 Throughput
analysis 2.8.7 Collision resolution 2.8.8 Opportunistic
transmission with intercell interference awareness 2.8.9
Performance examples References 38 40 46 48 49 51 51 52 52 53 53 57
58 59 61 62 65 65 68 70 3 Relaying and Mesh Networks 3.1 Relaying
Strategies in Cooperative Cellular Networks 3.1.1 The system model
3.1.2 System optimization 3.1.3 Relay strategy selection
optimization 3.1.4 Performance example 3.2 Mesh/Relay Networks
3.2.1 The system model 3.2.2 Exhaustive sleep 3.2.3 Practical
applications 3.2.4 Performance example 3.3 Opportunistic Ad Hoc
Relaying For Multicast 3.3.1 The system model 3.3.2 Proxy discovery
and route interference 3.3.3 Near-optimal multicast and
approximations 3.3.4 Performance examples References 73 73 73 75 79
84 85 86 88 94 95 97 98 99 101 103 107 4 Topology Control 4.1 Local
Minimum Spanning Tree (LMST) Topology Control 4.1.1 Basics of MST
topology control 4.1.2 Performance examples 4.2 Joint Topology
Control, Resource Allocation and Routing 4.2.1 JTCR algorithm 4.3
Fault-Tolerant Topology 4.3.1 The system model 4.3.2 Fault-tolerant
topology design 4.3.3 -Approximation algorithms 4.3.4 Performance
examples 113 115 115 118 118 121 123 124 124 127 132
CONTENTS 4.4 ix Topology Control in Directed Graphs 4.4.1 The
system model 4.4.2 Minimum-weight-based algorithms 4.4.3
Augmentation-based algorithms 4.4.4 Performance examples Adjustable
Topology Control 4.5.1 The system model 4.5.2 The r -neighborhood
graph Self-Conguring Topologies 4.6.1 SCT performance References
132 133 133 135 138 138 140 142 143 145 148 5 Adaptive Medium
Access Control 5.1 WLAN Enhanced Distributed Coordination Function
5.2 Adaptive MAC for WLAN with Adaptive Antennas 5.2.1 Description
of the protocols 5.3 MAC for Wireless Sensor Networks 5.3.1 S-MAC
protocol design 5.3.2 Periodic listen and sleep 5.3.3 Collision
avoidance 5.3.4 Coordinated sleeping 5.3.5 Choosing and maintaining
schedules 5.3.6 Maintaining synchronization 5.3.7 Adaptive
listening 5.3.8 Overhearing avoidance and message passing 5.3.9
Overhearing avoidance 5.3.10 Message passing 5.4 MAC for Ad Hoc
Networks 5.4.1 Carrier sense wireless networks 5.4.2 Interaction
with upper layers References 157 157 160 160 166 167 168 168 169
169 170 170 172 172 172 174 176 179 180 6 Teletrafc Modeling and
Analysis 6.1 Channel Holding Time in PCS Networks References 183
183 191 7 Adaptive Network Layer 7.1 Graphs and Routing Protocols
7.1.1 Elementary concepts 7.1.2 Directed graph 7.1.3 Undirected
graph 7.1.4 Degree of a vertex 7.1.5 Weighted graph 7.1.6 Walks and
paths 7.1.7 Connected graphs 7.1.8 Trees 7.1.9 Spanning tree 7.1.10
MST computation 7.1.11 Shortest path spanning tree 7.2 Graph Theory
193 193 193 193 194 194 195 195 195 196 197 199 201 212 4.5
4.6
x CONTENTS 7.3 7.4 Routing with Topology Aggregation Network
and Aggregation Models 7.4.1 Line segment representation 7.4.2
QoS-aware topology aggregation 7.4.3 Mesh formation 7.4.4 Star
formation 7.4.5 Line-segment routing algorithm 7.4.6 Performance
measure 7.4.7 Performance example References 214 215 217 220 220
221 222 224 225 228 8 Effective Capacity 8.1 Effective Trafc Source
Parameters 8.1.1 Effective trafc source 8.1.2 Shaping probability
8.1.3 Shaping delay 8.1.4 Performance example 8.2 Effective Link
Layer Capacity 8.2.1 Link-layer channel model 8.2.2 Effective
capacity model of wireless channels 8.2.3 Physical layer vs
link-layer channel model 8.2.4 Performance examples References 235
235 237 238 238 241 243 244 246 249 251 254 9 Adaptive TCP Layer
9.1 Introduction 9.1.1 A large bandwidth-delay product 9.1.2 Buffer
size 9.1.3 Round-trip time 9.1.4 Unfairness problem at the TCP
layer 9.1.5 Noncongestion losses 9.1.6 End-to-end solutions 9.1.7
Bandwidth asymmetry 9.2 TCP Operation and Performance 9.2.1 The TCP
transmitter 9.2.2 Retransmission timeout 9.2.3 Window adaptation
9.2.4 Packet loss recovery 9.2.5 TCP-OldTahoe (timeout recovery)
9.2.6 TCP-Tahoe (fast retransmit) 9.2.7 TCP-Reno fast retransmit,
fast (but conservative) recovery 9.2.8 TCP-NewReno (fast
retransmit, fast recovery) 9.2.9 Spurious retransmissions 9.2.10
Modeling of TCP operation 9.3 TCP for Mobile Cellular Networks
9.3.1 Improving TCP in mobile environments 9.3.2 Mobile TCP design
9.3.3 The SH-TCP client 9.3.4 The M-TCP protocol 9.3.5 Performance
examples 257 257 258 259 260 261 262 262 263 264 264 265 265 265
265 265 265 266 267 267 268 269 270 272 273 275
CONTENTS 9.4 9.5 Random Early Detection Gateways for Congestion
Avoidance 9.4.1 The RED algorithm 9.4.2 Performance example TCP for
Mobile Ad Hoc Networks 9.5.1 Effect of route recomputations 9.5.2
Effect of network partitions 9.5.3 Effect of multipath routing
9.5.4 ATCP sublayer 9.5.5 ATCP protocol design 9.5.6 Performance
examples References 10 Network Optimization Theory 10.1
Introduction 10.2 Layering as Optimization Decomposition 10.2.1 TCP
congestion control 10.2.2 TCP Reno/RED 10.2.3 TCP Vegas/Drop Tail
10.2.4 Optimization of the MAC protocol 10.2.5 Utility optimal MAC
protocol/social optimum 10.3 Crosslayer Optimization 10.3.1
Congestion control and routing 10.3.2 Congestion control and
physical resource allocation 10.3.3 Congestion and contention
control 10.3.4 Congestion control, routing and scheduling 10.4
Optimization Problem Decomposition Methods 10.4.1 Decoupling
coupled constraints 10.4.2 Dual decomposition of the basic NUM
10.4.3 Coupling constraints 10.4.4 Decoupling coupled objectives
10.4.5 Alternative decompositions 10.4.6 Application example of
decomposition techniques to distributed crosslayer optimization
10.5 Optimization of Distributed Rate Allocation for Inelastic
Utility Flows 10.5.1 Nonconcave utility ows 10.5.2 Capacity
provisioning for convergence of the basic algorithm 10.6 Nonconvex
Optimization Problem in Network with QoS Provisioning 10.6.1 The
system model 10.6.2 Solving the nonconvex optimization problem for
joint congestioncontention control 10.7 Optimization of Layered
Multicast by Using Integer and Dynamic Programming 10.7.1 The
system model 10.7.2 Lagrangian relaxation for integer programs
10.7.3 Group prot maximization by dynamic programming 10.8 QoS
Optimization in Time-Varying Channels 10.8.1 The system model
10.8.2 Dynamic control algorithm 10.9 Network Optimization by
Geometric Programming 10.9.1 Power control by geometric
programming: high SNR 10.9.2 Power control by geometric
programming: low SNR 10.10 QoS Scheduling by Geometric Programming
xi 276 276 277 280 280 280 280 281 282 287 287 289 289 290 290 291
292 292 295 298 298 301 303 306 307 307 308 310 310 313 315 319 319
322 323 323 325 326 327 329 329 331 331 332 337 338 340 340
xii CONTENTS 10.10.1 Optimization of OFDM system by GP 10.10.2
Maximum weight matching scheduling by GP 10.10.3 Opportunistic
scheduling by GP 10.10.4 Rescue scheduling by GP References 344 344
345 345 346 11 Mobility Management 11.1 Introduction 11.1.1
Mobility management in cellular networks 11.1.2 Location
registration and call delivery in 4G 11.2 Cellular Systems with
Prioritized Handoff 11.2.1 Channel assignment priority schemes
11.2.2 Channel reservation CR handoffs 11.2.3 Channel reservation
with queueing CRQ handoffs 11.2.4 Performance examples 11.3 Cell
Residing Time Distribution 11.4 Mobility Prediction in Pico- and
MicroCellular Networks 11.4.1 PST-QoS guarantees framework 11.4.2
Most likely cluster model Appendix: Distance Calculation in an
Intermediate Cell References 351 351 353 355 374 377 377 378 382
383 388 390 391 398 403 12 Cognitive Radio Resource Management 12.1
Channel Assignment Schemes 12.1.1 Different channel allocation
schemes 12.1.2 Fixed channel allocation 12.1.3 Channel borrowing
schemes 12.1.4 Simple channel borrowing schemes 12.1.5 Hybrid
channel borrowing schemes 12.1.6 Dynamic channel allocation 12.1.7
Centralized DCA schemes 12.1.8 Cell-based distributed DCA schemes
12.1.9 Signal strength measurement-based distributed DCA schemes
12.1.10 One-dimensional cellular systems 12.1.11 Reuse partitioning
(RUP) 12.2 Dynamic Channel Allocation with SDMA 12.2.1 Single-cell
environment 12.2.2 Resource allocation 12.2.3 Performance examples
12.3 Packet-Switched SDMA/TDMA Networks 12.3.1 The system model
12.3.2 Multibeam SDMA/TDMA capacity and slot allocation 12.3.3
SDMA/TDMA slot allocation algorithms 12.3.4 SDMA/TDMA performance
examples 12.4 SDMA/OFDM Networks with Adaptive Data Rate 12.4.1 The
system model 12.4.2 Resource allocation algorithm 12.4.3 Impact of
OFDM/SDMA system specications on resource allocations 12.4.4
Performance examples 12.5 Intercell Interference Cancellation SP
Separability 407 407 409 410 410 411 412 414 415 417 419 420 422
426 426 430 435 435 437 439 441 445 446 446 448 450 453 454
CONTENTS 12.6 12.7 12.8 12.9 12.5.1 Channel and cellular system
model 12.5.2 Turbo spacetime multiuser detection for intracell
communications 12.5.3 Multiuser detection in the presence of
intercell interference 12.5.4 Performance examples Intercell
Interference Avoidance in SDMA Systems 12.6.1 The BOW scheme 12.6.2
Generating beam-off sequences 12.6.3 Constrained QRA-IA Multilayer
RRM 12.7.1 The SRA protocol 12.7.2 The ESRA protocol Resource
Allocation with Power Preassignment (RAPpA) 12.8.1 Resource
assignment protocol 12.8.2 Analytical modeling of RAPpA Cognitive
and Cooperative Dynamic Radio Resource Allocation 12.9.1
Signal-to-interference ratio 12.9.2 System performance 12.9.3
Multicell operation 12.9.4 Performance examples Appendix 12A: Power
Control, CD Protocol, in the Presence of Fading Appendix 12B:
Average Intercell Throughput References 13 Ad Hoc Networks 13.1
Routing Protocols 13.1.1 Routing protocols 13.1.2 Reactive
protocols 13.2 Hybrid routing protocol 13.2.1 Loop-back termination
13.2.2 Early termination 13.2.3 Selective broadcasting (SBC) 13.3
Scalable Routing Strategies 13.3.1 Hierarchical routing protocols
13.3.2 Performance examples 13.3.3 FSR (sheye routing) protocol
13.4 Multipath Routing 13.5 Clustering Protocols 13.5.1
Introduction 13.5.2 Clustering algorithm 13.5.3 Clustering with
prediction 13.6 Cashing Schemes for Routing 13.6.1 Cache management
13.7 Distributed QoS Routing 13.7.1 Wireless links reliability
13.7.2 Routing 13.7.3 Routing information 13.7.4 Token-based
routing 13.7.5 Delay-constrained routing 13.7.6 Tokens 13.7.7
Forwarding the received tokens 13.7.8 Bandwidth-constrained routing
xiii 455 457 459 460 461 467 468 468 470 471 473 475 476 479 484
486 488 491 492 494 498 499 505 505 507 512 524 526 527 528 531 531
533 534 537 539 539 541 542 549 549 558 558 558 559 559 560 561 562
562
xiv CONTENTS 13.7.9 Forwarding the received tickets 13.7.10
Performance example References 562 564 567 14 Sensor Networks 14.1
Introduction 14.2 Sensor Networks Parameters 14.2.1 Pre-deployment
and deployment phase 14.2.2 Post-deployment phase 14.2.3
Re-deployment of additional nodes phase 14.3 Sensor networks
architecture 14.3.1 Physical layer 14.3.2 Data link layer 14.3.3
Network layer 14.3.4 Transport layer 14.3.5 Application layer 14.4
Mobile Sensor Networks Deployment 14.5 Directed Diffusion 14.5.1
Data propagation 14.5.2 Reinforcement 14.6 Aggregation in Wireless
Sensor Networks 14.7 Boundary Estimation 14.7.1 Number of RDPs in P
14.7.2 Kraft inequality 14.7.3 Upper bounds on achievable accuracy
14.7.4 System optimization 14.8 Optimal Transmission Radius in
Sensor Networks 14.8.1 Back-off phenomenon 14.9 Data Funneling
14.10 Equivalent Transport Control Protocol in Sensor Networks
References 573 573 575 576 576 577 577 578 578 581 585 586 587 590
591 593 593 596 598 598 599 600 602 606 607 610 613 15 Security
15.1 Authentication 15.1.1 Attacks on simple cryptographic
authentication 15.1.2 Canonical authentication protocol 15.2
Security Architecture 15.3 Key Management 15.3.1 Encipherment
15.3.2 Modication detection codes 15.3.3 Replay detection codes
15.3.4 Proof of knowledge of a key 15.3.5 Point-to-point key
distribution 15.4 Security management in GSM networks 15.5 Security
management in UMTS 15.6 Security architecture for UMTS/WLAN
Interworking 15.7 Security in Ad Hoc Networks 15.7.1 Self-organized
key management 15.8 Security in Sensor Networks References 623 623
625 629 631 635 637 637 637 637 638 639 643 645 647 651 652
654
CONTENTS xv 16 Active Networks 16.1 Introduction 16.2
Programable Networks Reference Models 16.2.1 IETF ForCES 16.2.2
Active networks reference architecture 16.3 Evolution to 4G
Wireless Networks 16.4 Programmable 4G Mobile Network Architecture
16.5 Cognitive Packet Networks 16.5.1 Adaptation by cognitive
packets 16.5.2 The random neural networks-based algorithms 16.6
Game Theory Models in Cognitive Radio Networks 16.6.1 Cognitive
radio networks as a game 16.7 Biologically Inspired Networks 16.7.1
Bio-analogies 16.7.2 Bionet architecture References 659 659 661 662
662 665 667 670 672 673 675 678 682 682 684 686 17 Network
Deployment 17.1 Cellular Systems with Overlapping Coverage 17.2
Imbedded Microcell in CDMA Macrocell Network 17.2.1 Macrocell and
microcell link budget 17.2.2 Performance example 17.3 Multitier
Wireless Cellular Networks 17.3.1 The network model 17.3.2
Performance example 17.4 Local Multipoint Distribution Service
17.4.1 Interference estimations 17.4.2 Alternating polarization
17.5 Self-Organization in 4G Networks 17.5.1 Motivation 17.5.2
Networks self-organizing technologies References 693 693 698 699
702 703 704 708 709 711 711 713 713 715 717 18 Network Management
18.1 The Simple Network Management Protocol 18.2 Distributed
Network Management 18.3 Mobile Agent-Based Network Management
18.3.1 Mobile agent platform 18.3.2 Mobile agents in multioperator
networks 18.3.3 Integration of routing algorithm and mobile agents
18.4 Ad Hoc Network Management 18.4.1 Heterogeneous environments
18.4.2 Time varying topology 18.4.3 Energy constraints 18.4.4
Network partitioning 18.4.5 Variation of signal quality 18.4.6
Eavesdropping 18.4.7 Ad hoc network management protocol functions
18.4.8 ANMP architecture References 721 721 725 726 728 728 730 735
735 735 736 736 736 736 736 738 743
xvi CONTENTS 19 Network Information Theory 19.1 Effective
Capacity of Advanced Cellular Networks 19.1.1 4G cellular network
system model 19.1.2 The received signal 19.1.3 Multipath channel:
nearfar effect and power control 19.1.4 Multipath channel: pointer
tracking error, rake receiver and interference canceling 19.1.5
Interference canceler modeling: nonlinear multiuser detectors
19.1.6 Approximations 19.1.7 Outage probability 19.2 Capacity of Ad
Hoc Networks 19.2.1 Arbitrary networks 19.2.2 Random networks
19.2.3 Arbitrary networks: an upper bound on transport capacity
19.2.4 Arbitrary networks: lower bound on transport capacity 19.2.5
Random networks: lower bound on throughput capacity 19.3
Information Theory and Network Architectures 19.3.1 Network
architecture 19.3.2 Denition of feasible rate vectors 19.3.3 The
transport capacity 19.3.4 Upper bounds under high attenuation
19.3.5 Multihop and feasible lower bounds under high attenuation
19.3.6 The low-attenuation regime 19.3.7 The Gaussian
multiple-relay channel 19.4 Cooperative Transmission in Wireless
Multihop Ad Hoc Networks 19.4.1 Transmission strategy and error
propagation 19.4.2 OLA ooding algorithm 19.4.3 Simulation
environment 19.5 Network Coding 19.5.1 Max-ow min-cut theorem
(mfmcT) 19.5.2 Achieving the max-ow bound through a generic LCM
19.5.3 The transmission scheme associated with an LCM 19.5.4
Memoryless communication network 19.5.5 Network with memory 19.5.6
Construction of a generic LCM on an acyclic network 19.5.7
Time-invariant LCM and heuristic construction 19.6 Capacity of
Wireless Networks Using MIMO Technology 19.6.1 Capacity metrics
19.7 Capacity of Sensor Networks with Many-to-One Transmissions
19.7.1 Network architecture 19.7.2 Capacity results References 747
747 749 750 752 753 755 757 757 761 762 764 765 768 769 773 773 775
776 776 777 778 779 780 783 784 784 787 788 789 792 793 794 794 795
798 800 805 805 807 809 20 Energy-efcient Wireless Networks 20.1
Energy Cost Function 20.2 Minimum Energy Routing 20.3 Maximizing
Network Lifetime 20.4 Energy-efcient MAC in Sensor Networks 20.4.1
Staggered wakeup schedule References 813 813 815 816 821 821
823
CONTENTS xvii 21 Quality-of-Service Management 21.1 Blind QoS
Assessment System 21.1.1 System modeling 21.2 QoS Provisioning in
WLAN 21.2.1 Contention-based multipolling 21.2.2 Polling efciency
21.3 Dynamic Scheduling on RLC/MAC Layer 21.3.1 DSMC functional
blocks 21.3.2 Calculating the high service rate 21.3.3
Heading-block delay 21.3.4 Interference model 21.3.5 Normal delay
of a newly arrived block 21.3.6 High service rate of a session 21.4
QoS in OFDMA-Based Broadband Wireless Access Systems 21.4.1
Iterative solution 21.4.2 Resource allocation to maximize capacity
21.5 Predictive Flow Control and QoS 21.5.1 Predictive ow control
model References 827 827 829 831 831 832 835 837 838 840 841 841
842 842 846 848 849 850 854 Index 859
Preface to the Second Edition Although the rst edition of the
book was not published long ago, a constant progress in research in
the eld of wireless networks has resulted in a signicant
accumulation of new results that urge the extension and modication
of its content. The major additions in the book are the following
new chapters: Chapter 1: Fundamentals, Chapter 2: Opportunistic
Communications, Chapter 3: Relaying and Mesh Networks, Chapter 4:
Topology Control, Chapter 10: Network Optimization and Chapter 12:
Cognitive Radio Resource Management. OPPORTUNISTIC COMMUNICATIONS
Multiuser diversity is a form of diversity inherent in a wireless
network, provided by independent time-varying channels across the
different users. The diversity benet is exploited by tracking the
channel uctuations of the users and scheduling transmissions to
users when their instantaneous channel quality is near the peak.
The diversity gain increases with the dynamic range of the
uctuations and is thus limited in environments with little
scattering and/or slow fading. In such environments, the multiple
transmit antennas can be used to induce large and fast channel
uctuations so that multiuser diversity can still be exploited. The
scheme can be interpreted as opportunistic beamforming and true
beamforming gains can be achieved when there are sufcient users,
even though very limited channel feedback is needed. Furthermore,
in a cellular system, the scheme plays an additional role of
opportunistic nulling of the interference created on users of
adjacent cells. This chapter discusses the design implications of
implementing this scheme in a wireless system. RELAYING AND MESH
NETWORKS In a wireless network with many sourcedestination pairs,
cooperative transmission by relay nodes has the potential to
improve the overall network performance. In a distributed multihop
mesh/relay network (e.g. wireless ad hoc/sensor network, cellular
multihop network), each node acts as a relay node to forward data
packets from other nodes. These nodes are often energy-limited and
also have limited buffer space. Therefore, efcient power-saving
mechanisms (e.g. sleeping mechanisms) are
xx PREFACE TO THE SECOND EDITION required so that the lifetime
of these nodes can be extended while at the same time the quality
of service (QoS) requirements (e.g. packet delay and packet loss
rate) for the relayed packets can be satised. In Chapter 3, a
queuing analytical framework is presented to study the tradeoffs
between the energy saving and the QoS at a relay node as well as
relaying strategies in cooperative cellular networks. In addition
integrated cellular and ad hoc multicast, which increases multicast
throughput through opportunistic use of ad hoc relays, is also
discussed. NETWORK TOPOLOGY CONTROL Energy efciency and network
capacity are perhaps two of the most important issues in wireless
ad hoc networks and sensor networks. Topology control algorithms
have been proposed to maintain network connectivity while reducing
energy consumption and improving network capacity. The key idea to
topology control is that, instead of transmitting with maximal
power, nodes in a wireless multihop network collaboratively
determine their transmission power and dene the network topology by
forming the proper neighbour relation under certain criteria. The
topology control affects network spatial reuse and contention for
the medium. A number of topology control algorithms have been
proposed to create a power-efcient network topology in wireless
multihop networks with limited mobility. In Chapter 4, we summarize
existing work in this eld. Some of the algorithms require explicit
propagation channel models, while others incur signicant message
exchanges. Their ability to maintain the topology in the case of
mobility is also rather limited. The chapter will discuss the
tradeoffs between these opposing requirements. NETWORK OPTIMIZATION
Network protocols in layered architectures have traditionally been
obtained on an ad hoc basis, and many of the recent crosslayer
designs are also conducted through piecemeal approaches. Network
protocol stacks may instead be systematically analyzed and designed
as distributed solutions to some global optimization problems.
Chapter 10 presents a survey of the recent efforts toward a
systematic understanding of layering as optimization decomposition,
where the overall communication network is modelled by a
generalized network utility maximization problem, where each layer
corresponds to a decomposed subproblem and the interfaces among
layers are quantied as functions of the optimization variables
coordinating the subproblems. There can be many alternative
decompositions, leading to a choice of different layering
architectures. This chapter will survey the current status of
horizontal decomposition into distributed computation and vertical
decomposition into functional modules such as congestion control,
routing, scheduling, random access, power control and channel
coding. Key results are summarized and open issues discussed.
Through case studies, it is illustrated how layering as
optimization decomposition provides a common language to
modularization, a unifying, top-down approach to design protocol
stacks and a mathematical theory of network architectures.
COGNITIVE RADIO RESOURCE MANAGEMENT Network optimization, including
radio resource management, discussed in Chapter 10, provides
algorithms that optimize system performance dened by a given
utility function. In Chapter 12, we present suboptimum solutions
for resource management that include high level of cognition and
cooperation to mitigate intercell interference. An important
segment of this topic dealing with the
PREFACE TO THE SECOND EDITION xxi exible spectra sharing is
covered in another of our books on Advanced Wireless Communications
focusing more on the physical layer, published by John Wiley &
Sons, Ltd in 2007. In addition to the new chapters, which represent
about 40 % of the book, other chapters have been also updated with
latest results. Savo Glisic Beatriz Lorenzo
1 Fundamentals 1.1 4G NETWORKS AND COMPOSITE RADIO ENVIRONMENT
In the wireless communications community we are witnessing more and
more the existence of the composite radio environment (CRE ) and as
a consequence the need for recongurability concepts based on
cognitive, cooperative and opportunistic algorithms. The CRE
assumes that different radio networks can be cooperating components
in a heterogeneous wireless access infrastructure, through which
network providers can more efciently achieve the required capacity
and quality of service (QoS) levels. Recongurability enables
terminals and network elements dynamically to select and adapt to
the most appropriate radio access technologies for handling
conditions encountered in specic service area regions and time
zones of the day. Both concepts pose new requirements on the
management of wireless systems. Nowadays, a multiplicity of radio
access technology (RAT) standards are used in wireless
communications. As shown in Figure 1.1, these technologies can be
roughly categorized into four sets: Cellular networks that include
second-generation (2G) mobile systems, such as Global System for
Mobile Communications (GSM) [1], and their evolutions, often called
2.5G systems, such as enhanced digital GSM evolution (EDGE),
General Packet Radio Service (GPRS) [2] and IS 136 in the US. These
systems are based on TDMA technology. Third-generation (3G) mobile
networks, known as Universal Mobile Telecommunications Systems
(UMTS) (WCDMA and cdma2000) [3] are based on CDMA technology that
provides up to 2 Mbit/s. Long-term evolution (LTE) [412] of these
systems is expected to evolve into a 4G system providing up to 100
Mbit/s on the uplink and up to 1 Gbit/s on the downlink. The
solutions will be based on a combination of multicarrier and
spacetime signal formats. The network architectures include macro,
micro and pico cellular networks and home (HAN) and personal area
networks (PAN). Broadband radio access networks (BRANs) [13] or
wireless local area networks (WLANs) [14] which are expected to
provide up to 1 Gbit/s in 4G. These technologies are based on OFDMA
and spacetime coding. Digital video broadcasting (DVB) [15] and
satellite communications. Ad hoc and sensor networks with emerging
applications. Advanced Wireless Technologies: Cognitive,
Cooperative & Opportunistic 4G Technology Second Edition Savo
G. Glisic C 2009 John Wiley & Sons, Ltd.
2 FUNDAMENTALS Sensor networks (self configur ation) PLMN PSTN
Ad hoc networks IP Network (mobile user agents) Private Network
Cellular multihop network macro/micro/ Pico/PAN/BAN Space-time
frequency coding (100Mb) Cellular multihop network Access Network
Reconfiguration & Dynamic Spectra Allocation BRAN/ WLAN/mesh
Access Space-timefrequency coding, (1Gbit) DVB satellite
Reconfigurable Mobile Terminals (Cognitive, Cooperative and
Opportunistic) Figure 1.1 Composite radio environment in cognitive,
cooperative and opportunistic 4G networks. In order to increase the
spectral efciency further, besides the spacetime frequency coding
in the physical layer, the new paradigms like cognitive [1620],
cooperative [2132] and opportunistic [3338] solutions will be used.
Although 4G is open for new multiple access schemes, the CRE
concept remains attractive for increasing the service provision
efciency and the exploitation possibilities of the available RATs.
The main assumption is that the different radio networks, GPRS,
UMTS, BRAN/WLAN, DVB and so on, can be components of a
heterogeneous wireless access infrastructure. A network provider
(NP) can own several components of the CR infrastructure (in other
words, can own licenses for deploying and operating different
RATs), and can also cooperate with afliated NPs. In any case, an NP
can rely on several alternate radio networks and technologies, for
achieving the required capacity and QoS levels, in a cost-efcient
manner. Users are directed to the most appropriate radio networks
and technologies, at different service area regions and time zones
of the day, based on prole requirements and network performance
criteria. The various RATs are thus used in a
4G NETWORKS AND COMPOSITE RADIO ENVIRONMENT 3 complementary
manner rather than competing each other. Even nowadays a mobile
handset can make a handoff between different RATs. The deployment
of CRE systems can be facilitated by the recongurability concept,
which is an evolution of a software-dened radio [39, 40]. The CRE
requires terminals that are able to work with different RATs, and
the existence of multiple radio networks offering alternate
wireless access capabilities to service area regions.
Recongurability supports the CRE concept by providing essential
technologies that enable terminals and network elements dynamically
(transparently and securely) to select and adapt to the set of RATs
that are most appropriate for the conditions encountered in specic
service area regions and time zones of the day. According to the
recongurability concept, RAT selection is not restricted to those
that are pre-installed in the network element. In fact, the
required software components can be dynamically downloaded,
installed and validated. This makes it different from the static
paradigm regarding the capabilities of terminals and network
elements. The networks provide wireless access to IP (Internet
protocols)-based applications and service continuity in the light
of intrasystem mobility. Integration of the network segments in the
CR infrastructure is achieved through the management system for the
CRE (MS-CRE) component attached to each network. The management
system in each network manages a specic radio technology; however,
the platforms can cooperate. The xed (core and backbone) network
will consist of public and private segments based on IPv4- and
IPv6-based infrastructures. A mobile IP (MIP) will enable the
maintenance of IP-level connectivity regardless of the likely
changes in the underlying radio technologies used that will be
imposed by the CRE concept. Figures 1.2 and 1.3 depict the
architecture of a terminal that is capable of operating in a CRE
context. The terminals include software and hardware components
(layer 1 and 2 functionalities) for operating with different
systems. The higher protocol layers, in accordance with their peer
entities in the network, support continuous access to IP-based
applications. Different protocol busters can further enhance the
efciency of the protocol stack. There is a need to provide the best
possible IP performance over wireless links, including legacy
systems. Within the performance implications of link
characteristics (PILC) of the IETF group, the concept of a
performance-enhancing proxy Terminal management system Network
discovery support Network selection Mobility management intersystem
(vertical) handover QoS monitoring Profile management user
preferences, terminal characteristics Application Enhanced for TMS
interactions and Information flow synchronization Transport layer
TCP/UDP Network layer protocol boosters & conversion IP Mobile
IP bandwidth reasignment GPRS support protocol Layers 2/1 UMTS
support protocol Layers 2/1 WLAN/BRAN Support protocol Layers 2/1
DVB-T Support protocol Layers 2/1 Figure 1.2 Architecture of a
terminal that operates in a composite radio environment.
4 FUNDAMENTALS Application Enhanced for TMS interactions and
Information flow synchronization Terminal management System Network
discovery support Network selection Mobility management intersystem
(vertical) handovers QoS monitoring Profile management
Functionality for software download, instalation, validation
Security, fault/error recovery Transport layer TCP/UDP protocol
boosters & conversion Network layer IP, Mobile IP Active
configurations Repository Reconfigurable modem Interface
Reconfiguration commands Monitoring information Software components
for communication through The selected RATs bandwidth reasignment
Rat-specific and generic software components an parameters Figure
1.3 Architecture of a terminal that operates in the recongurability
context. (PEP) [4144] has been chosen to refer to a set of methods
used to improve the performance of Internet protocols on network
paths where native TCP/IP performance is degraded due to
characteristics of a link. Different types of PEPs, depending on
their basic functioning, are also distinguished. Some of them try
to compensate for the poor performance by modifying the protocols
themselves. In contrast, a symmetric/asymmetric boosting approach,
transparent to the upper layers, is often both more efcient and
exible. A common framework to house a number of different protocol
boosters provides high exibility, as it may adapt to both the
characteristics of the trafc being delivered and the particular
conditions of the links. In this sense, a control plane for easing
the required information sharing (cross-layer communication and
congurability) is needed. Furthermore, another requirement comes
from the appearance of multihop communications, as PEPs have been
traditionally used over the last hop, so they should be adapted to
the multihop scenario. Most communications networks are subject to
time and regional variations in trafc demands, which lead to
variations in the degree to which the spectrum is utilized.
Therefore, a services radio spectrum can be underused at certain
times or geographical areas, while another service may experience a
shortage at the same time/place. Given the high economic value
placed on the radio spectrum and the importance of spectrum
efciency, it is clear that wastage of radio spectrum must be
avoided. These issues provide the motivation for a scheme called
dynamic spectrum allocation (DSA), which aims to manage the
spectrum utilized by a converged radio system and share it between
participating radio networks over space and time to increase
overall spectrum efciency, as shown in Figures 1.4 and 1.5.
Composite radio systems and recongurability, discussed above, are
potential enablers of DSA systems. Composite radio systems allow
seamless delivery of services through the most appropriate
4G NETWORKS AND COMPOSITE RADIO ENVIRONMENT Time or region RAN3
RAN1 RAN3 RAN2 RAN3 RAN3 RAN2 RAN2 RAN1 RAN2 RAN1 RAN3 RAN3 Time or
region RAN1 RAN1 RAN2 RAN1 RAN3 RAN2 RAN1 RAN1 RAN1 RAN1 RAN1
Frequency RAN2 RAN2 RAN2 RAN2 RAN1 RAN2 RAN1 RAN2 RAN2 RAN1 RAN1
RAN2 RAN1 RAN2 RAN2 RAN1 Fragmented Frequency RAN2 RAN1 Frequency
RAN2 Contiguous RAN1 Fixed 5 Time or region Figure 1.4 Fixed
spectrum allocation compared to contiguous and fragmented DSA. 2483
2400 2200 1900 1880 1710 960 880 854 470 230 217 WLAN UMTS GSM GSM
Analog TV and DVB-T DAB Contiguous DSA (a) Contiguous DSA WLAN
Analog TV and DVB-T GSM GSM UMTS WLAN Analog TV and DVB-T DAB
Fragmented DSA (b) Contiguous DSA WLAN GSM UMTS WLAN GSM GSM UMTS
Analog TV and DVB-T DAB Fragmented DSA (c) Figure 1.5 DSA operation
congurations: (a) static (current spectrum allocations); (b)
continuous DSA operations; (c) discrete DSA operations.
6 FUNDAMENTALS access network, and close network cooperation
can facilitate the sharing not only of services but also of
spectrum. Recongurability is also a very important issue, since
with a DSA system a radio access network could potentially be
allocated any frequency at any time in any location. It should be
noted that the application layer is enhanced with the means to
synchronize various information streams of the same application,
which could be transported simultaneously over different RATs. The
terminal management system (TMS) is essential for providing
functionality that exploits the CR environment. On the
user/terminal side, the main focus is on the determination of the
networks that provide, in a cost-efcient manner, the best QoS
levels for the set of active applications. A rst requirement is
that the MS-CRE should exploit the capabilities of the CR
infrastructure. This can be done in a reactive or proactive manner.
Reactively, the MS-CRE reacts to new service area conditions, such
as the unexpected emergence of hot spots. Proactively, the
management system can anticipate changes in the demand pattern.
Such situations can be alleviated by using alternate components of
the CR infrastructure to achieve the required capacity and QoS
levels. The second requirement is that the MS-CRE should provide
resource brokerage functionality to enable the cooperation of the
networks of the CR infrastructure. Finally, parts of the MS-CRE
should be capable of directing users to the most appropriate
networks of the CR infrastructure, where they will obtain services
efciently in terms of cost and QoS. To achieve the above
requirements the MS architecture shown in Figure 1.6 is required.
The architecture consists of three main logical entities:
Monitoring, service-level information and resource brokerage
(MSRB). Resource management strategies (RMS). Session managers
(SMs). Short-term operation Mid-term operation MS-CRE Profile and
service-level information Resource brokerage Session manager
Management Plane interface Service configuration traffic
distribution Netwotk configuration Status monitoring MSRB
Management Plane interface Mobile terminal Managed network
(component of CR infrastructure) Legacy element and network
management systems User and control plane interface Figure 1.6
Architecture of the MS-CRE. RMS Management Plane interface
PROTOCOL BOOSTERS Session manager MSRB RMS 7 MS-CRE 1.
Identification of new condition in service area 2. Extraction of
status of Network and of SLAs 3a. Offer request 3b. Offer request
3c. Offer request 4a. Optimization request 4b. Determination of new
service provision pattern (QoS levels, traffic distribution to
networks) Computation of Tentative reconfigurations 4c. Reply 5.
Solution acceptance phase. Reconfiguration of managed Network and
managed components Figure 1.7 MS-CRE operation scenario. The MSRB
entity identies the triggers (events) that should be handled by the
MS-CRE and provides corresponding auxiliary (supporting)
functionality. The RMS entity provides the necessary optimization
functionality. The SM entity is in charge of interacting with the
active subscribed users/terminals. The operation steps and
cooperation of the RMS components are shown in Figures 1.7 and 1.8,
respectively. In order to gain an insight into the scope and range
of possible recongurations, we review the network and protocol
stack architectures of the basic CRE components as indicated in
Figure 1.1. 1.2 PROTOCOL BOOSTERS As pointed out in Figure 1.2, an
element of the reconguration in 4G networks are protocol boosters.
A protocol booster is a software or hardware module that
transparently improves protocol performance. The booster can reside
anywhere in the network or end systems, and may operate
independently (one-element booster) or in cooperation with other
protocol boosters (multielement booster). Protocol boosters provide
an architectural alternative to existing protocol adaptation
techniques, such as protocol conversion. A protocol booster is a
supporting agent that by itself is not a protocol. It may add,
delete or delay protocol messages, but never originates, terminates
or converts that protocol. A multielement protocol booster may dene
new protocol messages to exchange among themselves, but these
protocols are originated and terminated by protocol booster
elements, and are not visible or
8 FUNDAMENTALS MSRB Service configuration traffic distribution
Network configuration 1. Optimization request 2. Service
configuration and traffic distribution: Allocation to QoS and
networks 3a. Request for checking the feasibility of solution 3b.
Computation of Tentative network reconfiguration 3c. Reply on
feasibility of solution 4. Selection of best Feasible solution 5.
Reply 6. Solution acceptance phase 7. Network configuration Figure
1.8 Cooperation of the RMS components. Protocol messages Host X
Booster A Booster B Host Y Booster messages Figure 1.9 Two-element
booster. meaningful external to the booster. Figure 1.9 shows the
information ow in a generic two-element booster. A protocol booster
is transparent to the protocol being boosted. Thus, the elimination
of a protocol booster will not prevent end-to-end communication, as
would, for example, the removal of one end of a conversion (e.g. a
TCP/IP header compression unit). In what follows we will present
examples of protocol busters.
PROTOCOL BOOSTERS 9 1.2.1 One-element error detection booster
for UDP UDP has an optional 16-bit checksum eld in the header. If
it contains the value zero, it means that the checksum was not
computed by the source. Computing this checksum may be wasteful on
a reliable LAN. On the other hand, if errors are possible, the
checksum greatly improves data integrity. A transmitter sending
data does not compute a checksum for either local or remote
destinations. For reliable local communication, this saves the
checksum computation (at the source and destination). For wide-area
communication, the single-element error detection booster computes
the checksum and puts it into the UDP header. The booster could be
located either in the source host (below the level of UDP) or in a
gateway machine. 1.2.2 One-element ACK compression booster for TCP
On a system with asymmetric channel speeds, such as broadcast
satellite, the forward (data) channel may be considerably faster
than the return (ACK) channel. On such a system, many TCP ACKs may
build up in a queue, increasing round-trip time and thus reducing
the transmission rate for a given TCP window size. The nature of
TCPs cumulative ACKs means that any ACK acknowledges at least as
many bytes of data as any earlier ACK. Consequently, if several
ACKs are in a queue, it is necessary to keep only the ACK that has
arrived most recently. A simple ACK compression booster could
ensure that only a single ACK exists in the queue for each TCP
connection. (A more sophisticated ACK compression booster allows
some duplicate ACKs to pass, allowing the TCP transmitter to get a
better picture of network congestion.) The booster increases the
protocol performance because it reduces the ACK latency and allows
faster transmission for a given window size. 1.2.3 One-element
congestion control booster for TCP Congestion control reduces
buffer overow loss by reducing the transmission rate at the source
when the network is congested. A TCP transmitter deduces
information about network congestion by examining acknowledgments
(ACKs) sent by the TCP receiver. If the transmitter sees several
ACKs with the same sequence number, then it assumes that network
congestion caused a loss of data messages. If congestion is noted
in a subnet, then a congestion control booster could articially
produce duplicate ACKs. The TCP receiver would think that data
messages have been lost because of congestion, and would reduce its
window size, thus reducing the amount of data it injects into the
network. 1.2.4 One-element ARQ booster for TCP TCP uses ARQ to
retransmit data unacknowledged by the receiver when a packet loss
is suspected, such as after a retransmission timeout expires. If we
assume the network of Figure 1.9 (except that Booster B does not
exist), then an ARQ booster for TCP will: (a) cache packets from
Host Y; (b) if it sees a duplicate acknowledgment arrive from Host
X and it has the next packet in the cache; then it deletes the
acknowledgment and retransmits the next packet (because a packet
must have been lost between the booster and Host X); and (c) delete
packets retransmitted from Host Y that have been acknowledged by
Host X. The ARQ booster improves performance by shortening the
retransmission path. A typical application would be if Host X were
on a wireless network and the booster were on the interface between
the wireless and wireline networks.
10 FUNDAMENTALS 1.2.5 A forward erasure correction booster for
IP or TCP For many real-time and multicast applications, forward
error correction coding is desirable. The two-element FZC booster
uses a packet forward error correction code and erasure decoding.
The FZC booster at the transmitter side of the network adds parity
packets. The FZC booster at the receiver side removes the parity
packets and regenerates missing data packets. The FZC booster can
be applied between any two points in a network (including the end
systems). If applied to an IP, then a sequence number booster adds
sequence number information to the data packets before the rst FZC
booster. If applied to TCP (or any protocol with sequence number
information), then the FZC booster can be more efcient because: (1)
it does not need to add sequence numbers and (2) it could add new
parity information on TCP retransmissions (rather than repeating
the same parities). At the receiver side, the FZC booster could
combine information from multiple TCP retransmissions for FZC
decoding. 1.2.6 Two-element jitter control booster for IP For
real-time communication, we may be interested in bounding the
amount of jitter that occurs in the network. A jitter control
booster can be used to reduce jitter at the expense of increased
latency. At the rst booster element, timestamps are generated for
each data message that passes. These (a) (b) Figure 1.10
Three-dimensional amplitude patterns of a two-element uniform
amplitude array for d D 2, directioned towards (a) 0 D 0 , (b) 0 D
60 .
GREEN WIRELESS NETWORKS (a) (c) (b) 11 (d) Figure 1.11
Three-dimensional amplitude patterns of a ten-element uniform
amplitude array for d D =4, directioned towards (a) 0 D 0 , (b) 0 D
30 , (c) 0 D 60 , (d) 0 D 90 . timestamps are transmitted to the
second booster element, which delays messages and attempts to
reproduce the intermessage interval that was measured by the rst
booster element. 1.2.7 Two-element selective ARQ booster for IP or
TCP For links with signicant error rates using a selective ARQ
protocol (with selective acknowledgment and selective
retransmission) can signicantly improve the efciency compared to
using TCPs ARQ (with cumulative acknowledgment and possibly
go-back-N retransmission). The two-element ARQ booster uses a
selective ARQ booster to supplement TCP by: (1) caching packets in
the upstream booster, (2) sending negative acknowledgments when
gaps are detected in the downstream booster and (3) selectively
retransmitting the packets requested in the negative
acknowledgments (if they are in the cache). 1.3 GREEN WIRELESS
NETWORKS 4G wireless networks might be using a spatial notching
(angle ) to suppress completely antenna radiation towards the user,
as illustrated in Figures 1.10 and 1.11. These solutions will be
referred to as green wireless networks for obvious reasons. In
order to ensure the connectivity in the case when the antenna lobe
is not directed towards the access point, a multihop communication,
with the possibility of relaying, is required. In addition, to
reduce the overall transmit power a cooperative transmit diversity,
discussed in Section 19.4, and adaptive MAC protocol, discussed in
Chapter 5 can be used. REFERENCES [1] M. Mouly and M.-B. Pautet,
The GSM System for Mobile Communications, Palaiseau, France,
1992.
12 FUNDAMENTALS [2] R. Kalden, I. Meirick and M. Meyer,
Wireless Internet access based on GPRS, IEEE Pers. Commun., vol. 7,
no. 2, April 2000, pp. 818. [3] 3rd Generation Partnership Project
(3GPP), http://www.3gpp.org. [4] P. Mogensen, Wei Na, I. Z. Kovacs,
F. Frederiksen, A. Pokhariyal, K. I. Pedersen, T. Kolding, K. Hugl
and M. Kuusela, LTE capacity compared to the Shannon bound, in
Vehicular Technology Conf., 2225 April 2007, pp. 12341238. [5] H.
Holma, A. Toskala, K. Ranta-aho, and J. Pirskanen, High-speed
packet access evolution in 3GPP Release 7 [Topics in Radio
Communications], IEEE Commun. Mag., vol. 45, no. 12, December 2007,
pp. 2935. [6] A. Hoikkanen, A techno-cconomic analysis of 3G
long-term evolution for broadband access, in Conf. on Telecommun.
Techno-Economics, 1415 June 2007, pp. 17. [7] A. Racz, A. Temesvary
and N. Reider, Handover performance in 3GPP long term evolution
(LTE) systems, in Mobile and Wireless Commun. Summit, 16th IST, 15
July 2007, pp. 15. [8] M. Valkama, L. Anttila and M. Renfors, Some
radio implementation challenges in 3G-LTE context, in Signal
Processing Advances in Wireless Commun., IEEE 8th Workshop on SPAWC
2007, 1720 June 2007, pp. 15. [9] J. J. Sanchez, D.
Morales-Jimenez, G. Gomez and J. T. Enbrambasaguas, Physical layer
performance of long term evolution in cellular technology, in 16th
IST Mobile and Wireless Commun. Summit, 15 July 2007, pp. 15. [10]
J. Berkmann, C. Carbonelli, F. Dietrich, C. Drewes and Wen Xu, On
3G LTE terminal implementation standards, algorithms, complexities
and challenges, in International Wireless Commun. and Mobile
Computing Conf., IWCMC 08, 68 August 2008, pp. 970975. [11] C.
Spiegel, J. Berkmann, Zijian Bai, T. Scholand and C. Drewes, MIMO
schemes in UTRA LTE, a comparison, in IEEE Vehicular Technology
Conf., 1114 May 2008, pp. 22282232. [12] S.-E. Elayoubi, O. Ben
Haddada and B. Fourestie, Performance evaluation of frequency
planning schemes in OFDMA-based networks, IEEE Trans. Wireless
Commun., vol. 7, no. 5, part 1, May 2008, pp. 16231633. [13] J.
Khun-Jush et al., HiperLAN2: broadband wireless communications at 5
GHz, IEEE Commun. Mag., vol. 40, no. 6, June 2002, pp. 130137. [14]
U. Varshney, The status and future of 802.11-based WLANs, IEEE
Comp., vol. 36, no. 6, June 2003, pp. 102105. [15] Digital Video
Broadcasting (DVB), http://www.dvb.org, January 2002. [16] V.
Stavroulaki, P. Demestichas, A. Katidiotis and D. Petromanolakis,
Evolution in equipment management concepts: from recongurable to
cognitive wireless terminals, in 16th IST Mobile and Wireless
Commun. Summit, 15 July 2007, pp. 15. [17] R. Muraleedharan and L.
A. Osadciw, Increasing QoS and security in 4G networks using
cognitive intelligence, in IEEE Globecom Workshops, 2630 November
2007, pp. 16. [18] I. F. Akyildiz, W.-Y. Lee, M. C. Vuran and S.
Mohanty, Next generation/dynamic spectrum access/cognitive radio
wireless networks: a survey, Computer Networks J., Elsevier, vol.
50, September 2006, pp. 21272159. [19] M. Muck, D. Bourse, K.
Moessner, N. Alonistioti, P. Demestichas, E. Nicollet, E.
Buracchini, D. Bateman, Z. Boudis, E. Patouni, V. Stavroulaki, A.
Trogolo and P. Goria, End-to-end recongurability in heterogeneous
wireless systems software and cognitive radio solutions enriched by
policy- and context-based decision making, in 16th IST Mobile and
Wireless Commun. Summit, 15 July 2007, pp. 15. [20] B. Aazhang, J.
Lilleberg and G. Middleton, Spectrum sharing in a cellular system,
in 2004 IEEE Eighth Int. Symp. on Spread Spectrum Techniques and
Applications, 30 August2 September 2004, pp. 355359. [21] K.
Doppler, A. Osseiran, M. Wodczak and P. Rost, On the integration of
cooperative relaying into the WINNER system concept, in 16th IST
Mobile and Wireless Commun. Summit, 15 July 2007, pp. 15.
REFERENCES 13 [22] Qian Zhang, Qing Chen, Fan Yang, Xuemin Shen
and Zhisheng Niu, Cooperative and opportunistic transmission for
wireless ad hoc networks, IEEE Network , vol. 21, no. 1, January
February 2007, pp. 1420. [23] M. Katz and F. H. P. Fitzek,
Cooperative techniques and principles enabling future 4G wireless
networks, in EUROCON 2005, The Int. Conf. on Computer as a Tool,
vol. 1, 2005, pp. 2124. [24] H. Paloheimo, J. Manner, J. Nieminen
and A. Yla-Jaaski, Challenges in packet scheduling in 4G wireless
networks, in 2006 IEEE 17th Int. Symp. on Personal, Indoor and
Mobile Radio Commun., September 2006, pp. 16. [25] Carlos Leonel
Flores Mayorga,, Francescantonio della Rosa, Satya Ardhy Wardana,
Gianluca Simone,, Marie Claire Naima Raynal, Joao Figueiras and
Simone Frattasi, Cooperative positioning techniques for mobile
localization in 4G cellular networks, in IEEE Int. Conf. on
Pervasive Services, 1520 July 2007, pp. 3944. [26] D. Niyato and E.
Hossain, A cooperative game framework for bandwidth allocation in
4G heterogeneous wireless networks, in 2006 IEEE Int. Conf. on
Commun., vol. 9, June 2006, pp. 43574362. [27] M. Dohler, D.-E.
Meddour, S.-M. Senouci and A. Saadani, Cooperation in 4G hype or
ripe?, in IEEE Technology and Society Mag., vol. 27, no. 1, Spring
2008, pp. 1317. [28] S. Frattasi, M. Monti and R. Prasad, A
cooperative localization scheme for 4G wireless communications, in
2006 IEEE Radio and Wireless Symp., 1719 January 2006, pp. 287290.
[29] C. Politis, T. Oda, S. Dixit, A. Schieder, H.-Y. Lach, M. I.
Smirnov, S. Uskela and R. Tafazolli, Cooperative networks for the
future wireless world, IEEE Commun. Mag., vol. 42, no. 9, September
2004, pp. 7079. [30] V. Marques, R. L. Aguiar, C. Garcia, J. I.
Moreno, C. Beaujean, E. Melin and M. Liebsch, An IP-based QoS
architecture for 4G operator scenarios, IEEE Wireless Commun. [see
also IEEE Pers. Commun.], vol. 10, no. 3, June 2003, pp. 5462. [31]
Luan Huang, Kar Ann Chew and R. Tafazolli, Network selection for
one-to-many services in 3G-broadcasting cooperative networks, in
2005 IEEE 61st Vehicular Technology Conf., vol. 5, 30 May1 June
2005, pp. 29993003. [32] S. M. S. Masajedian and H. Khoshbin,
Cooperative location management method in next generation cellular
networks, in Ninth Int. Symp. on Computers and Communications, ISCC
2004, vol. 1, 28 June1 July 2004, pp. 525530. [33] S. Sorrentino
and U. Spagnolini, A predictive opportunistic scheduler for 4G
wireless systems, in 16th IST Mobile and Wireless Commun. Summit,
15 July 2007, pp. 15. [34] A. K. F. Khattab and K. M. F. Elsayed,
Opportunistic scheduling of delay sensitive trafc in OFDMA-based
wireless networks, in Int. Symp. on a World of Wireless, Mobile and
Multimedia Networks, WoWMoM 2006, 2629 June 2006, 10 pp. [35] Yu
Wu, Ji-bo Wei, Yong Xi, Byung-Seo Kim and Sung Won Kim,
Opportunistic scheduling with statistical fairness guarantee in
wireless networks, in IEEE 18th Int. Symp. on Personal, Indoor and
Mobile Radio Commun., PIMRC 2007, 37 September 2007, pp. 15. [36]
S. Mangold, Zhun Zhong, K. Challapali and Chun-Ting Chou, Spectrum
agile radio: radio resource measurements for opportunistic spectrum
usage, in IEEE Global Telecommun. Conf. 2004, GLOBECOM 04, vol. 6,
29 November3 December 2004, pp. 34673471. [37] S. Sanayei, A.
Nosratinia and N. Aldhahir, Opportunistic dynamic subchannel
allocation in multiuser OFDM networks with limited feedback, in
IEEE Information Theory Workshop, 2004, 2429 October 2004, pp.
182186. [38] W. Ajib and D. Haccoun, An overview of scheduling
algorithms in MIMO-based fourthgeneration wireless systems, IEEE
Network , vol. 19, no. 5, SeptemberOctober 2005, pp. 4348. [39] S.
Glisic, Advanced Wireless Communications: 4G Cognitive and
Cooperative Broadband Technology, 2nd edition. John Wiley &
Sons, Ltd: Chichester, London, 2007.
14 FUNDAMENTALS [40] J. Mitola III and G. Maguire Jr, Cognitive
radio: making software radios more personal, IEEE Pers. Commun.,
vol. 6, no. 4, August 1999, pp. 1318. [41] J. Border et al.,
Performance enhancing proxies intended to mitigate link-related
degradations, RFC 3135, June 2001. [42] D. C. Feldmeier et al.,
Protocol boosters, IEEE J. Selected Areas Commun., vol. 16, no. 3,
April 1998, pp. 437444. [43] M. Garca et al., An experimental study
of snoop TCP performance over the IEEE 802.11b WLAN, in 5th Int.
Symp. on Wireless Personal Multimedia Commun., vol. III, Honolulu,
HA, October 2002, pp. 10681072. [44] L. Mu oz et al., Optimizing
Internet ows over IEEE 802.11b wireless local area networks: n a
performance enhancing proxy based on forward error correction, IEEE
Commun. Mag., vol. 39, no. 12, December 2001, pp. 6067.
2 Opportunistic Communications As pointed out in Chapter 1,
opportunistic signaling will be used in 4G networks to increase
further the spectral efciency of these systems. In this chapter we
discuss a number of different solutions that are based on that
principle. 2.1 MULTIUSER DIVERSITY Multiuser diversity is provided
in wireless networks by independent time-varying channels across
the different users. The diversity benet is exploited by tracking
the channel uctuations of the users and scheduling transmissions to
users when their instantaneous channel quality is highest. The
diversity gain increases with the dynamic range of the uctuations
and is thus limited in environments with slow fading. In such
environments, multiple-transmit antennas can be used to induce
large and fast channel uctuations so that multiuser diversity can
be improved [1]. The scheme can be interpreted as opportunistic
beamforming and beamforming gains can be achieved when there are
sufcient users. In a cellular system, the scheme plays an
additional role of opportunistic nulling of the interference
created on users of adjacent cells. Let us assume a simple model of
the downlink of a cellular wireless communication system with a
base station (transmitter) having a single antenna communicating
with K users (receivers). The time-slotted block-fading channel
model in baseband is given by yk (t) D h k (t)x(t) C zk (t); T k D
1; 2; : : : ; K (2.1) T In Equation (2.1), x(t) 2 C is the vector
of T transmitted symbols, yk (t) 2 C is the vector of T received
symbols of user k, h k (t) 2 C is the fading channel gain from the
transmitter to receiver k and fzk (t)gt is an independent and
identically distributed (i.i.d.) sequence of zero mean
circularsymmetric Gaussian random vectors CN (0; 2 IT ). Here we
assume that the channel is constant over time slots of length T
samples and that the transmit power level is P D const at all
times, i.e. E[kx(t)k2 ] D P T . Advanced Wireless Technologies:
Cognitive, Cooperative & Opportunistic 4G Technology Second
Edition Savo G. Glisic C 2009 John Wiley & Sons, Ltd.
16 OPPORTUNISTIC COMMUNICATIONS Total Throughput in bps/Hz 2,2
AWGN Channel Rayleigh Fading 2 1,8 1,6 1,4 1,2 1 0,8 0 2 4 6 8 10
Number of Users 12 14 Figure 2.1 Sum capacity of two channels,
Rayleigh fading and AWGN, with average SNR D 0 dB. We can view this
downlink channel as a set of parallel Gaussian channels, one for
each fading state. The sum capacity of this channel, dened by the
maximum achievable sum of long-term average data rates transmitted
to all the users, can be achieved by a simple time division
multiple access (TDMA) strategy: at each fading state, transmit to
the user with the strongest channel [2]. The sum capacity of the
downlink channel is presented in Figure 2.1, as a function of the
number of users, for the case when users undergo independent
Rayleigh fading with average received signal-to-noise ratio (SNR) D
0 dB. One can see that the sum capacity increases with the number
of users in the system. On the other hand, the sum capacity of a
nonfaded downlink channel, where each user has a xed additive white
Gaussian noise (AWGN) channel with SNR = 0 dB, is constant
irrespective of the number of users. In a system with many users
with independently varying channels, it is likely that at any time
there is a user with a channel much stronger than the average SNR.
By transmitting to users with strong channels at all times, the
overall spectral efciency of the system can be made high,
signicantly higher than that of a nonfaded channel with the same
average SNR. In order to exploit such multiuser diversity it is
necessary that: Each receiver should track its own channel SNR
through, say, common downlink pilot, and feed back the
instantaneous channel quality to the base station. The base station
has the ability to schedule transmissions among the users as well
as to adapt the data rate as a function of the instantaneous
channel quality. 2.2 PROPORTIONAL FAIR SCHEDULING The concept of
multiuser diversity brings about two issues: fairness and delay.
When users fading statistics are the same, the strategy above
maximizes not only the total capacity of the system but also the
throughput of individual users. When the statistics are not
symmetrical, the multiuser diversity concept provides maximum
long-term average throughputs. In practice, there are latency
requirements, in which case the average throughputs and limited
delay is the performance metric of interest. In the sequel the
objective will be to address these issues while at the same time
exploiting the multiuser diversity gain inherent in a system, with
users having independent, uctuating channel conditions. In a
further discussion the feedback of the channel quality of user k in
time slot t
PROPORTIONAL FAIR SCHEDULING 17 Requested rates in bps/Hz 1 0,9
0,8 0,7 0,6 0,5 0,4 0,3 0,2 0 50 100 150 200 250 Time Slots
Requested rates in bps/Hz Figure 2.2 For symmetric channel
statistics of users, the scheduling algorithm reduces to serving
each user with the largest requested rate. 1,1 1 0,9 0,8 0,7 0,6
0,5 0,4 0,3 0,2 0 50 100 150 200 250 Time Slots Figure 2.3 In
general, with asymmetric user channel statistics, the scheduling
algorithm serves each user when it is near its peak within the
latency time scale tc . to the base station will be expressed in
terms of a requested data rate Rk (t). This is the data rate that
the kth users channel can support at the time. The scheduling
algorithm keeps track of the average throughput Tk (t) of each user
in a past window of length tc , and in time slot t the scheduling
algorithm transmits to the user k with the largest Rk (t)=Tk (t)
among all active users in the system. The average throughputs Tk
(t) can be updated using an exponentially weighted low-pass lter Tk
(t C 1) D (1 (1 1=tc )Tk (t) C Rk (t)=tc ; 1=tc )Tk (t); k D k : k
6D k (2.2) The operation of the algorithm is illustrated in Figures
2.2 and 2.3. The sample paths of the requested data rates of two
users are plotted as a function of time slots (each time slot is
1.67 ms). In Figure 2.2, the two users have identical fading
statistics. If the scheduling time scale tc is much larger than the
correlation time scale of the fading dynamics, then by symmetry the
throughput of each user Tk (t) converges to the same quantity. The
scheduling algorithm reduces to always
OPPORTUNISTIC COMMUNICATIONS Average Throughput in bps/Hz 18
2,2 2 1,8 1,6 1,4 1,2 Mobile Fixed Equal time scheduling 1 0,8 0 5
10 15 Number of Users 20 25 Figure 2.4 Multiuser diversity gain in
xed and mobile environments. picking the user with the highest
requested rate. Thus, each user is scheduled when its channel is
good and at the same time the scheduling algorithm is perfectly
fair in the long term. In Figure 2.3, one users channel is much
stronger than the other users on the average, although both
channels uctuate due to multipath fading. Always picking the user
with the highest requested rate means giving all the system
resources to the statistically stronger user and would be unfair.
Under the scheduling algorithm dened by Equation (2.2), users
compete for resources not only based on their requested rates but
after normalization by their respective average throughputs. The
user with the statistically stronger channel will have a higher
average throughput. Thus, the algorithm schedules a user when its
instantaneous channel quality is high relative to its own average
channel condition over the period tc . In other words, data are
transmitted to a user when the channel is near its own peaks.
Multiuser diversity benet can still be exploited because channels
of different users uctuate independently so that if there is a
sufcient number of users in the system, there is likely to be a
user near its peak at any one time. The parameter tc is tied to the
latency time scale of the application. Peaks are dened with respect
to this time scale. If the latency time scale is large, then the
throughput is averaged over a longer time scale and the scheduler
can afford to wait longer before scheduling a user when its channel
hits a really high peak. The theoretical properties of this
scheduling algorithm are further explored later in this chapter. It
will be shown that this algorithm guarantees a fairness property
called proportional fairness. Figure 2.4 gives some insights into
the issues involved in realizing multiuser diversity benets in
practice. The plot shows the total throughput of the downlink under
the proportional fair scheduling algorithm in the following two
simulated environments: Fixed. Users are static but there are
movements of objects around them (2 Hz Rician, def D E direct =E
specular D 5). Mobile. Users move at walking speeds (3 km/h,
Rayleigh). The total throughput increases with the number of users
in both the xed and mobile environments, but the increase is more
dramatic in the mobile case. While the channel fades in both cases,
the dynamic range and the rate of the variations is larger in the
mobile environment than in the xed one. This means that over the
latency time scale (1.67 s in these examples), the peaks of the
channel uctuations are likely to be higher in the mobile
environment, and the peaks are what determines the performance of
the scheduling algorithm. Thus, the inherent multiuser diversity is
more limited in the xed environment.
OPPORTUNISTIC BEAMFORMING 2.3 19 OPPORTUNISTIC BEAMFORMING The
effectiveness of multiuser diversity depends on the rate and
dynamic range of channel uctuations. In environments where the
channel uctuations are small, the multiuser diversity gain can be
increased by inducing faster and larger uctuations. In Reference
[1] multiple-transmit antennas at the base station are used for
these purposes, as illustrated in Figure 2.5. For such a system
with N transmit antennas at the base station, let h nk (t) be the
complex channel gain from antenna n to the kth user in time slot t.
In time slot t, the same block of symbols x(t) is p transmitted
from all of the antennas except that it is multiplied by a complex
number n (t)e jn (t) N at antenna n, for n D 1; : : : ; N , such
that nD1 n (t) D 1, preserving the total transmit power. The
received signal at user k is given by N n (t)e jn (t) h nk (t) x(t)
C zk (t) yk (t) D (2.3) nD1 Thus, the overall channel gain seen by
receiver k is now N n (t)e jn (t) h nk (t) h k (t) :D (2.4) nD1 In
Equation (2.4), the n (t) denote the fractions of power allocated
to each of the transmit antennas and the n (t) the phase shifts
applied at each antenna to the signal. By varying over time n (t)
from 0 to 1 and n (t) from 0 to 2, uctuations in the overall
channel can be induced even if the physical channel gains h nk (t)
do not uctuate much. Each receiver k reports back the value of x
(t) a(t) 1a(t)e jq(t) h2k (t) h1k (t) User k Figure 2.5 The same
signal is transmitted over the two antennas with time-varying phase
and powers.
20 OPPORTUNISTIC COMMUNICATIONS Average Throughput in bps/Hz
2,2 2 1,8 1,6 1,4 Mobile Fixed, Opp, BF Fixed Equal time scheduling
1,2 1 0,8 0 5 10 15 20 25 30 Number of Users Figure 2.6 Amplication
in multiuser diversity gain with opportunistic beamforming in a xed
environment. SNRjh k (t)j2 = 2 of its own channel to the base
station and the base station schedules transmissions to users
accordingly. The rate of variation of fn (t)g and fn (t)g in time
is a design parameter of the system. On one side it should be as
fast as possible to provide full channel uctuations within the
latency time scale of interest. On the other hand, the variation
should be slow enough to allow the channel to be reliably estimated
by the users and the SNR information feedback. Further, the
variation should be slow enough to ensure that the channel seen by
the users does not change abruptly and thus maintains stability of
the channel tracking loop. To illustrate the performance of this
scheme, we now consider the xed environment of Figure 2.5 with two
antennas of equal and constant (over time) power split and phase
rotation over [0; 2 ] (with one complete rotation in 30 ms as in
Reference [1]). Figure 2.6 plots the improved performance as a
function of number of users. This improvement is due to the fact
that the channel is changing faster and the dynamic range of
variation is larger over the time scale of scheduling (1.67 s in
this example). 2.4 OPPORTUNISTIC NULLING IN CELLULAR SYSTEMS For
wide-band cellular systems with full frequency reuse, it is
important to consider the effect of intercell interference on the
performance of the system, particularly in interference-limited
scenarios. In a cellular system, the channel quality of a user is
measured by the signal-to-interference plus noise ratio (SINR). In
a fading environment, the energies in both the received signal and
the received interference uctuate over time. Since the multiuser
diversity scheduling algorithm allocates resources based on the
channel SINR (which depends on both the channel amplitude and the
amplitude of the interference), it automatically exploits both the
uctuations in the energy of the received signal as well as that of
the interference: the algorithm tries to schedule resource to a
user whose instantaneous channel is good and the interference is
weak. Thus, multiuser diversity naturally takes advantage of the
time-varying interference to increase the spatial reuse of the
network. From this point of view, power and phase randomization at
the base station transmit antennas plays an additional role: it
increases not only the amount of uctuations of the received
signal
OPPORTUNISTIC NULLING IN CELLULAR SYSTEMS 21 to the intended
users within the cells, but also the amount of the uctuations of
the interference the base station causes in adjacent cells. Hence,
opportunistic beamforming has a dual benet in an
interference-limited cellular system. In fact, opportunistic
beamforming performs opportunistic nulling simultaneously, while
randomization of power and phase in the transmitted signals from
the antennas allows near-coherent beamforming to some user within
the cell; it will create near nulls at some other user in adjacent
cells. This in effect allows interference avoidance for that user
if it is currently being scheduled. In a slow at fading scenario
under power and phase randomization at all base stations, the
received signal of a typical user being interfered by J adjacent
base stations is given by J g j (t)u j (t) C z(t) y(t) D h(t)x(t) C
(2.5) jD1 where x(t) is the signal of interest, u j (t) is the
interference from the jth base station and z(t) is additive
Gaussian noise. All base stations have the same transmit power P
and N transmit antennas and are performing power and phase
randomization independently; h(t) and g j (t) are the overall
channel gains from the base stations N n (t)e jn (t) h n h(t) :D
(2.6) nD1 N n j (t) D e j g j (t) :D n j (t) gn j (2.7) nD1 where h
n and gn j are the slow fading channel gains to the user from the
nth transmit antenna of the base station of interest and the
interfering base station j, respectively. Averaging over the signal
x(t) and the interference u j (t), the (time-varying) SINR of the
user can be computed to be Pjh(t)j2 SINR(t) D J P (2.8) jg j (t)j2
C 2 jD1 The SINR varies because of both the variations of the
overall gain from the base station of interest as well as those
from the interfering base station. In a system with many other
users, the N proportional fair scheduler will serve this user while
its SINR is at its peak P nD1 jh n (t)j2 = 2 , i.e. when the
received signal is the strongest and the interference is completely
nulled out. Thus, the opportunistic nulling and beamforming
technique has the potential to shift a user from a low-SNR
interference-limited regime to a high-SNR noise-limited regime. How
close the performance of opportunistic beamforming and nulling in a
nite-size system is to this asymptotic limit depends on the
probability that the received signal is near beamformed and all the
interference is near null. In the interference-limited regime when
P= 2 1, the performance depends mainly on the probability of the
latter event. This probability is larger when there are only one or
two base stations contributing most of the interference, as is
typically the case. In contrast, when there is interference from
many base stations, interference averaging occurs and the
probability that the total interference is near null is much
smaller. Interference averaging, which is good for CDMA networks,
is actually unfavorable for the opportunistic scheme described
here, since it reduces the likelihood of the nulling of the
interference and hence the likelihood of the peaks of the SINR. In
a typical cell, there will be a distribution of users, some closer
to the base station and some closer to the cell boundaries. Users
close to the base station are at high SNR and are noise-limited;
the contribution of the intercell interference is relatively small.
These users benet
22 OPPORTUNISTIC COMMUNICATIONS mainly from opportunistic
beamforming (diversity gain plus a 3 dB power gain if there are two
transmit antennas). Users close to the cell boundaries, on the
other hand, are at low SNR and are interference-limited; the
average interference power can be much larger than the background
noise. These users benet both from opportunistic beamforming and
from opportunity nulling of intercell interference. Thus, the
cell-edge users benet more in this system than users in the
interior. This is rather desirable from a system fairness point of
view, as the cell-edge users tend to have poorer service. This
feature is particularly important for a system without soft handoff
(which is difcult to implement in a packet data scheduling system).
To maximize the opportunistic nulling benets, the transmit power at
the base station should be set as large as possible, subject to
regulatory and hardware constraints. 2.5 NETWORK COOPERATION AND
OPPORTUNISTIC COMMUNICATIONS In this section we consider a network
architecture with B cooperating and B noncooperating base stations
(BSs). The cluster of cooperating BSs cooperate to improve the
capacity with fairness to the users in the network. The
noncooperating BSs do not cooperate with BSs in this cluster and
cause interference to the users, which cannot be mitigated. As an
example, in Figure 2.7 [3], there are three BSs (marked with
circles), which cooperate and provide service to the users in the
shaded region. The other BSs cause interference to these users. In
the sequel we focus only on the activities of the cooperating BSs.
During every time slot each coordinating BS selects one beam out of
the set of L beams available to it and services one user. In this
way, the cluster of B cooperating BSs supports B users
simultaneously. The system operates in two steps. In the rst step,
each BS transmits a pilot signal using a certain beam and collects
the SINR reports from the users. The BS then transmits these
reports to a centralized unit which, in the second step, schedules
users based on their current and past SINR reports for data packet
transmission. This process is repeated for the entire time period
over which data are transmitted. The number of users K being
serviced during this time period is assumed constant. Figure 2.7
Three-base intercell coordination scenario.
NETWORK COOPERATION AND OPPORTUNISTIC COMMUNICATIONS 23 In a
given timeslot n during the rst step, each BS transmits a pilot
signal using one of the L beams. The received signal power at user
i from the BS b on the lth beam is Si [n] D P(i;b) jh (i;b;l) [n]j2
, where h (i;b;l) [n] is the channel gain from BS b to the user i
on beam l and P(i;b) is the received signal power at the user i
with the combined effect of path loss and shadowing. The
interference plus noise power is INi [n] D Ii [n] C Noi [n], where
I () is the interference power from the noncooperating BSs and No()
is the thermal noise power. The user measures the received SINR,
(i;b;l) [n] D Si [n]=INi [n] and transmits this SINR value to the
BS. The centralized scheduler collects the SINR reports from all
the BSs and obtains the matrix K B , where the (i; b)th element
represents the SINR feedback by user i on the beam from BS b. In
the second phase the centralized scheduler using these data
opportunistically schedules B users on each of the B BSs, and each
BS then transmits to the scheduled user. The data rate R(i;b;l) , a
BS b, provides a given user i on beam l and is calculated by
R(i;b;l) D log2 1 C Si [n] D log2 1 C B I Ni [n] C (i;b;l) [n] 1 C
2 P(i;b) h (i;b;l) [n] b6Db B (i;b;l) [n] (2.9) b6Db In a scenario
where the BSs cooperate and opportunistically beamform and schedule
users the main steps are [3]: (1) Choose U users to serve: (kU k D
K u ; B < K u < K ) using one of the following criteria: a)
Select users that experience large inter-packet delay, i [n]. b)
Select users who have low short-term throughput Ti [n]. c) Select
users who have the largest i [n]=Ti [n] ratio. (2) The centralized
scheduler using the user-base-beam selection algorithm described
below selects, a subset of B users from the above chosen group of K
u users, and the corresponding BSs and beams on which they may be
serviced. This algorithm generates a set of B triplets where the
triplet (i j ; b j ; l j ) represents the case where user i j is
best served by BS b j on the beam l j ; j 2 f1 Bg: (3) Each BS
generates the above assigned beam and collects the SINR reports
from all the users. All the BSs transmit these reports to the
centralized scheduler. (4) The centralized scheduler using the K B
SINR matrix and the short-term throughput vector T schedules B
users for service. Steps 1,2 and 3 constitute the rst phase of the
system and step 4 constitutes the second phase. In the
user-base-beam selection (UBBS) algorithm, mentioned above, B users
from the initially chosen set of K u users and the indices of the
BSs and the beams on which these users may be serviced are chosen.
The beams must be chosen such that they least interfere with the B
selected users. One possible approach to attain this objective is
through the instantaneous system throughput, which is the sum of
the achievable user rates. For B D 3 we state this formally as arg
max i 1 ;i 2 ;i 3 2f1K u g;b1 ;b2 ;b3 2f1Bg;l1 ;l2 ;l3 2f1Lg f (i;
b; l) (2.10)
24 OPPORTUNISTIC COMMUNICATIONS with i D fi 1 ; i 2 ; i 3 g, b
D fb1 ; b2 ; b3 g, l D fl1 ; l2 ; l3 g and f () is the sum of the
rates of the three users. By using Equation (2.9) we have f (i; b;
l) D log2 1C (i 1 ;b1 ;l1 ) 1C 1C (i 1 ;b2 ;l2 ) C 1C (i 1 ;b3 ;l3
) (i 2 ;b2 ;l2 ) 1C (i 2 ;b1 ;l1 ) C (i 2 ;b3 ;l3 ) (i 3 ;b3 ;l3 )
1C (i 3 ;b1 ;l1 ) C (i 3 ;b2 ;l2 ) In the absence of current SINRs,
the short-term averaged SINRs (i;b;l) can be used instead of in
Equation (2.10). The short-term SINR is computed as (i;b;l) [n] D n
1 tD0 n t 1 (i;b;l) [t]((i;b) [t]; l)a n 1 n t 1 tD0 ((i;b) [t];
l)a (i;b;l) (2.11) where (i;b;l) [t] represents the index of the
beam on which user i was serviced by base station b at time instant
t. Thus, this alternative criterion for selecting the set of
triplets is stated as arg max i 1 ;i 2 ;i 3 2f1K u g;b1 ;b2 ;b3
2f1Bg;l1 ;l2 ;l3 2f1Lg (i; b; l) f (2.12) with (i; b; l) given by
Equation (2.10) using (i;b;l) instead of (i;b;l) . f The
optimization of Equation (2.12) is practically infeasible when B
and L are large. The following, computationally less intensive,
modied UBBS algorithm was proposed in Reference [3] for scenarios
where B and L are large. A group of users U is initially chosen and
arranged in order of preference for service using one of the
criteria stated earlier. The main steps of this modied procedure
are: (1) Choose the rst user i 1 and obtain the BS and beam index
fb1 ; l1 g on which the user i 1 experiences maximum throughput.
Formally these indices are obtained as (b1 ; l1 ) D arg max (i;b;l)
(2.13) b2f1Bg;l2f1Lg 1 D (i1 ;b1 ;l1 ) (2.14) (2) Choose the second
user i 2 from the remainder of the set (U the beam and BS using the
following criterion: arg max i 1 ), and the indices for 2 (i; b;
l); f (2.15) b2 2f1Bg;l2 2f1Lg with i D fi 1 ; i 2 g and b D fb1 ;
b2 g; l D fl1 ; l2 g and 2 (i; b; l) D f 1C (i1 ;b1 ;l1 ) 1 C (i1
;b2 ;l2 ) 1C (i2 ;b2 ;l2 ) 1 C (i2 ;b1 ;l1 ) f 2 D 2 (i; b; l) In
this way the newly chosen user i 2 causes minimal interference to
user i 1 and vice versa. The terms 1 , 2 represent the virtual SINR
of the system as the system is loaded with users and BSs. We also
use the constraint that 2 > 1 to ensure that the addition of a
new user does not decrease the overall virtual SINR of the system.
In the same way, the remaining set of users and the beams on which
they are serviced by the BSs are chosen. When this constraint of
increasing the virtual SINR of the system fails, the beams are
chosen at random. The scheduler obtains the terms (i;b;l) , which
are the SINR reports of the users from the BSs on the beams given
by the UBBS algorithm. Then the rate R(i;b;l) that user i obtains
from BS b using
NETWORK COOPERATION AND OPPORTUNISTIC COMMUNICATIONS 25 beam l
is computed. Using this information, B users are chosen whose
weighted sum of rates is maximum, with the weights given by the
reciprocal of the users short-term throughputs T []. This criterion
is also computationally intensive, so the following alternative is
considered. The users are ordered according to their weighted rates
as (R=T )i1 ;b1 (R=T )i2 ;b2 (R=T )i K ;bk and the rst B users are
chosen while maintaining the constraint that only one user is
scheduled from a BS. 2.5.1 Performance example We will illustrate
the performance of the above methods for two different hexagonal
cellular networks, as shown in Figures 2.7 and 2.8. The number of
coordinating BSs B in these two networks are 3 and 9 respectively.
The hexagonal cells are divided into three 120 sectors and each of
these sectors are covered by a BS with a smart antenna system. The
BSs are located at the centre of each of the hexagonal cells and
the users are distributed randomly in the shaded regions. The
received signal at the user is modelled as in Section 2.5 to take
into account path loss, shadow fading and correlated Rayleigh
fading effects. The path loss is based on the Hata urban
propagation model with a path loss coefcient of 3.5. A lognormal
shadow fading process with zero mean and a standard deviation of 8
dB was used to characterize the variations in the signal due to
environmental clutter. Two different correlated Rayleigh fading
channels corresponding to Doppler rates with user velocities of 1
m/s and 8 m/s are analyzed. The antennas at the BSs consist of a
uniform linear array (ULA) of antenna elements. Each of these
elements is fed equal currents and only their phases are varied.
Eleven such equispaced radiation patterns were generated using a
four-element array. The normalized array factor considered in the
simulations is j A( )j D jsin(N =2)= sin( =2)j=N , where N is the
number of antenna array elements and is the azimuth in radians. The
radiation pattern of each element is shown by the dashed line in
Figure 2.9. The