modeling and dimensioning of mobile...

MODELING ANDDIMENSIONING OFMOBILE NETWORKSFROM GSM TO LTE

Maciej StasiakPoznan University of Technology, Poland

Mariusz GłabowskiPoznan University of Technology, Poland

Arkadiusz WisniewskiOrange, Poland

Piotr ZwierzykowskiPoznan University of Technology, Poland

MODELING ANDDIMENSIONING OFMOBILE NETWORKS

MODELING ANDDIMENSIONING OFMOBILE NETWORKSFROM GSM TO LTE

Maciej StasiakPoznan University of Technology, Poland

Mariusz GłabowskiPoznan University of Technology, Poland

Arkadiusz WisniewskiOrange, Poland

Piotr ZwierzykowskiPoznan University of Technology, Poland

This edition first published 2011

©2011 John Wiley & Sons Ltd.

Registered officeJohn Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, United Kingdom

For details of our global editorial offices, for customer services and for information about how to apply forpermission 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 anyform or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UKCopyright, 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 beavailable in electronic books.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand names andproduct names used in this book are trade names, service marks, trademarks or registered trademarks of theirrespective owners. The publisher is not associated with any product or vendor mentioned in this book. Thispublication is designed to provide accurate and authoritative information in regard to the subject matter covered. It issold on the understanding that the publisher is not engaged in rendering professional services. If professional adviceor other expert assistance is required, the services of a competent professional should be sought.

Library of Congress Cataloging-in-Publication Data

Stasiak, Maciej.Modeling and dimensioning of mobile networks : from GSM to LTE / Maciej Stasiak, Mariusz Głabowski,

Arkadiusz Wisniewski.p. cm.

Includes bibliographical references and index.ISBN 978-0-470-66586-2 (cloth)1. Wireless communication systems. 2. Mobile communication systems. 3. Wireless metropolitan

area networks. 4. Computer networks–Scalabiltiy. 5. Cell phone systems. I. Głabowski, Mariusz.II. Wisniewski, Arkadiusz. III. Title.

TK5103.2.S85 2011004.6–dc22

2010026286

A catalogue record for this book is available from the British Library.

Print ISBN 9780470665862 (H/B)ePDF ISBN: 9780470976043oBook ISBN: 9780470976036ePub ISBN: 9780470975992

Set in 10/12 Times by Thomson Digital, Noida, India

www.wiley.com

Contents

List of Figures xiii

List of Tables xvii

Preface xix

PART I MOBILE NETWORK STANDARDS

1 Global System for Mobile Communications 3

1.1 Introduction 31.2 System Architecture 41.3 Time Structure of the GSM System 71.4 Logical Channels 71.5 High-Speed Circuit Switched Data (HSCSD) 91.6 Packet Transmission based on GPRS 101.7 Packet Transmission based on EDGE 121.8 Traffic Management Mechanisms in Cellular Networks 13

1.8.1 Directed Retry Handover 131.8.2 Traffic Handover 131.8.3 Queuing 14

References 14

2 Universal Mobile Telecommunication System 15

2.1 Introduction 152.2 System Architecture 172.3 Wideband Access with WCDMA Coding and Multiplexing – Essentials 20

2.3.1 Channelization Codes and Scrambling Codes 222.3.2 Bearers in the UMTS System 252.3.3 Frame Structure in the UMTS System 26

2.4 Channels in the WCDMA Radio Interface 262.4.1 Logical Channels 262.4.2 Transport Channels 272.4.3 Physical Channels 28

2.5 Modulation 292.5.1 Modulation in the Downlink 292.5.2 Modulation in the Uplink 30

vi Contents

2.6 Signal Reception Techniques 302.7 Radio Resource Management in the UMTS System 32

2.7.1 Power Control 322.7.2 Handover Control 342.7.3 Call Admission Control 352.7.4 Packet Scheduler 372.7.5 Load Control 38

2.8 High-Speed Packet Data Transmission 382.8.1 High-Speed Downlink Packet Access (HSDPA) 382.8.2 High-Speed Uplink Packet Access (HSUPA) 40

2.9 Services 40References 41

3 Long-Term Evolution 43

3.1 Introduction 433.2 System Architecture 443.3 Transmission Techniques in the LTE System 45

3.3.1 Long-Term Evolution OFDMA in the Downlink Direction 453.3.2 Long-Term Evolution SC-FDMA in the Uplink 483.3.3 Long-Term Evolution MIMO 48

3.4 Channels in the Radio Interface of the LTE System 493.4.1 Long-Term Evolution Logical Channels 503.4.2 Long-Term Evolution Transport Channels 513.4.3 Long-Term Evolution Physical Channels 51

3.5 Radio Resource Management in LTE 523.5.1 Admission Control 523.5.2 Frequency Domain Packet Scheduling 523.5.3 Interference Management and Power Settings 523.5.4 Discontinuous Transmission and Reception (DTX/DRX) 53

References 53

PART II TELETRAFFIC ENGINEERING FOR MOBILE NETWORKS

4 Basic Definitions and Terminology 57

4.1 Introduction 574.2 Call Stream 57

4.2.1 Poisson Stream and its Properties 574.2.2 Mathematical Model of Poisson Stream 58

4.3 Service Stream 614.3.1 Definition 614.3.2 Mathematical Model of Service Stream 61

4.4 Markov Processes 644.4.1 Stochastic Processes 644.4.2 Markov Process as a Call Service Process in the

Full-Availability Group 64

Contents vii

4.5 The Concept of Traffic 694.5.1 Introductory Information 694.5.2 Traffic and Traffic Intensity 704.5.3 Definitions of the Average Intensity of Carried Traffic 704.5.4 Definition of the Average Intensity of Offered Traffic 73

4.6 Quality of Service in Telecommunication Systems 734.6.1 Basic GoS Parameters in Loss Systems 734.6.2 Traffic Load-Carrying Capacity and Traffic Load of Communication

Systems 74References 74

5 Basic Elements of Traffic Engineering used in Mobile Networks 77

5.1 Introduction 775.2 Erlang Model 77

5.2.1 Assumptions of the Model 775.2.2 Diagram of the Service Process 785.2.3 State Equations 785.2.4 Occupancy Probability of Arbitrarily Chosen i Channels in the

Group – Erlang Distribution 795.2.5 Blocking Probability – Erlang Formula 795.2.6 Loss Probability 795.2.7 Occupancy Probability of x Precisely Determined Channels in a

Group – Palm-Jacobaeus Formula 805.2.8 Erlang Tables 805.2.9 Group Conservation Principle 81

5.2.10 Recursive Properties of the Erlang Formula 825.2.11 Traffic Carried by the Full-Availability Groups 825.2.12 Traffic Carried by One Channel of the Full-Availability Group 82

5.3 Engset Model 835.3.1 Assumptions of the Model 835.3.2 Diagram of the Service Process 845.3.3 State Equations 845.3.4 Occupancy Probability of Arbitrarily Chosen i Channels in the

Group – Engset Distribution 845.3.5 Blocking Probability 855.3.6 Loss Probability 855.3.7 Alternative Notation of the Engset Formula 865.3.8 Relationship between the Erlang and Engset Distributions 865.3.9 Occupancy Probability of x Precisely Determined Channels in a

Group 875.3.10 Traffic Carried by the Full-Availability Group 875.3.11 Recursive Properties of the Engset Formula 875.3.12 Commentary to the Average Traffic Intensities 88

5.4 Comments 90References 90

viii Contents

6 Modeling of Systems with Single-Rate Overflow Traffic 91

6.1 Introduction 916.2 Basic Information on Overflow Systems 91

6.2.1 Simplified Classification of Groups in Telecommunication Networks 916.2.2 Alternative Paths 916.2.3 Overflow Traffic 92

6.3 Models of Alternative Groups 936.3.1 Analytic Model of the System with Overflow Traffic 936.3.2 State Equations 946.3.3 Determination of Overflow Traffic Parameters – Riordan Formulas 966.3.4 Comment on Riordan Formulas 98

6.4 Equivalent Groups 986.4.1 Formulation of the Problem of Dimensioning Alternative Groups 986.4.2 Equivalent Random Traffic (ERT) Method 996.4.3 Comments on the ERT Method 1006.4.4 Overflow Group Decomposition Scheme 1006.4.5 Fredericks–Hayward Method 102

6.5 Modeling of Overflow Traffic in Systems with Finite Number of TrafficSources 103


7 Models of Links Carrying Multi-Service Traffic 107

7.1 Introduction 1077.2 Multi-Dimensional Erlang Distribution 108

7.2.1 Assumptions 1087.2.2 Process Diagram at the Microstate Level 1087.2.3 Reversibility of the Multi-Dimensional Erlang Process 1097.2.4 Multi-Dimensional Erlang Distribution at Microstate Level 1117.2.5 Macrostate Probability 1117.2.6 Interpretation of Macrostate Distribution 1127.2.7 Blocking and Loss Probability 1127.2.8 Recursive Notation of the Multi-Dimensional Erlang

Distribution 1127.2.9 Interpretation of the Recursive Notation of the Multi-Dimensional

Erlang Distribution 1137.2.10 Service Streams at the Macrostate Level 114

7.3 Full-Availability Group with Multi-Rate Traffic 1157.3.1 Assumptions 1157.3.2 Diagram of Markov Process at the Microstate Level 1157.3.3 Reversibility of the Process at the Microstate Level 1167.3.4 Macrostate Probability 1167.3.5 Recursive Notation of the Occupancy Distribution of the

Full-Availability Group with Multi-Rate Traffic 117

Contents ix

7.3.6 Blocking Probability and Loss Probability 1187.3.7 Recursive Properties of the Kaufman–Roberts Distribution 1187.3.8 Delbrouck Formula 1197.3.9 Service Streams in the Full-Availability Group with

Multi-Rate Traffic 1197.3.10 Convolution Algorithm 121

7.4 State-Dependent Systems 1227.4.1 Assumptions 1237.4.2 Diagram of the State-Dependent Process at the Microstate

Level 1237.4.3 Reversibility of the State-Dependent Multi-Dimensional Process 1247.4.4 Approximation of the State-Dependent Process by the Reversible

Process 1257.4.5 Generalized Kaufman–Roberts Distribution 1267.4.6 Blocking Probability 1277.4.7 Interpretation of the Generalized Kaufman–Roberts Distribution 127

7.5 Systems with Finite and Infinite Number of Traffic Sources 1287.5.1 Assumptions 1287.5.2 The Multi-Service Erlang-Engset Model 1297.5.3 Calculation Algorithm 1307.5.4 Comments 131

7.6 Limited-Availability Group 1327.6.1 Basic Model of the Limited-Availability Group 1327.6.2 Generalized Model of the Limited-Availability Group 1357.6.3 Comments 138

7.7 Full-Availability Group with Reservation 1397.7.1 Bandwidth Reservation 1397.7.2 Blocking Probability Equalization Rule 1407.7.3 Occupancy Distribution in the Group with Reservation 1407.7.4 Comments 1407.7.5 Modified Model of the Full-Availability Group with Reservation 1417.7.6 Comments 145

7.8 Full-Availability Group with Threshold Mechanism 1467.8.1 Single-Threshold Models 1467.8.2 Multi-Threshold Models 1497.8.3 Comments for Single-Threshold and Multi-Threshold Systems 153

7.9 Full-Availability Group with Compression Mechanism 1547.9.1 Description of the Model 1547.9.2 Comments 157

7.10 Full-Availability Group with Priorities 1587.10.1 Description of the Basic Model 1587.10.2 System with Two Priorities 1597.10.3 System with h Priorities 1617.10.4 Comments 161

References 162

x Contents

8 Modeling of Systems with Multi-Rate Overflow Traffic 165

8.1 Introduction 1658.2 Single-Service Model of the Full-Availability Group with Overflow Traffic 165

8.2.1 Assumptions of the Model 1668.2.2 Parameters of Overflow Traffic 1678.2.3 Occupancy Distribution and Blocking Probability in the

Alternative Group with Multi-Rate Traffic 1678.3 Dimensioning of Alternative Groups with Multi-Rate Traffic 1688.4 Multi-Service Model of the Full-Availability Group with Overflow

Traffic 1708.5 Comments 172References 173

9 Equivalent Bandwidth 175

9.1 Interrupted Poisson Process 1769.2 Markov Modulated Poisson Process 1779.3 Interrupted Bernoulli Process 1789.4 Comments 1809.5 Self-Similar Traffic 1809.6 Example Methods for Determining Equivalent Bandwidth 182

9.6.1 Methods for Loss Systems 1839.6.2 Methods for Queuing Systems 1869.6.3 Determination of the Equivalent Bandwidth for Self-Similar

Traffic 1869.7 Bandwidth Discretization 187

9.7.1 Comments 188References 189

10 Models of the Nodes in the Packet Network 191

10.1 Introduction 19110.1.1 Parameters of the Queuing System 19110.1.2 Classification of Queuing Systems 19110.1.3 Kendall’s Notation 192

10.2 Little’s Law 19310.3 Model of the M/M/1 System with Single-Server and Infinite Queue 196

10.3.1 Assumptions of the Model 19610.3.2 Diagram of the Service Process 19610.3.3 State Equations 19710.3.4 Characteristics of the M/M/1 System 19810.3.5 Tail Probability 200

10.4 Model of the M/M/1/N-1 System with Single-Server and LimitedQueue Size 20010.4.1 Assumptions for the Model 20010.4.2 Diagram of the Service Process 20110.4.3 State Equations 201

Contents xi

10.5 Model of the M/M/m System with m Servers and Infinite Queue Size 20310.5.1 Assumptions of the Model 20310.5.2 Diagram of the Service Process 20310.5.3 State Equations 20410.5.4 Occupancy Probability of all Servers 20510.5.5 Traffic Characteristics of the M/M/m System 205

10.6 Model of the M/M/m/N System with Limited Queue Size and LimitedNumber of Servers 20610.6.1 Assumptions of the Model 20610.6.2 Traffic Characteristics of the M/M/m/N System 206

10.7 Model of the M/G/1 System with Single-Server and Infinite Queue Size 20710.7.1 Pollaczek–Khinchin Formula 20810.7.2 Characteristics of the M/G/1 System 212

10.8 M/D/1 System 21310.9 Queuing Systems with One Server and Nonpre-Emptive Priorities 214

10.10 The M/G/R PS Model – Model of Buffers in the UMTS System 21710.10.1 Assumptions of the Model 21710.10.2 System Dimensioning based on the M/G/R PS Model 219

References 219

PART III APPLICATION OF ANALYTICAL MODELS FOR MOBILENETWORKS

11 Modeling and Dimensioning of the Radio Interface 223

11.1 Modeling of Resource Allocation in the Radio Interface 22311.1.1 Hard and Soft Capacity of the Mobile System 22311.1.2 Resource Allocation in Mobile Systems with Hard

Capacity 22411.1.3 Resource Allocation in Mobile Systems with Soft Capacity 225

11.2 Cellular System with Hard Capacity Carrying Single-ServiceTraffic 23011.2.1 Erlang Model of the Radio Interface 23111.2.2 Engset Model of the Radio Interface 231

11.3 Cellular System with Soft Capacity Carrying Single-Service Traffic 23311.3.1 Erlang Model of the Radio Interface 23311.3.2 Engset Model of the Radio Interface 234

11.4 Cellular System with Hard and Soft Capacity Carrying a Mixture ofMulti-Service Traffic Streams 23411.4.1 Model of the Radio Interface Servicing PCT1 Traffic Streams 23511.4.2 Model of the Radio Interface Servicing PCT2 Traffic Streams 23611.4.3 Model of the Radio Interface Servicing PCT1 and PCT2 Traffic

Streams 23911.4.4 Threshold Model of the Radio Interface 24111.4.5 Priorities in the Radio Interface 249

xii Contents

11.5 High-Speed Packet Data Transmission (HSPA) Traffic in the RadioInterface of the UMTS Network 25211.5.1 Description of the Model 25211.5.2 Calculation Algorithm 252


12 Modeling and Dimensioning of the Iub Interface 257

12.1 Introduction 25712.2 Example Architecture of the Iub Interface 25712.3 Modeling of the Iub Interface 259

12.3.1 Basic Algorithm for Dimensioning of the Iub Interface 26012.3.2 Dimensioning of the Iub Interface with Priorities 26112.3.3 Dimensioning of the Iub Interface Carrying HSPA Traffic 262


13 Application of Multi-Rate Models for Modeling UMTS Networks 267

13.1 Introduction 26713.2 Models of Group of Cells Carrying Multi-Rate Traffic 267

13.2.1 Fixed-Point Method 26713.2.2 Model of the Group of Cells in the Uplink Direction 27113.2.3 Model of the Group of Cell in the Downlink Direction 28013.2.4 Models of Group of Cells in the Uplink and Downlink Directions 281

13.3 Models of Traffic Overflow 28213.3.1 Model of Intercell Overflow of Single-Rate Traffic 28213.3.2 Model of Single-Rate Traffic Overflow between Macro

and Microcells 28413.3.3 Model of Intercell Overflow of Multi-Rate Traffic 28613.3.4 Comments 289

13.4 Handover Mechanisms 28913.4.1 The Model of the System Optimizing the Arrangement

of Connections 29013.4.2 Assumptions for the Model 29113.4.3 Group of Cells with Soft Handover Mechanism 294


Conclusion 301

Appendix A 303

Index 311

List of Figures

1.1 Radio link FDMA/TDMA access technology used in the GSM system 51.2 GSM system network structure 61.3 Time structure in the GSM system [1] 71.4 Time frame for reception and transmission in the mobile station 81.5 Logical channels in the GSM system 91.6 GSM system architecture supporting the HSCSD technology [7] 102.1 Radio transmission techniques in the terrestrial-based segment of the

IMT-2000 system [2] 162.2 Frequency allocation plan for the UMTS system in Europe 162.3 Duplex in FDD and TDD mode 172.4 UMTS network architecture under version R99 182.5 UMTS network architecture in R4 version 192.6 UMTS network architecture in R5 and R6 version 212.7 Spreading and transmission of the DS-CDMA signal 222.8 Despreading and reception of the DS-CDMA signal 232.9 OVSF channelization code tree 24

2.10 Bearers in the UMTS network 252.11 Frame format for the UMTS system 262.12 Channels in the UMTS system 272.13 Mapping of logical, transport and physical channels 292.14 Generation of WCDMA signal for the downlink 292.15 Generation of WCDMA signal for the uplink 302.16 RAKE receiver operation 312.17 Macrodiversity reception in the UMTS system 322.18 Power control mechanisms in WCDMA 332.19 Operation of the soft handover 352.20 Operation of the softer handover 362.21 Operation of high-speed retransmission from Node B (HSDPA/HSUPA) 39

3.1 Network architecture evolution 443.2 Long-term evolution network architecture 453.3 Long-term evolution frame structure in the downlink direction 463.4 The downlink time–frequency resource grid 473.5 Block diagram of the radio transmitter and LTE receiver in the downlink operating

in the OFDMA mode 473.6 Block diagram of the radio transmitter and the LTE receiver in the uplink

operating in the SC-FDMA mode 48

xiv List of Figures

3.7 Distribution of reference signals in the transmission with one and tworadio antennas 49

3.8 Mapping of the downlink logical and transport channels 503.9 Frequency domain scheduling principle 524.1 State transition diagram of the call service process in a group of links 644.2 Determination of the probability p0 (t) 654.3 Formation of state equations in the Markov process 674.4 Formation of state equations in the birth-and-death process 684.5 Example results of hypothetical observation of a group with the capacity V 715.1 State transition diagram for Erlang model 785.2 State transition diagram for Engset model 846.1 Alternative paths for the AB group 926.2 Types of traffic in the overflow system 936.3 Model of the system with overflow traffic 946.4 A fragment of the diagram of the Markov process in the system with overflow

traffic (illustration of Equation (6.5)) 956.5 A fragment of the diagram of the Markov process in the system with overflow

traffic (illustration of Equation (6.6)) 956.6 A fragment of the diagram of the Markov process in the system with overflow

traffic (illustration of Equation (6.7)) 966.7 Overflow diagram in the ERT method 996.8 Decomposition of the (V, R, Z) system into Z subsystems (V/Z, R/Z, 1) 1017.1 The full-availability group with several call streams 1087.2 A fragment of the diagram of the multi-dimensional Markov process in the

full-availability group 1097.3 Fragment of the diagram of the birth-and-death process in the full-availability

group 1137.4 Fragment of a diagram of the rescaled birth-and-death process in the full-

availability group 1137.5 Fragment of the diagram interpreting the recursive notation of the

multi-dimensional Erlang distribution 1147.6 Full-availability group with several call streams with different demands 1157.7 A fragment of one-dimensional Markov chain in the full-availability

group with two call streams (t1 = 1, t2 = 2) 1197.8 A fragment of the one-dimensional Markov chain in the full-availability group

with multi-rate traffic 1207.9 A fragment of the Markov process diagram in the state-dependent system 124

7.10 The microstate quadrangle in the state-dependent system 1247.11 A fragment of the one-dimensional Markov chain in the state-dependent

system with two call streams (t1 = 1, t2 = 2) 1287.12 Multi-rate system with two types of different PCT1 and PCT2 traffic streams 1297.13 The limited-availability group 1327.14 Call arrangements in the limited-availability group 1337.15 Generalized limited-availability group model 1357.16 Possible allocations of x free BBUs in the links of two types 1367.17 Reservation threshold in the full-availability group with multi-rate traffic 139

List of Figures xv

7.18 The Markov process in the full-availability group without reservation(V = 3, t1 = 1, t2 = 2) 141

7.19 The Markov process in the full-availability group with reservation(V = 3, t1 = 1, t2 = 2, Q = 1) 141

7.20 Modified diagram of the Markov process in the full-availability groupwith reservation (V = 3, t1 = 1, t2 = 2, Q = 1) 142

7.21 Transferring service streams of class 1 to state n (A1t1 = 0) 1437.22 Full-availability group with single-threshold mechanism 1467.23 A fragment of the Markov process diagram in the single-threshold system 1497.24 Example of a multi-threshold system with one class of calls 1507.25 Blockable states in the multi-threshold system 1527.26 Example full-availability group with compression mechanism 157

8.1 A fragment of the telecommunications network with single-service trafficoffered to primary groups 166

8.2 A fragment of telecommunications network with multi-service trafficoffered to primary groups 171

8.3 Decomposition of primary group with multi-rate traffic 1729.1 Traffic control levels in the present-day cellular network 1769.2 Example interrupted Poisson process 1769.3 Example Markov modulated Poisson process 1789.4 Example interrupted Bernoulli process 1799.5 Equivalent bandwidth evaluation algorithm for the Lindenberger-Tidblom

method 1849.6 Bandwidth discretization 188

10.1 A simple queuing system 19210.2 The M/M/3 queuing system 19310.3 Call arrival and service process in the queuing system 19410.4 M/M/1 queuing system 19610.5 Markov process equilibrium in the M/M/1 system 19610.6 M/M/1/N-1 queueing system 20110.7 Markov process equilibrium (birth-and-death process) in the M/M/1/N-1

system 20110.8 M/M/m queuing system 20310.9 Markov process equilibrium (birth-and-death diagram) in the M/M/n system 20310.10 Time diagram for a defined embedded Markov chain 20810.11 Residual service time of currently serviced call 21611.1 The phenomenon of soft capacity 22411.2 Available resources in a system with soft capacity 22911.3 Resource allocation in the radio interface 22911.4 Resource allocation in the multi-rate radio interface 23011.5 Model of the radio interface with hard capacity that services PCT1 traffic 23111.6 Model of the radio interface with hard capacity that services PCT2 traffic 23211.7 Model of the radio interface with soft capacity that services PCT1 traffic 23311.8 Model of the radio interface with soft capacity that services PCT2 traffic 23411.9 Model of the radio interface in the cell servicing multi-rate PCT1 traffic 23511.10 Model of the radio interface in the cell servicing PCT2 multi-rate traffic 237

xvi List of Figures

11.11 Model of the radio interface in the cell servicing multi-rate traffic of thetype PCT1 and PCT2 239

11.12 Resources of the cell with threshold mechanism for i class calls 24311.13 Relation between traffic classes and groups of users 24911.14 The capacity division in the HSDPA radio interface 25412.1 Access part of the UMTS network 25812.2 One of the most common methods for establishing a connection between

the UMTS base station and Radio Network Controller is the one applyingIMA technology 259

13.1 Concept of the application of the fixed-point method for intercell interference 26813.2 The generalized group of cells model 27213.3 Simplified group of cells 27813.4 Two-cell simplified model 27813.5 Model of the cell in the downlink direction 28013.6 Model of intercell overflow 28313.7 Model of overflow between microcells and macrocell 28513.8 Model of a system with intercell overflow of multi-rate traffic 28713.9 Connection handover in the group of cells 29013.10 The formation of assemblies of neighboring cells 29213.11 Fragment of the mobile network with soft handover mechanism 295

List of Tables

1.1 Frequency range and the number of available channels for the GSM 900 and1800 MHz systems 4

1.2 Data rate performance in the radio interface in the HSCSD technology in relationto used number of channels and coding rate [8] 10

1.3 Coding schemes in the GPRS system [10] 121.4 Coding schemes in EDGE technology [10] 132.1 Spreading factors in the downlink channel and the corresponding

transmission speeds 232.2 Spreading factors in the uplink channel and the corresponding

transmission speeds 242.3 A comparison of the properties of DCH channels (R99), HS-DSCH (HSDPA)

and E-DSH (HSUPA) 402.4 Service classes of the UMTS system and their basic parameters [8] 415.1 Erlang tables 816.1 Types of traffic in the telecommunication network 939.1 Parameters of ON/OFF sources 1809.2 Example values of the coefficients c1 and c2 presented in [2] 1849.3 Parameters for FBM traffic 187

11.1 Examples of R99 services and load factors [5] 22611.2 HSUPA services and load factors [6] 22712.1 An example of service class mapping into ATM classes 25913.1 Sets of neighboring cells 273A.1 Modeling of the radio interface in the single cell carrying single-service traffic 304A.2 Modeling of the radio interface in the single cell carrying multi-service traffic 305A.3 Modeling of the group of cells carrying multi-service traffic 306A.4 Modeling of the Iub interface in the UMTS network carrying multi-service

traffic 307A.5 Modeling of a group of cells carrying multi-service with call

handover mechanism 308A.6 Modeling of the group of cells carrying overflow traffic 309A.7 Modeling of the HSPA traffic in the UMTS network 310

Preface

The first connection in the GSM network was set up in 1991 and this year marks the onsetof the dynamic development of cellular telephony we are experiencing today. The unques-tionable success of the GSM telephony has motivated further research in the field and thedevelopment new technologies for cellular telephony. Initially, it was assumed that cellularnetworks would also provide their users with multimedia services and would offer accessto the Internet. Subsequent research eventually succeeded in working out a standard for thethird-generation telephony (UMTS). However, UMTS network telephony has been develop-ing at a much slower pace than its GSM predecessor, hampered by substantially high costsof rendering a network operational. Currently, services provided by the UMTS network areoffered by most cellular network operators. The extension of a second-generation cellularnetwork and the construction of a third-generation network must involve installation of newsystems in the network that make it possible to increase the number of subscribers and tointroduce new services, primarily those that have their quality parameters defined. Within thiscontext, the need to design optimum network resources through a substantial reduction inthe costs of network modernization is a topical and particularly important issue. The designprocess involves the determination of the network topology (the number of nodes, that of thestructure, etc.), defining its resources (capacity) and preparing appropriate management andtraffic-distribution strategies in such a way as to ensure that the total investment overlays forthe creation of a network can be kept as low as possible. In order to design a network properly,and dimension its elements appropriately, knowledge of the characteristic methodologies ofparticular types of designed networks is indispensable. These methodologies, frequently verycomplex, stem from the problems and issues formulated in, for example, traffic theory, thetheory of graphs, the theory of stochastic processes, combinatorics and integer programming.Knowledge of appropriate design and optimization methods facilitates effective functioningof cellular networks that are open to expansion and reconfiguration.

The creation of the GSM network, then the UMTS network, and, in the near future, theLTE network, has been, and still is a challenge for traffic theoreticians and engineers, whichis particularly observable in 3G and 4G networks that service a mixture of traffic streams withmulti-rate traffic. The analysis of present-day cellular networks servicing multi-rate trafficand guaranteeing a defined quality level of call service indicates a need for further researchaimed at working out effective dimensioning methods and evaluation methods for traffic load-carrying capacity in such networks. The dimensioning process in 3G and 4G systems shouldallow designers to determine the capacity of particular elements of the system that will make itpossible, given the assumed load of the system, to ensure the assumed level of GoS (Grade ofService) and QoS (Quality of Service). The dimensioning of a system also involves taking intoconsideration mutual dependencies of call-service processes in interfaces in the access part of

xx Preface

the network (for example, in the radio interface and in the Iub interface in the UMTS system).Such an approach results in the creation of new traffic models that enable us to dimensionand optimize 3G and 4G systems even when demands for resources are of a time-dependent(changeable) character.

The cellular network design process thus depends on the construction of the radio infras-tructure part and the fixed part of the network, along with a determination of the capacityof individual elements of the system necessary to maintain complex quality parameters. It istherefore necessary to proceed with such activities as the evaluation of the capacity in radiointerfaces and other interfaces in the access network of the system (for example, the Iub inter-face in the UMTS system). Determination of the capacity of radio interfaces in a network withsoft and hard handover is by no means trivial. Because of the multi-service character of traffic,a response to this strong challenge is to work out appropriate methods for the evaluation oftraffic capacities of the interfaces located in the “nonradio” part of the access network. Theradio interfaces and other interfaces of the access network in 2G and 3G systems (as wellas in the concept of the 4G system) are fundamental for the desired traffic effectiveness inthe network. Due to the low capacity of the radio interface, which is also limited by inter-ference from neighboring cells and by capacity as well as the organization (of resources) ofjoined “non radio” interfaces, operators take advantage of a number of mechanisms that allowthem to increase the traffic effectiveness of the system such as compression (e.g. for servicesin HSDPA, HSUPA), reservation of resources for predefined classes of calls, introduction ofthresholds making resources allocation conditional on the interference load at the level of thecontrol function controlling access to resources, priorities for selected services or group ofsubscribers, optimum allocation of connections in a group of cells (implementation of retryhandover and hard handover mechanisms), traffic overflow between neighboring cells and cellsof other networks covering a given area of a given operator (for example, from 3G networkcells to 2G cells).

This book aims to provide extensive information on modeling appropriate interfaces andgroups of interfaces of 3G and 4G access networks that service a mixture of traffic characterizedby different properties and different GoS and QoS requirements. In particular, the book presentsthe following analytical models of systems servicing multi-rate traffic in interfaces betweenthe access network and implemented traffic-management mechanisms:

• prioritization models – with the possibility of introducing a hierarchy in the service processfor particular call classes and pushing out calls with lower priority;

• reservation models – with the possibility of introducing a mechanism for the reservation ofpredefined classes of calls depending on the load of the system;

• threshold models – with the possibility of a change in service parameters (the number of allo-cated resources, service time) depending on predefined thresholds indicating the admissibleload level in the interface;

• compression models – introducing a change in the parameters of serviced calls in a cell(compression of allocated resources) in the event of a lack of free resources for newcalls;

• overflow models (with traffic overflow to neighboring cells) – multi-rate traffic overflowsfrom cells (sectors) with the heaviest load or from other network of the operator;

• models for soft handover, softer handover and soft-softer handover connections in whichthe mobile station is connected to a number of base stations.

Preface xxi

The models presented in the book form the basis of a coherent methodology for the di-mensioning of the elements of the mobile network most sensitive to traffic load, namely theinterfaces of the radio access network. This book provides the reader with extensive infor-mation on problems and issues in the traffic engineering, dimensioning and optimization ofUMTS and LTE networks.

The authors have been involved in teaching and research investigations on the optimizationand design of the UMTS network at the Faculty of Electronics and Communications of PoznanUniversity of Technology for a number of years. Over the years, many original models workedout by the authors have been published in research journals and conference proceedings. Thefinal version of the book has also been influenced by the authors’ experience from a numberof projects carried out for operators of cellular networks between 2003 and 2008.

Chapter 1 outlines the basic conceptual framework of a GSM system. The architecture ofthe system and its time structure with logical channels are discussed. Particular attention isgiven to data transmission technologies: HSCSD, GPRS and EDGE. The chapter also brieflydelineates traffic-management mechanisms that are important for the GSM network, such asdirected retry handover, traffic handover and queuing.

Chapter 2 briefly discusses the most important elements of the UMTS network. Systemarchitecture is presented as well as the basic operating principles of the WCDMA radio interfaceand channel and scrambling codes. Bearers and frame structure in the UMTS system are definedand explained. Special attention is given to a description of logical, transport, physical channelsin the WCDMA radio interface. The chapter discusses essential methods for radio resourcemanagement in the UMTS system, including power control, handover control, call admissioncontrol, packet scheduler and load control. The HSPA technology (high-speed packet datatransmission) for the uplink and the downlink is discussed and a brief presentation of theclassification and categories of the most important services available in the UMTS network isprovided.

Chapter 3 presents the concept of the evolution of multi-service systems towards the fourthgeneration system, known as the LTE system. Possible changes in the system architectureare discussed as well as a number of available proposals for new transmission techniques,including LTE MIMO. A classification of transport and physical logical channels potentiallyavailable for the LTE network is highlighted. The chapter also briefly presents a concept forresource management in this system.

Chapter 4 provides a discussion of basic issues and questions related to the analysis ofsingle-rate systems. Basic concepts and properties of the call stream, service stream, Markovprocess and the birth and death process are defined and explained. The concepts of telecom-munications traffic and traffic intensity are introduced and explained. The last section presentsbasic parameters for the quality of service and grade of service. Particular attention is given tothe determination of blocking and the loss probability in telecommunications systems.

Chapter 5 familiarizes the reader with the most common models of the full availability groupwith single-rate traffic, which is known in engineering and the theory of traffic. A method forthe determination of the occupancy distribution on the basis of the analysis of the birth anddeath process is presented and discussed. The chapter considers service of traffic streamsgenerated both by an infinite number of traffic sources (Erlang model) and by a finite numberof traffic sources (Engset model). All important parameters for full-availability groups, such asblocking probability and the loss probability, the occupancy probability of precisely determined

xxii Preface

channels and carried traffic, are defined and discussed. The differences between the Erlang andthe Engest model are identified and commented on.

Chapter 6 formulates and discusses the problem of traffic overflow in single-service systems.The basics for the analysis of overflow systems and the concept of the primary group and theoverflow group are presented. Special attention is given to a method for the determination of theparameters of overflow traffic: mean value, variance and the peakedness coefficient. Methodsand algorithms for dimensioning of alternative groups, the equivalent random traffic methodand the Fredericks-Hayward method are discussed in detail.

Chapter 7 analyzes of multi-dimensional, single-service and multi-service systems. Prop-erties and characteristics multi-dimensional Markovian processes at the level of the so-calledmicro and macro states are presented and discussed. Particular attention is given to methodsfor the determination of the occupancy distribution in state-independent systems. Recursiveand convolution algorithms for the determination of the occupancy distribution and othercharacteristics of the full-availability group carrying a mixture of different multi-rate trafficare presented and discussed. The following state-dependent systems are modeled: group withreservation, limited-availability group, threshold systems, systems with compression, and sys-tems with priorities. The chapter also describes methods for modeling multi-rate systems fortraffic generated by both a finite and infinite number of traffic sources.

Chapter 8 presents the concept of overflow traffic in multi-service systems. The occupancydistribution in the alternative group is determined. This is a mixture of multi-rate traffic charac-terized by different values of the peakedness coefficient. A method for the determination of theparameters of multi-rate traffic that overflows from primary groups, that is the average valueand variance, is presented. The chapter also proposes a method for dimensioning alternativegroups with overflow traffic from many primary groups servicing multi-rate traffic.

Chapter 9 presents an approach for dimensioning systems with virtual channel switching.The basics of modeling traffic sources with variable bit rates are given. Traffic source modelsof the type: ON/OFF IPP, ON/OFF IBP, MMPP and self-similar traffic sources are presented.The chapter also includes a presentation of example models for the determination of equiv-alent bandwidth and of the method for bandwidth discretization, which forms the base fordimensioning of broadband networks.

Chapter 10 discusses the issues concerning modeling of basic queuing systems. Classifica-tion and notation pertaining to queuing systems are discussed. Basic dependencies between themost important parameters for queuing systems are identified and defined. The most importantpart of the chapter demonstrates the parameters and characteristics of the following queuingsystems: M/M/1, M/M/1/N, M/M/m, M/M/m/N, M/G/1, M/D/1 and the M/G/1 system withpriorities without preemption. The chapter focuses on the M/G/R PS model used for modelingnodes in modern packet networks.

Chapter 11 is devoted to modeling and dimensioning radio interfaces in cellular networks.First, the method for bandwidth discretization in systems with soft capacity is discussed, withparticular attention given to the method for determining the basic bandwidth unit (BBU) forthe radio interface. Then, methods for modeling the radio interface with single-rate and multi-rate traffic are presented. The possibility of service in the interface with traffic from finite andinfinite traffic sources is considered. The influence of interference upon the soft capacity ofcell is taken into consideration through the application of the fixed-point methodology.

Chapter 12 discusses methods for modeling the Iub interface in the UMTS network on thebasis of a model of the full-availability group with multi-rate traffic servicing traffic from a

Preface xxiii

finite and infinite number of sources. The proposed model considers the possibility of introduc-ing priorities for selected classes of calls. It also considers the feasibility of service-orientedsolutions for traffic that undergoes compression, including the average throughput availablefor HSPA subscribers.

Chapter 13 sums up the book, providing a number of computational algorithms for selectedelements of the radio access network, commonly used in engineering practice. The chapterdiscusses, for example, a model of a group of cells that services multi-rate traffic from finiteand infinite traffic sources, handover model and traffic overflow between sectors of a givencell and between different cells. In the model of a group of cells the influence of interferenceon the soft capacity of cells has been taken into consideration through the application of thefixed-point methodology.

Acknowledgements

We would like to thank the first readers of our book: Professor Tadeusz Czachorski, JanuszWiewiora and Tomasz Olszewski for offering valuable advice that helped shape the book inits final form, and our families for their patience, understanding and support.

Maciej StasiakMariusz Głabowski

Arkadiusz WisniewskiPiotr Zwierzykowski

Part IMobile NetworkStandards

1Global System for MobileCommunications

1.1 Introduction

The Global System for Mobile Communications, or GSM,1 is the so-called second-generation(2G) cellular mobile phone system. Its predecessors, first generation (1G) mobile phonesystems, were analog systems such as the Nordic Mobile Telephone (NMT) system andthe Advanced Mobile Phone System (AMPS) [1]. Designers of analog systems did notrealize though, that cellular telephony would in a short time become a universal and popu-lar service and thus the systems in question had a rather limited capacity. Moreover, most ofthem were incompatible with one another, which resulted in serious limitations because usersof a given network were exclusively subscribers of a given operator.

The GSM standard was developed thanks to a European initiative aiming at creating auniform, open cellular mobile phone system. Originally, the standard was to be applied andimplemented in the 12 countries of the European Common Market. Accordingly, the EuropeanConference of Postal and Telecommunications Administrations (CEPT) created the GroupeSpecial Mobile (GSM) in 1982 to develop a standard for a mobile telephone system thatwould operate in the 900 MHz bandwidth [2]. In 1987, a group of 15 operators from acrossEurope signed a memorandum of understanding in which they agreed to implement the GSMtechnology to develop a common cellular telephone system across Europe. In 1989 responsi-bility for GSM was transferred to the European Telecommunication Standard Institute, and thefirst phase of the GSM 900 specifications was completed one year later (GSM 900 Phase 1).Towards the end of phase 1 recommendations for the GSM 1800, operating in this bandand aiming to service densely populated urban areas, were worked out. In October 1995, itwas announced that work on the second phase of the GSM standard (GSM Phase 2) hadbeen completed.

1Originally, the acronym GSM designated the Groupe Spécial Mobile, the name of the organization that was createdto develop a standard for a mobile telephone system.

Modeling and Dimensioning of Mobile Networks: From GSM to LTEMaciej Stasiak, Mariusz Głabowski, Arkadiusz Wisniewski and Piotr Zwierzykowski© 2011 John Wiley & Sons, Ltd.

4 Modeling and Dimensioning of Mobile Networks

Table 1.1 Frequency range and the number of available channels forthe GSM 900 and 1800 MHz systems

Feature / Bandwidth GSM 900 GSM 1800

Uplink (MHz) 890–915 1710–1785Downlink (MHz) 935–960 1805–1880Number of available channels 124 374

Work on the specifications for the GSM system continued in the following years (GSMPhase 2+). This included such specifications as the standard for the intelligent networks for theGSM CAMEL (Customized Application for the Mobile Network Enhanced Logic), fast datatransmission with packet switching HSCSD (High Speed Circuit Switched Data), technologyof packet transmission of data GPRS (General Packet Radio Service), and later also EDGE(Enhanced Data rates for GSM Evolution).

Since 1999, the leading role in the development and standardization of the GSM system hasbeen taken by the Third Generation Partnership Project (3GPP) set up by regional standard-ization bodies working on new concepts in telecommunications systems in December 1998.Its scope of operation covers establishing norms and publishing technical reports on GPRSand EDGE data transmission as well as working out standards for the third-generation mobilenetworks [3].

One of the major objectives in designing the GSM system was to develop a digital system thatwould enable voice transmission, short text messaging (SMS) and data transmission. Moreover,the system was to handle international roaming. Table 1.1 shows the frequency bandwidthsadopted in the GSM 900 and 1800 MHz systems for the uplink (from the mobile station tothe base station) and for the downlink direction (from the base station to the mobile station).Full duplex transmission in both bandwidths is carried out based on separate frequency ranges.Each of the bands is divided into channels with a bandwidth of 200 KHz. For the GSM 900system there are 124 available (separately for the uplink and for the downlink direction), andfor the GSM 1800 there are 374 channels.

The designers of the GSM system decided to provide access to the radio link using frequencydivision multiple access (FDMA) and time division multiple access (TDMA) simultaneously(Figure 1.1). Each carrier frequency is then divided into eight time slots. In order to set up aconnection in the GSM system it is necessary to assign each user a defined frequency channeland a time slot in which the signal can be transmitted or received.

1.2 System Architecture

In the GSM system, as shown in Figure 1.2, three basic subsystems can be distinguished [4]:

• base station subsystem (BSS);• core network (CN);• user equipment (UE).

The interfaces between particular elements of the system are defined. These interfaces deter-mine rules for cooperation between devices.

modeling and dimensioning of mobile...

Documents