1. 1. The Complete Wireless Communications Professional: A Guide for Engineers and Managers
2. 2. The Complete Wireless Communications Professional: A Guide for Engineers and Managers William Webb Artech House Boston London
3. 3. Library of Congress Cataloging-in-Publication Data Webb, William, 1967 The complete wireless communications professional : a guide for engineers and managers / William Webb p. cm. (Artech House mobile communications library) Includes bibliographical references and index. ISBN 0-89006-338-9 (alk. paper) 1. Wireless communication systems. 2. Mobile communication systems. I. Title. II. Series. TK5103.2.W42 1999 621.6845dc21 98-51802 CIP British Library Cataloguing in Publication Data Webb, William, 1967 The complete wireless communications professional : a guide for engineers and managers(Artech House mobile communications library) 1. Wireless communication systems. I. Title 621.382 ISBN 0-89006-338-9 Cover design by Lynda Fishbourne 1999 ARTECH HOUSE,INC. 685 Canton Street Norwood, MA 02062 All rights reserved. Printed and bound in the United States of America. No part of this book may be reproduced or utilized in any form or by any means,electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system,without permission in writing from the publisher. All terms mentioned in this book that are known to be trademarks or service marks have been appropriately capitalized. Artech House cannot attest to the accu- racy of this information. Use of a term in this book should not be regarded as affecting the validity of any trademark or service mark. International Standard Book Number: 0-89006-338-9 Library of Congress Catalog Card Number: 98-51802 10987654321
4. 4. Contents Preface What is a complete wireless professional? xiii Introduction xiii Format of this book xiv Acknowledgments xv Part I Introductorymaterial 1 1 Some interesting history 3 1.1 Introduction 3 1.2 Early history 4 1.3 Some key milestones in mobile radio history 8 1.4 Recent history 10 References 16 Part II Mobile radio systems 17 2 The basics of mobile radio 19 2.1 Introduction 19 v
5. 5. vi The Complete Wireless Communications Professional 2.2 Basic principles of propagation 20 2.3 Radio spectrumutilization 32 2.4 Basic system design 37 2.4.1 System overview 37 2.4.2 Voice encoding37 2.4.3 Securetransmission 46 2.4.4 Overcomingchannelimperfections48 2.4.5 Frequency and phasemodulation55 2.4.6 Clock recovery 64 2.4.7 Carrier recovery 66 2.4.8 Multipleaccess68 2.5 Packet and circuit transmission 79 2.6 Theoretical capacity of mobile radio systems 80 References 82 Further reading 83 3 Cellular radio technologies 85 3.1 The range of cellular systems 85 3.2 GSM 88 3.2.1 System architecture 88 3.2.2 Locatinga subscriberandstarting calls 91 3.2.3 Transmissionwithin GSM 93 3.3 cdmaOne 101 3.4 Other systems 106 References 106 4 Private mobile radio systems 107 4.1 Introduction 107 4.2 Simple private radio systems 114
6. 6. Contents vii 4.3 TETRA 119 4.3.1 Introduction 119 4.3.2 System operation120 4.3.3 Technicalparameters 123 4.4 Other systems 125 References 125 5 Other mobile radio systems 127 5.1 Introduction 127 5.2 Cordless systems 128 5.2.1 Overview of cordless telephony128 5.2.2 Digital enhancedcordless telephone 130 5.2.3 Personal handiphonesystem 134 5.3 Wireless local loop systems 135 5.3.1 Introduction to wireless local loop135 5.3.2 Access technologies:radioandcable137 5.3.3 WLLand cellular:the differences142 5.3.4 Technologiesfor WLLand LMDS/MVDS144 5.4 Satellite systems for telephony 149 5.4.1 Introduction 149 5.4.2 Concept 150 5.4.3 Economics of satellitesystems153 5.5 TV, radio, and other systems 154 References 159 6 Interfacing with fixed networks 161 6.1 The need for fixed networks 161 6.2 Fixed network architectures 162 6.3 Fixed network protocols 170
7. 7. viii The Complete Wireless Communications Professional 6.4 Fixed mobile convergence 178 6.4.1 Introduction 178 6.4.2 Definingfixed-mobile convergence179 6.4.3 Possiblesolutions 179 6.4.4 The future of the FMC marketplace 185 References 187 Part III The mobile network operator 189 7 Designing a mobile radio network 191 7.1 Technical design 191 7.1.1 Introduction 191 7.1.2 Networkplanning193 7.1.3 Radio planning196 7.1.4 Microcells and picocells198 7.1.5 Interconnection 199 7.1.6 Operations and maintenance planning203 7.1.7 Supplierselection204 7.1.8 Networkdeployment205 7.2 Applying for a license 205 7.3 The mobile radio equipment manufacturer 208 References 210 8 Economics of a mobile radio network 211 8.1 Understanding financial information 211 8.1.1 Introduction to accounting211 8.1.2 The profit and lossaccount 212 8.1.3 The balancesheet215 8.1.4 The funds flow statement 218 8.1.5 Performingfirst pass modeling221
8. 8. Contents ix 8.2 The business case223 8.2.1 The overall structure of the business case223 8.2.2 The networkcosts223 8.2.3 The operatingexpenses227 8.2.4 Revenue 227 8.2.5 Financing 233 8.2.6 Summary 235 References 238 9 Operating a mobile radio network 239 9.1 Introduction 239 9.2 Monitoring the network 240 9.3 Tariff policies and their implications 244 9.4 Capacity enhancement 246 9.4.1 Introduction 246 9.4.2 The available capacity enhancement techniques 246 9.4.3 Dual-band operation247 9.4.4 Techniques affectingthe cluster size247 9.4.5 Using more cells251 9.4.6 Which capacity enhancement techniques should be used when? 252 10 Large users of mobile radio networks 255 10.1 Introduction 255 10.2 Railways 257 10.2.1 Introduction 257 10.2.2 Current railwaycommunications withinEurope258 10.2.3 Railway requirements 259 10.2.4 PMR versus cellular261
9. 9. x The Complete Wireless Communications Professional 10.3 Police 263 10.3.1 Introduction 263 10.3.2 Descriptionof requirements 264 10.3.3 Selectionof radio system 265 10.4 Other emergency services 266 10.5 Other users 267 11 Future mobile radio systems 269 11.1 Progress in radio systems 269 11.2 The third generation vision 270 11.3 Designing the third generation system 274 References 277 Part IV Regulators and governments 279 12 Radio spectrum 281 12.1 Introduction 281 12.2 The management of radio spectrum 282 12.3 Modern allocation and assignment methods 287 12.4 Implications for the mobile radio operator 290 12.5 Government policy 294 References 297 13 Standardization 299 13.1 Introduction 299 13.2 Standards-making bodies 300 13.3 Writing standards 304 References 307
10. 10. Contents xi Part V Becoming a better wireless professional309 14 Areas of conflict 311 14.1 Introduction 311 14.2 TETRAversus GSM 313 14.2.1 Backgroundto the debate 313 14.2.2 Evaluationof thetechnologies 313 14.2.3 Economiccomparison 315 14.2.4 Analyzing the debate318 14.3 DECT versus PHS 319 14.3.1 Backgroundto the debate 319 14.3.2 The keyissues 319 14.3.3 Analyzing the debate321 14.4 CDMA versus TDMA 322 14.4.1 Backgroundto the debate 322 14.4.2 The capacityof CDMA versus TDMA 323 14.4.3 Otherissues introducedintothedebate325 14.4.4 Analyzing the debate327 14.5 Handling conflict 327 References 328 15 Management 329 15.1 Introduction 329 15.2 An overview of management 330 15.3 Understanding corporatestrategy 332 References 338 16 The complete wireless professional339 16.1 Introduction 339
11. 11. xii The Complete Wireless Communications Professional 16.2 Conferences and publications 340 16.3 Links with research organizations 346 16.4 Qualifications 349 References 350 Appendix A Erlang B macro 351 Appendix B Mandating standards 353 B.1 Introduction 353 B.2 The key issues 354 B.3 Case studies 359 B.4 Implications 361 B.5 Conclusions 362 References 363 Glossary 365 About the author 385 Index 387
12. 12. Preface What is a complete wireless professional? Since we cannot know all that is to be known of everything, we ought to know a little about everything. Blaise Pascal (16231662) Introduction When thinking about mobile radio engineersthereis atendencyto assume that the engineeringfunctionrelatessolelyto the technical aspects of the network, suchas the equipment designor the network design. That is certainlyakey part of the role of a mobile radio engineer. However, increasinglyengineersare requiredto interact withprofession- als from other divisions. The complete wireless professionalshould know about mobile networks;fixednetworks;other types of mobile systems;regulatoryandgovernment policy;the requirements ofthe users;and financial, legal, and marketingissues. Otherwise, there is a tendencyfor the engineering, finance, andmarketingdepartments to be xiii
13. 13. xiv The Complete Wireless Communications Professional completelyseparate entitiesthat are unable to communicate accurately with each other. The net result is products that do not fit the marketing requirementsor are not financiallyviable, although theymay be master- pieces of advanced engineeringdesign. This book looks at the range of topics that complete wireless commu- nications professionals needto understandinorder to perform theirtask well. Clearly, above all else, theyneedto have an engineeringknowledge of how mobile radio systems work. Sucha body of knowledge is con- tainedin many excellent text booksandreference works;the intention here is to provide the salient points ineacharea and a guide to further reading. In otherareas such as finance, the complete wireless professional onlyrequires anunderstanding of the key issues, and the brief description provided in this bookmay be sufficient. To some extent, this bookgathers referencematerial from awide range of technical, managerial, and financial sourcesandrepresentsin summary form those issuesthat are key to the completewireless profes- sional. By encompassingthe informationina suitable frameworkand providing additional chapters onareas such as the resolutionof conflicts and career structure, it is hopedthat the effect of the whole is greater than the sum of its parts. Above all, complete wireless professionals needto understandthe world around them and apply this knowledge to engineeringissues. This bookdescribes the worldof mobile radio. Format of this book This book is divided into five parts. This first part provides introductorymaterial inthe form of achap- ter discussingthe relevant historyof mobile communications. The secondpart looks at mobile radio systems, consideringthe basics of mobile radio, the designof cellular andprivate radio sys- tems, and the issues concernedwithinterworkingwith the fixed network. The thirdpart looks at the role of amobile radio operatoranddis- cusses the designof mobile radio networks, the operationof these
14. 14. Preface xv networks, the needs of large user groups, and the future of mobile radio systems. The fourthpart looks at the regulatoryand government decisions that impact mobile radio, includingthe management of radio spec- trum, the standardizationof mobile radio systems, andthe effect of government policyonthe mobile radio community. The final part focuses onbecomingabetter engineer byconsidering the resolutionof conflictssuchas the time division /code division multiple access(TDMA/CDMA) debate, onthe need for understandingmana- gerial issues, andfinally on the way to become acomplete wireless professional throughprofessional vehiclesandcareer structure. Acknowledgments In writing this book, I drew upon all my experiencegainedover years in the industry. I have learnedsomethingfrom almost everyonewithwhom I have come into contactandthank all of those withwhom I have had dis- cussions. Special thanks are due to a number of key individuals. During my time at Multiple Access Communications, ProfessorRaySteele, Pro- fessorLajosHanzo, Dr. Ian Wassell, and Dr. JohnWilliams, amongothers, taught me much about the workings of mobile radio systems. At Smith System Engineering, RichardShenton, Dr. Glyn Carter, and Mike Shan- non provided valuable knowledge, as have contacts witha number of others inthe industry, includingMichel Mouly(Independent Consult- ant), DirkMunning (DeTeCon), Michael Roberts(Alcatel), Mike Watkins and Les Giles (Racal-BRT), Mike Goddardand Jim Norton(Radiocommu- nications Agency), and PhillipaMarks (NERA). At NetCom Consultants, Steve Woodhouse, DonPearce, MarkCornish, andothers taught me about WLL and cellular operators. While at Motorola, Raghu Rau has provided invaluable guidance. Many presentations andpapers from those involved in the mobile radio industrycontributedto myunderstanding. Most importantly, mywife, Alison, has perseveredas I workedmany eve- nings and spent much time away from home inmy quest to absorb knowledge in as many spheres as possible, to gainadditional qualifica- tions, and to try to become acompletewirelesscommunications profes- sional myself.
15. 15. IPART Introductory material
16. 16. 1CHAPTER Some interestingContents history1.1 Introduction 1.2 Early history 1.3 Some key milestones in mobile radio history When I want to understand what is happen- ing today, or to try and decide what will1.4 Recent history happen tomorrow, I look back. Oliver Wendell Holmes, Jr. 1.1 Introduction The complete wireless professional does not needto be a historian. One will be able to designmobile radio networks or products equally well whether aware or not that the first demonstrationof aradio transmission was made by Hertz. However, the complete wireless professional woulddo well to learn from some of the lessons of mobileradio over its history. Because of the dramatic change in the design and deployment of mobile radio over the last 100 years and the change in eco- nomics and uses of radio systems, the most relevant lessons are thosefrom most recent times. Hence, this chapter provides ashort summary of the historical backgroundand 3
17. 17. 4 The Complete Wireless Communications Professional tries to select someof the keylessons that canbe learned. An excellent descriptionof the historyof mobileradio development, including a detailedanalysis of the last 15 years of cellular deployment, canbe found 1in [1], while a very readable biography of the founder of Motorola, Paul Galvin [2], describeshowMotorolashapedmuchof the development of mobile radio from 1930onward. This sectionprovides ashort summaryof the topics that the references describe inmuchgreater detail. The key issues that the completewirelessprofessional shouldlearn from this chapter are: History, particularlyrecent history, has a number of important les- sons that do not seem to have beenlearnedby many in the mobile radio business; Standardization is far from an assuredroute to success, withthe globalsystem for mobile communications (GSM) being the only stan- dardized product among a wide range of standards to have been highly successful; Standardization of GSM took13 years intotal; New radio systems suchas cdmaOne can emerge from unexpected sources and, with the right political andeconomicbacking, can become important globally. 1.2 Early history Some of the key historical developmentsleadingto todays mobile radio systems are shownin Figure 1.1. The first radio transmissions are attributedto HeinrichHertzin1885. The first radio system was very simple, usinga switch and an induction coil to generate asparkacross two electrodes. The receiver was a loop made from copper wire around35 cm in diameter, witha small gap in the loop. Whenthe transmittergeneratedaspark, a small spark was seento jump the gap in the receivingcoil. Sparktransmissionwas used as the basis for most radio equipment until around1915 andis the reasonwhy radio operators are sometimes knownas sparks. Hertzs work aroused 1. This is a superbly written and highly authoritative bookand comes highly recommended.
18. 18. Some interesting history 5 Technical milestones Applications 1880 Theoretical prediction of radio waves Generation of radio waves (Hertz) Tunedcircuit (Lodge) 1890 Aerial/earth system (Marconi) 1900 Cross-channel tests (UK) Royal Navy (UK) Speech transmission (Fessenden) Transmission to automobile (US) Thermionic valve (Fleming) Merchant shipping(UK)1910 Transatlantic telegraph service Valve transmitter (Meissen) Radio direction finding(UK) Aircraft use for artillery spotting(UK)1920 Transportables (UK) Police use (Detroit, US) Fishing boats (Norway, UK)First international spectrum 1930 Aviation navigation andcontrolconference Frequency modulation (Armstrong) Telephones on ocean liners (UK) 1940 Private mobile radio systems (US) Operator controller mobile phones (US)1950 Cellular concept (Bell Labs) Junction transistor (Schockley) 1960 Digital integratedcircuits 1970 Automatic mobile telephones Cellular test (US) Solid state switches 1980Microprocessor Cellular services (Japan) 1990 Digital cellular networks (Europe) Iridium satellite services launched Figure 1.1 Key historical milestones. much scientificinterest, but it tookthe emergenceof ascientist withbusi- ness acumento move these discoveriesinto the commercial domain.
19. 19. 6 The Complete Wireless Communications Professional Perhaps the father of todays mobile radio systemswas Guglielmo Marconi, bornin Italy in 1874. Marconi was first ascientistwho made the important breakthroughof redesigningthe transmitter from agap betweentwo electrodesto connectingone electrodeto the earthand the other to ametal pole placedon topof a mast (the aerial of today). By this method, he was able to demonstrate transmissions over arange of a few kilometers ratherthanjust across the laboratory. Marconi triedto approach the Italian Post Officeforsponsorshipof his workbut was unsuccessful. He thenmoved to the UnitedKingdom, where the British Post Office was preparedto provide sponsorship. This proved to be a shrewd move because the first applicationof mobileradio was in shipping and, at the time, the UnitedKingdom had the worlds largest shipping fleet. Marconis first salesof radio systems were in1900whenthe Royal Navy ordered32 sets at the equivalent of todays cost of $500,000per set with annual royaltypayments of $270,000per set. It was revenue from deals suchas this that enabled Marconi to founda companyand develop new radio products. For some time, the Marconi companywas the worlds largest producerof radio equipment. Marconiwas awarded the Nobel Prize for Physicsin1907 anddiedin 1937. The company he foundedis now part of the GeneralElectricCorporation (GEC). Other important developments includedthe inventionof the thermi- onic diode in1904, whichlead to practical high-vacuum triodes by1912, facilitatingthe use of narrower band transmissions andmaking the trans- missionof speechapossibility. The superheterodyne receiver was devel- opedby Armstrongand Fessendenin1912, andby 1933 Armstronghad developedthe concept of frequencymodulation. The First WorldWar provedan important vehicle for demonstrating the value of mobile radio inmilitarymaneuvers, especiallyfor use by spotter planes providingreports for artillery. The needfor smaller and lighter transmittersfor planes hastenedthe reductioninsize of radios to the extent that theycouldbe carriedina backpack. After the First WorldWar, the main impetus for developments came from broadcasting. The rapid increase inthe number of radio stations, especiallyinthe UnitedStates, resultedina commercializationof receiv- ers. It also resultedinefforts to coordinatethe use of radio spectrum, and the first international spectrum management conferencetookplace in Washingtonin 1927. This standardizedthe use of frequencies upto 1.5 MHzthe highest frequencythought to be of practical use for radio
20. 20. Some interesting history 7 transmission. Some earlyexperimentswere also undertakeninwhat has become known as private mobile radio. In 1921, the Detroit policeexperi- mentedwith voice transmissionto cars, but onlyin a one-way format, and the MetropolitanpoliceinLondonconductedasimilar experiment in 1923. By the SecondWorldWar, domesticradio receivers were relatively complex, with a high sensitivityandselectivityandthe abilityto receive radio stations from aroundthe world. During this period, two-way radios slowlydevelopedto the extent that some policeforces were equipping police cars withradio transmitters, but the high power consumptionand weight prohibitedtheir universal acceptance. The SecondWorldWar resultedinthe mass productionof mobile radio equipment in order to equipthe increasingnumber of militaryair- craft and ships. Infantry backpack radios became more popular and around 50,000 were manufacturedinthe UnitedKingdom during the war. After the war, the manufacturers were lookingfor amarket for their large productioncapabilityand startedto target the private mobileradio 2 (PMR) market. Mobile phones were fittedintaxis from 1950 onward, and the basic dispatchform of communications still usedtoday, and describedinmore detail inChapter 4, was developed. The next stepforwardwas the development of the transistor, dra- maticallyreducingthe size and power consumptionof radio systems and enabling mass productionof circuit boards to reduceprices. By1965, the first pocket-sizedmobilephones were produced, allowingmarket growth so that, for example, everypolicemancouldbe equippedwith a mobile radio rather than every car. The penetrationof PMR at this time startedto owe more to the licensingand regulatorypoliciesof the government rather than the equipment or market acceptance, and even today this has resultedina situationwhere the UnitedStates has more than four times the percentage penetrationof PMR thanthe UnitedKingdom. Regulatory policiesandthe implicationsfor the mobileradio engineer are discussed in more detail inChapter 12. 2. Private mobile radio is known as specialized mobile radio(SMR) in the United States and as private businessradio (PBR) in the United Kingdom, although PMR is by far the most widely used abbreviation and will be adopted throughout this book.
21. 21. 8 The Complete Wireless Communications Professional 1.3 Some key milestones in mobile radio history In describingthe first cellular systems it is important to rememberthat there is always a thindividing line betweenPMR and what is today known as mobile radio, typicallycellular radio systems. Bothare basically transmitter/receiver units;the differences typicallylie inthe serviceswith which theyare equipped. This is a topic to which we will returninmore detail in Section14.2. The PMR systems describedpreviouslywere typi- callytechnicallyable to provide some form of mobileradio service but were normallyprohibitedfrom interconnectionto the telephone system (and still are today) bothto prevent longer calls (whichare typical of con- nectingto a landline subscriber), whichwould therefore congest the radio spectrum, and to preserve the licensed(at the time monopoly) provision of cellular radio services. The first mobilephone service was introduced by AT&T in 25 U.S. citiesin1946 andcalled Mobile Telephone Service (MTS). It was not a cellular system because onlysingle cells were usedand opera- tor interventionwas requiredto set upcalls. This was followedbythe ImprovedMobileTelephoneService (IMTS), which also onlyused single cells but allowed automatic call set-upusingtone signaling. However, a short- age of spectrum and a lack of government interestincorrectingthis situa- tionpreventedthis system from providingany significant capacityandit made little impact. In Sweden, the first Europeanmobile radio system was introducedin 1955 byTeleverket;with modifications, this system existeduntil around 1981, whenits subscriber base hadgrown to 20,000users. Inthe United Kingdom, the first commercial system, calledSystem 1, was introducedin 1965 inLondon. It was expensive, had limitedcapacityand many draw- backs, but was still heavilyoversubscribed. The next variant, System 2, was never deployed, but System 3 reducedthe voice channel bandwidth from 100 kHzto 25 kHz, increasingthe capacity. This still fell alongway short of the demand. It was not until the early1980sthat cellular mobile phone systems were deployed, finallyprovidingthe dramatic increase in capacityrequiredto make mobile radio a mass-market product. The cellu- lar concept is describedinmore detail inChapter 2. Much of the work on cellular systems was pioneeredinthe United States. The cellular concept was first developedbyBell Labs in 1948, and its parent company, AT&T, lobbiedthe government for radio spectrum for
22. 22. Some interesting history 9 some time. In1977, this eventuallyresultedinan assignment inthe 800-MHzband, which is still one of the keyfrequencybands for mobile radio. Trials basedon this license tookplace until 1981andprovided very encouragingresults. The trials convincedthe U.S. regulator, the Federal CommunicationsCommission (FCC), that cellular was a viable concept. There thenfollowedalong periodduringwhich the FCC triedto deter- mine how to best assignlicensesfor cellular, the start of aprotractedand still ongoingprocess of selectingthe optimum way to assignlicenses that is describedinmore detail inSection12.3. Inthis case, awarding the licensesbasedonthe quality of the application(the beauty contest process)failedto workdue to the huge number of applicants, making evaluation highly difficult. The FCC overcame this usingthe lottery approach of selectinglicenses at random;it proved to be a highly unsatis- factoryapproachthat resultedinthe substantial tradingof licenses after the award. In fact, the UnitedStates tookover sevenyears to award all its licenses;all the delays resultedina high cost to the U.S. economyinfore- gone revenue and growth. U.S. cellular deploymentswere basedonthe AdvancedMobilePhone Service (AMPS)astandard still inwidespreaduse today. However, the AMPS standard only definedthe air interface;most operators useddifferent approachesto switchingand billing, with the result that roamingbetweendifferent regions inthe UnitedStates, of which there were more than90, was not possible. For amore detailed descriptionof cellular ingeneral andAMPS in particular, see [3]. Developments were also takingplace in other parts of the world. In the Nordic countries, the NordicMobileTelephone at 450 MHz (NMT450) was beingdevelopedwith the advantage that it would allow roamingto other Scandinavian countries. NMT900 was subsequently introducedbecause the capacityof the 450-MHzfrequenciesproved insufficient. Other Europeancountriesadoptedarange of systems, some developedwithin the countryand only usedfor that country. Others adopteda modifiedversionof AMPSknown as a Total Access Communica- tions System (TACS) that operatedinthe 900-MHzband. In Europe, this stage of mobile radio development, lastingfrom around1985to 1991, was generallymarkedby monopolyprovision. Most countriesonlypro- vided a license to the existingstate Post and TelecommunicationOrganization (PTO). Only the UnitedKingdom tookthe unusual steps of introducing competitionbyissuingtwo licenses andpreventing the PTO from owning either of these licenses (althoughtheywere allowed to take a minority
23. 23. 10 The Complete Wireless Communications Professional shareholding). Competitionis nowrecognizedas important inthe provi- sionof mobile radio services, andthe EuropeanCommission (EC) mandates that members must have a competitive mobile radio environment. 1.4 Recent history The recent historyof mobileradio since1991has beendominatedby the introductionof digital mobile radio andthe attempts to standardize third generationsystems. Keywithinthis historyare the roles of the GSM, CDMA technologies, and the thirdgenerationconcept andthe success(or otherwise) of standardization. It is this historythat is probably of greatest interest to the complete wireless professional because the lessons of this periodare likelyto be highly relevant over the comingyears. GSM Mobile radio since 1991has beendominatedby the GSM system. However, in 1991, it was far from clear that this wouldbe the case. Stan- dardizationof GSM startedin1982 withinthe ConferenceEuropeennedes Administrations des Postes et Telecommunications (CEPT)the European spectrum management body. Although CEPT had standardizedmany products inthe past, they were far from successful. Typically, CEPT stan- dardizationwas ledby engineers withlittle regardfor commercial reality and with a desire to see their ownideas incorporatedinto the standard. The standards that were developed, suchas X25, were oftenambiguous and resultedinvarious national implementations, preventinginterwork- ing in the form envisaged. The trackrecordof other standards bodies was also not good. In most cases, the development of standards tookso long that national solutions hadalready beendevelopedand the acceptance of the standard was low. Hence, there was little reasonto supposethat GSM standardizationwould be successful (andindeed, little reasonto suppose that other standards, suchas the digitalenhanced cordless telephone (DECT) or the terrestrial trunkedradio (TETRA), would be successful simply because GSM was). Standardization will be discussedinmore detail in Chapter 13, but an important lessonof historyis that standardizationof complete mobile radio systemsis morelikelyto fail thanto succeed. The reasons for the success of GSM are varied. One key point was that standardizationbegan earlybeforealmostanymanufacturer had startedto developtheir owndigital mobile radio standard. This prevented different manufacturers from goingdifferent waysand ensuredthat there
24. 24. Some interesting history 11 was sufficient time for the relativelyslowstandardizationprocess to pro- duce results before the product was required. The transfer of the stan- dardizationfrom CEPT to the EuropeanTelecommunication Standards Institute (ETSI) helpedto produce new rules about the manner in which the standardizationwould proceed, makingthe standardizationmore practical. Another factor was that the EuropeanCommissionwas in the process of mandatingGSM for use by cellular operators andthat many Europeancountries were comingto adead-end with the analog system, which was expensive due to its proprietarynature and unable to provide the additional capacity required. The inclusionof manufacturers intothe standards bodies was also a very important development comparedto previous CEPT standardizationthat included only the PTOs. The fact that there was no major competingstandard, particularlyfrom the United States, was also helpful. Finally, the vision of the participants, who fore- saw the increase insemiconductor complexity, andthe reductionincosts helpedprovide a standard that was state of the art at its time of completion. A more detailedassessment ofthe development of GSM is provided by [1]. The progress of GSM has beenfar from smooth, withmany delays en route. The standardizationproved much more complexthanoriginally anticipatedand tooka total of 13 years from the inceptionin1982to the final delivery of the Phase Two specifications withall intendedfeatures in 1995. Other complexstandards suchas DECT and TETRA have taken equally longperiods of time. Consideringthat the next generationof mobile radio standards will be even more complex, the longtime takento developthe GSM standards shouldnot be forgottenwhenremembering the timescalessuggestedbythose involved in thirdgenerationsystems. GSM is now installedinwell over 100 countries. Newoperatorsmak- ing a decisionabout the technologytheyshouldadopt oftenselect GSM because of the competitive supplyof equipment, the large base of exper- tise indeployingthe network, and the fact that users canroam to other countries. But clearlythis was not the case for the first operators, who experiencedsomethingof achicken-and-eggproblem;that is, once suc- cessful, the standardbecomesevenmore successful, but how does it become successful inthe firstinstance?In the case of GSM, operatorsin Europe were mandated to use the technology, so hence alarge volume of sales was guaranteed, spurringthe manufacturers to produce com- petitive offerings. Manyhave complainedabout this mandating. For
25. 25. 12 The Complete Wireless Communications Professional non-Europeanmanufacturers it provideda closedmarket, andfor opera- tors inEurope it removedchoice. The issues surroundingtechnology mandates are discussedinAppendix B. It couldbe argued that this combinationof astandard startedearly, a guaranteed market base, a lackof competitionfrom otherstandards, and full Europeancooperation, not to mentionthe careful andskilledworkof those performingthe standardization, is unusual and that standards are more likelyto fail thansucceed. Indeed, there is muchevidence that this is the case. Unfortunately, many engineers onlylookbackinto historyas far as GSM and conclude that all European standards will be successful. This is probablya rather selective use of history, as the followingexamples demonstrate. DSRR After the success of GSM, the EC startedstandardizing digital versions of almost everypossibleradio system includingPMR, short range, cordless, andpaging. Short range systems currentlyhave a small but steady market around the world. These systems do not use a base sta- tionbut communicate directlybetweenmobilesina walkie-talkie mode (see, e.g., [4]). They are widely usedin places suchas building sites and department stores where anumber of people workin a relatively small area. The EC decidedthat Europe neededa digital standard for these applications and startedthe digitalshort-rangeradio (DSRR) standardiza- tionproject. The scope of DSRR rapidlygrew from asimple back-to- back radio system to one where terminals couldrelaymessagesto other terminals and had securityfeaturesandcomplexdigital encoding. With GSM, the approach of making the phone highly complexhad worked because the economiesof scaleallowedthis complexityto be addedat lit- tle cost. The DSRR standards body failedto realize that DSRR would have much lesser economiesofscale;inany case, these advanced facilities were not requiredby the user, who valued low cost above all. The DSRR stan- dard was completedin1993, but no product has ever beenproducedto this standard. Instead, Motorolahas introduceda short-rangebusinessradio (SRBR) based on very simple analog transmissionandwith the user manually selectingone of threechannels. SRBR has proved very success- ful in meetingthe market requirementsfor simplicityand low price. Standards bodies are not very good at producingsimple standards. Withmultiple parties attendingthe standardization, the requirements from eachtends to get added to the total specification. It is difficult to restrict the capabilitieswitharguments about economicviabilitybecause
26. 26. Some interesting history 13 these are hard to quantify. Complexstandards can be advantageous when there is a large market but oftencause the failure of astandard. TETRA and APCO25 The Europeanstandard for PMR, TETRA may be moving down the same route. TETRA is discussedinmore detail in Section4.3, andits role infuture mobile communications is discussed in Section14.2. Yet again, the TETRA specificationhas proved to be highly complex, with TETRA providing many more facilities than requiredby most of the users. Manufacturers are counteringthis to some extent by building equipment that does not have all the features inthe specification, but whether this will be sufficient to generate alarge enough market to cover the development costsis far from clear. The UnitedStates hasa project similar to the EuropeanTETRA project known as APCO25 that is being standardizedwithin the Telecommunica- tions IndustryAssociation (TIA) TR8 committee(standardizationcommit- tees are explainedinmore detail inSection13.2). APCO has very similar goals to TETRA and, like TETRA, is targetedprimarilyat emergencyserv- ices users. The key difference betweenTETRA and APCO is that TETRA uses time divisionmultipleaccess (TDMA) while APCO25 uses frequency divisionmultiple access (FDMA)access methods are explainedin Section2.4.8. Telepoint Another interestinglessonis that of telepoint. The cordless and telepoint applicationis discussedinmore detail inSection5.2 and, again, Garrard[1] provides an excellent analysis of their history. Tele- point shows that modifyinga standard from itsoriginal purposeis danger- ous and that the success of mobile radio is not simplyborne out of adesire for anything that can communicate without wires but for a product pro- viding particular features. After the success of cellular inthe United Kingdom, the government was keento introduce more competition, and other industryplayers were keento enter the market. The cordless stan- dard, CT-2, developedfor indoorextensions to fixedlines, seemedto offer a way to meet these requirements. Bydeployinga large number of cord- less base stations incities, the telepoint operatorsthought that theycould provide a service similar to cellular withthe added advantage to the users that theycoulduse the same phone in their home, communicatingwith their home base station. Telepoint licenses were fiercelycontestedinthe UnitedKingdom and about 10,000base stations were deployedaround the country. Users, however, were not impressedbya service that could
27. 27. 14 The Complete Wireless Communications Professional onlybe used in cities, that couldnot accept incomingcalls, and where the handsets were just as expensive as cellular. It seemedobvious to many at the time that telepoint coverage is too expensive to provide and that coverage is a critical issue to users. The tele- point operators appear to have beenblinded by a desire to operatea cellular-typenetworkand the equipment manufacturers bya desire to sell more products. Evenmore bizarre was the entryinto the U.K. market of another operator (HutchisonRabbit) after the first four operatorshad failed. Estimates are that Rabbit never managed to have as many subscrib- ers as it did base stations. Historyhas shown that mobile radio canbe highly successful, but onlyif it provides the service that the subscriber wants. CDMA and TDMA One of the key debates of recent years has been whether code division multiple access (CDMA) or TDMA is the more appro- priate access scheme for mobileradio. This debate is exploredinmore detail in Section14.4. Infact, and little realizedbymany engineers, the debate has been less about the ideal access scheme andmore about whether operators shouldselect GSM or astandard developedby the U.S. company Qualcomm, now calledcdmaOne (previouslyreferredto as IS-95). The debate, althoughovertlytechnical, has reallybeenan issue of trying to market cdmaOne as better thanGSM. When cdmaOne was announced in the early1990s, it seemedunlikelythat it would succeed. The company that designedit was relativelysmall and little knownin the world of mobile radio. Standards were already establishedinEurope (GSM), and the U.S. standard (digital AMPS, or D-AMPS) was supported by the key manufacturers. However, today, cdmaOne is probably the worlds secondstandard, after GSM. Understanding how this occurredis an interestinghistorical lesson. Probably, Qualcomm couldonlyhave succeededwithcdmaOne in the UnitedStates. This is one of the few countriesthat: Allows any manufacturer to developa product that they can then offer as astandard (unlike Europe where there canonlybe one jointlydevelopedstandard); Has a large enough home market to produce goodeconomiesof scale inthe case that the standard is not acceptedelsewhere; Is sufficientlyadvanced in the development of cellular technologies that much of the rest of the worldlooks towardthem for leadership.
28. 28. Some interesting history 15 Clearly, Qualcomm would not have (and has not) succeededin Europe where the Europeanstandard was mandated. Key in the success of Qualcomm was their linkage with a number of other manufacturers and the desire of the UnitedStates to have a homegrownproduct rather than importingthe Europeanproduct and hence losingleadershipin cellular. Other countries suchas SouthKoreaalso rebelledagainst Europeandominance and insistedthat their operatorsdeploycdmaOne, with equipment purchasedfrom SouthKoreanproducers inorder to encourage local industry. Qualcomms astute use of partnerships and exploitationof the backlashagainst Europeandominance enabled cdmaOne to become akeyglobal standard. Politics andnational sensitivi- ties are likelyto playa key role in the future development of mobile radio standards. Thirdgeneration Third generationsystemsare intendedto be the replacement technologyfor existingsecondgenerationdigital systems suchas GSM. The concept of thirdgenerationis describedinmore detail in Chapter 11. Here, it is interestingto examine its progressto date. When thirdgenerationwas first announced, the key attribute was the abilityto provide service to all users, includingcellular, cordless, PMR, and satel- lite, all within the same system. One of the uses was data rates up to 2 Mbps in some environments, althoughat the start of the standardiza- tionthis was not a key requirement. Duringthe time that the standards bodies were making little progress tryingto agree onthe basic structure of thirdgeneration, GSM was quietlyevolving to provide service to nearly all users includingPMR, satellite, cordless, andothers. Suddenly, the thirdgenerationstandards committees realizedthat their requirements had mostlyalreadybeenmetwiththe exceptionof the 2-Mbps data. This now became the key requirement forthirdgenerationwithout any real indicationthat it was requiredby the users or that it was practical to provide given the limitationsof radio spectrum. Third generationstandardizationis ongoingand it will be interesting to watch the development of the standard. However, those doingthe standardizationseem to have trouble recognizingthat the directionof the standardizationmight needto change. Nor do they appear to have learnedthe lessonfrom GSM that suchstandardizationmight take 13 years or more. Muchof the standardizationis performedbyengineers who would do well to lookat recent historical experience.
29. 29. 16 The Complete Wireless Communications Professional References [1] Garrard,G., Cellular Communications:World-Wide Market Developments , Norwood, MA: Artech House,1997. [2] Petrakis,H. M., The Founders TouchThe Life of Paul Galvin of Motorola , Chicago: Motorola University Press,1991. [3] Bell SystemTechnical J. , Vol. 58, No 1, Jan. 1979. [4] Walker,J., ed., Advances in Mobile Information Systems , Norwood, MA: Artech House,1998.
30. 30. IIPART Mobile radio systems In this sectionthe keyfundamentals of mobile radio systemsare introduced. The intentionis not to write acomprehensive textbookcoveringall areas of mobile radio technologymanyexcellent books exist alreadybut to provide an overview. Armed with the knowledge in this part, the complete wireless professional shouldbe able to under- stand the key issues and know where to look to find more detailedinformationif it is required. This sectionassumes abasic knowledge of electrical engineeringprinciples andsome simple mathematical capability. For those who require amore basic introductionto the principles of mobileradio engineering, UnderstandingCellularRadio by W. Webb (Norwood, MA: ArtechHouse, 1998) is asuit- able introductorytext.
31. 31. 2CHAPTER The basics of mobileContents radio2.1 Introduction 2.2 Basic principles of propagation 2.3 Radio spectrum utilization Mans business here is to know for the sake of living, not to live for the sake of knowing.2.4 Basic system design Frederic Harrison 2.5 Packet and circuit transmission 2.6 Theoretical 2.1 Introductioncapacity of mobile radio systems The basic principles of mobile radio are best understoodbyfirst studyingthe propagation mechanisms by which the signal passes from the transmitter to the receiver. From propa- gation, this sectionexamines the shortage of radio spectrum and the complexsystem designs requiredto provide sufficient system capacity. Then the designof a typical system is examined, providing a goodunderstanding of the basics of mobile radio communications. The complete wireless professional needs to have an understanding of: 19
32. 32. 20 The Complete Wireless Communications Professional The likelyreceivedsignal strengthandthe effectof slowand fast fading and intersymbol interference (ISI) on the receivedsignal (these terms are explainedlater inthis chapter); The lackof radio spectrum and its implicationsfor mobile radio sys- tem design, including the means whereby spectrum canbe reused; The overall block-diagram designof a mobile radio system andthe needfor eachof the blocks; The capacity(i.e., the number of subscribers)that canactually be achieved from mobile radio systems. This sectionprovides abasic guide to these issues. 2.2 Basic principles of propagation An understanding of radio propagationis essential to the complete wire- less professional because the lossinsignal causedby propagationlimits the receivedsignal strength, impactingonthe quality of the receivedsig- nal. Many of the building blocks of mobile radio systems, as introducedin Section2.4, are usedsolelyto overcome the problems introducedby propagation. These building blocks include error coding, equalization, and to some extent the choice of multiple accessscheme. There are many detailedtreatisesonpropagation, and the topic is coveredina wide range of books suchas [13], in most cases inahighly mathematical fashion. Here the keyissues are introduced. If the receivedsignal strengthat a mobile radio is plottedagainst time, then the trace wouldshow a great deal of complexitythat wouldtypically take substantial effort to understand. To simplifythe analysis of radio propagation, engineers generallyconsiderthe receivedsignal strengthto be a compositeof threediscrete effects knownas path loss, slowfading, and fast fading. Although suchcharacterizationdoes not exactlyreflect reality, it has proved to be sufficientlyuseful andaccurate to model mobile radio systemsandis inwidespread use to date. Each of these sepa- rate elements is nowexamined.
33. 33. Thebasics of mobile radio 21 Pathloss Pathloss is the simplest of all the propagationmechanismsto understand and reflectsthe fact that the signal drops as the distance from the transmitter increases. Theoryshows that if the transmitter were in free space(i.e., some distance away from any object), thenthe signal would radiate in an expanding sphere from the point sourceof the trans- mitter. Since the surfaceareaof the sphere is proportional to the radius squared, the receivedsignal power at a distance d from the transmitteris proportional to 1/ d . Free-space loss cannot occur onthe Earthsince one2 half of the expanding sphere is under the ground which has a certain reflectionandtransmissioncoefficient dependingonthe material making up the surface of the Earthat that particular point. Of more relevance is the fact that there will be obstructionsonthe groundin the form of build- ings, hills, and vegetation, for example. These absorb and reflect the sig- nal, resultingina receivedsignal strengththat is muchlower than that predictedusingfree-spaceloss. Becauseof the complexityof modeling every building, general guidelines are adoptedas to the loss likelyto be experienced. Measurements have shown that in an urban environment, if the path loss is modeledas beingproportional to 1/ d , or in some cases3. 5 1/ d , then the results achievedbest reflect real life. Empirical models,4 suchas that from Hata, introducedinChapter 7, take the analysis one stage further by modifyingthe exponent accordingto the height of the mobile antenna, with the exponent falling by around 0.6 for eachorder of magnitude increase inthe height of the mobile antenna. This reflectsthe fact that as the mobile antennarises, buildings and other obstructions have increasinglyless effect and, hence, the path loss cancome closerto free space. There are a few isolatedcases where pathloss exponents lower than the exponent of 2 predictedbyfree space are experienced. These typically occur inconstrainedspaces, oftenincorridors inabuilding. Here, the sig- nal does not expand on the surface of asphere because the walls of the corridorcause the signal travelingtoward them to be reflectedbackinto the corridor. Because the signal is nowmoving forwardon a surface that is not expanding (assumingthe corridor stays the same width and height), theorywould predict that no loss insignal strengthwill occur. In practice, somesignal leaks throughthe corridor walls and exponents of around 1.6 to 1.8 canbe realized. Slow fading The word fading is usedto describe adropin the received signal strength, overandabove that which would be expectedbasedupon path loss. This loss occurs temporarily. There are two phenomenain
34. 34. 22 The Complete Wireless Communications Professional mobile radio that cause fading to occur, one that causes fades lastingof the order of a fewseconds and one causingfades lastingof the order of a few microseconds. The former is termedslowfading, while the latter is termedfast fading. To see aslow-fadingwaveform it would be necessary to take a set of measurements made by a mobile;to remove the effectsof path loss bycorrectingfor the distance from the transmitter at any point, using a formulasuchas that determinedbyHata; and thento filter the remainingsignal suchthat any high-frequencychanges, lastingless than a second, were removed. The resultingwaveform would typicallyshow a signal fallingby around 8 dB or so over a periodof afew seconds and then risingback up to the mean level. This fading phenomenonis causedby the receiver temporarilypass- ing behind obstacles that partiallyblockthe signal from the transmitter. A clear example of this is realizedwhendriving down a streetthat has detachedhouses betweenthe base stationand the mobile. Whenbehind the houses the signal strengthwill be reduced, whereas when between them the signal strengthwill rise backto the expectedlevel. The depth of the fade will dependon both the amount of loss of the signal inpassing through the building and the strengthof signals receivedby other mecha- nisms suchas reflection. The durationof the fade will depend onthe time it takes the mobile to traverse the building. Measurements have shown that if a plot of the slow-fadingwaveform is periodicallysampledandthe probabilityof any particular level of signal strengthplotted, thenthe resultswill followa log-normal distribution (i.e., a normal distributionplottedonalogarithmic scale)witha standard deviation of around8 dB. Fast fading Fast fading can easilybe seenonthe plot of receivedsignal strengthif onlya small portionof the plot, of durationsay1 sec, is exam- ined. Alternatively, it can be shown by filteringout all low-frequency changes in the receivedsignal. The cause of fast fading is multipathpropa- gation. In a complexurban environment, the mobile will receive many copies of the transmittedsignal eachtravelingvia a different path. Some will come directto the mobile, otherswill reflect offbuildings, cars, or other objects. The simplest arrangement of amobile receivinga direct and a reflectedsignal from the transmitter is showninFigure 2.1. In real life, the receipt of adirect signal is relativelyrare, andthe mobile is likelyto receive numerous reflections from all around. Each of these signals has followedadifferent pathfrom the transmitter to the receiver andso is likelyto have a different signal strength. The lengthof each path will also
35. 35. Thebasics of mobile radio 23 Reflected path Line-of-sight path Figure 2.1 Multipathpropagation. be different, withthe result that eachwave will take a slightlydifferent time to arrive at the mobile. This will result inthe phase of the carrier wave being different for eachof the receivedwaveforms. Imagine the simple case where there are two receivedwaves, bothof the same signal strength, and that one results from areflectionof amov- ing object, so that over time the distance this secondwave travels increases. The result is shownin Figure 2.2, where the top trace shows the first wave, the secondtrace the reflectedwave, and the thirdtrace the compositesignal as seenby the receiver. It is clear that at the point at which the waves are in exact antiphase there is complete cancellation of the receivedsignal, resultingina fade that extends to zero received signal strength. At a transmissionfrequencyof 900 MHz, the transition from constructive interferenceto destructive interferenceas shownin Figure 2.2 would take half a wavelength, approximately15 cm. Hence, eachtime the mobile moves a full wavelength, around 30 cm, it is likelyto pass througha fade. A mobile will travel this distance ina very short periodof timehence, the term fast fading. This can alternativelybe representedbya process of vector additionas shown in Figure 2.3, where the magnitude of the vector represents the strengthof the signal, and the angle of the vector from the origin(inan anticlockwise direction) representsthe phase differencebetweenthe referencesignal (the strongestpath) and the particular receivedsignal. There can be a number of vectors eachcorrespondingto adifferent receivedpath. In the case that the receivedsignals are exactlyinphase (as is the case at the start of Figure 2.2), the vector diagram consistsof two vectors superimposedontopof each other. In the case that the vectors are in exact antiphase (as is the case at the end of Figure 2.2), the two vectors are of equal magnitude but opposite polarity. Clearly, additionof the two vectors inthe first case wouldresult ina single vector alongthe originof
36. 36. 24 The Complete Wireless Communications Professional 2 1 0 - 1 - 2 2 1 0 - 1 - 2 4 2 0 - 2 - 4 Figure 2.2 The toptrace representsthe wantedsignal, the second the interferingsignal, and the thirdthe receivedwaveform. twice the magnitude, whereas in the secondcase the result of the addition would be no remainingsignal. Figure 2.3 shows the vector diagrams and resultingadditionfor the cases ofphase differences of 90degrees and 180 degrees. It real life, the prospectof exact cancellationis veryremote since no two waves will typicallyhave the same signal strength. However, the probabilityof a partial cancellationis veryhigh, with the result that the signal will fluctuate by as much as 40 dB from the mean level while the mobile passes throughfades. Figure 2.4 shows what a typical fast fading signal would looklike. Fast fading is also known as Rayleigh fading, after the physicist who developedthe statistics that canbe used to describe it.
37. 37. Thebasics of mobile radio 25 Above: Vectors corresponding to Above: Vectors corresponding to paths shifted by 90 paths shifted by 180 Below: The resulting signal, shifted by Below: The resulting complete cancellation of the signal45 with a magnitude of 1.41 Figure 2.3 Vectorial representationof multipathsignal reception. Mathematically, Rayleighfading is modeledby the combinationof in-phase and quadrature signals, bothhaving a normal distributionwith 2 variance s . Then the amplitude of the receivedsignal is given by () () ()ak ak ak=+ (2.1)22 i q where a and a are the in-phase and quadrature waveforms, respectively.i q The mathematics of derivingthe distributionof the amplitude are com- plex and detailedin[4]. In the case that there is no dominant path to the mobile, the probabilityof the signal having a particular amplitude a is given by
38. 38. 26 The Complete Wireless Communications Professional 10 5 0 - 5 - 10 - 15 - 20 - 25 - 30 - 35 - 40 Time (01 sec) Figure 2.4 Rayleigh fading waveform for a mobile moving at walking speedover 1 sec. () - a2 paa e = (2.2)2 s 2 sRayl eigh 2 A graph of this equationfor the case s = 1 is shown in Figure 2.5, where it can be seenthat the highest probabilityis for the signal to have no fading; that there is a significant probabilityof fades as deepas 22 dB; and that the tail of the graph asymptoticallyapproaches the x-axis, result- ing in a small probabilityof an infinitelydeepfade. If there is not a line-of-sight (LOS) path betweenthe transmitter and the receiver, and typicallythere is not, thenthere are onlytwo mecha- nisms by which the radio signal can propagate from the transmitterto the receiver, namely, reflectionand diffraction. At the frequenciesusedfor a 0.8 0.6 0.4 0.2 0 - 30 - 26 - 22 - 18 - 14 - 10 - 6 - 22 610 Signal strength (dB) Figure 2.5 Probabilityof aRayleigh fading signal strength.
39. 39. Thebasics of mobile radio 27 mobile radio system, diffractionresultsinveryhigh levels of path loss; hence, reflectionis byfar the most important phenomenon. Reflectionsimplyresultsfrom the signal reflectingfrom the surfaces that it encounters ratherthanbeing absorbedby them. The amount of reflectiondepends onthe reflective, absorptive, and transmissive characteristicsof the surfacethe wave encounters. Sheet metal pro- vides near-perfect reflection, while glass and paper provide near-perfect transmission. Materialssuchas brickresult insome reflection, some absorption, and some transmission. Typically, the surfaces encountered in real life are rough;hence, any reflections are diffuse, spreadingthe signal over a larger area but resultingina lower signal strength. Diffractionallows signals to bendover the edge of obstacles. Modeling diffractionis highlycomplexand typicallyis only tractable for the ideal- ized knife-edgecase where the obstacle encounteredcanbe considered to have minimal width [1, 2]. The key issue with diffractionis the angle through which the signal can diffract. Whena large angle of diffraction can be achieved, it is possible for the signal to re-formafter passing around an obstruction. Withlower angles, ashadow is formedbehind the obstruction. For the knife-edge, Figure 2.6 shows the variationof sig- nal strengthof a diffractedsignal withparameter v, while Figure 2.7 shows how the signal strengthfor agiven diffractedangle varies with frequency. The parameter v is given by ()2 = +h dd 1 2 dd (2.3)1 2 where d is the distance from the transmitter to the obstruction, d the dis-1 2 tance from the obstructionto the receiver, h the height of the obstruction, and the wavelength of the transmittedfrequency. For apossible GSM networkdeployment ina citywith d = 1,000m and d = 200m, at1 2 1,800 MHzthe parameter v = 0.28 h. Hence, using Figure 2.6, if h = 0 (i.e., the LOS path grazes the topof a building) the diffractionloss will be around 5 dB, while if the height is 10m (correspondingto atotal angle of diffractionof 3.43 degrees) the loss will be around22 dB. A little further analysis soonshows that diffractionangles of greater than1 degree are likelyto result ininsufficient signal strengthat the frequencies of interest. As Figure 2.7 shows, the loss is frequencydependent, but as the loss is
40. 40. 28 The Complete Wireless Communications Professional 5 0 - 5 - 10 - 15 - 20 - 25 V Figure 2.6 Variationof diffractionloss withparameter v. - 6 - 7 - 8 - 9 - 10 - 11 - 12 - 13 Frequency (MHz) Figure 2.7 Variationof diffractionloss foraparticular obstruction with frequency. already severe, the frequencyvariationis unlikelyto be the overrid- ing issue.
41. 41. Thebasics of mobile radio 29 Wideband channels There is a further difficultythat may be caused by fast fading. If the reflectioncomes from somefaraway object, thenthe delay of the reflectedsignal will be large. If this delay is greater thanthe time takento transmit a bit of information, thenwhen the reflectedsignal finallyarrives it carries different informationto the directsignal. The result of this is that the previous bit transmitted, or symbol, interferes with the current symbol, generatingaphenomenaknown as ISI. ISI is problematic inmobile radio systemsthat have transmissionbandwidths greater thanaround 100 kHz. In order to understandwhether the prob- lem might occur, consider the followingexample. Assume that the symbol rate is 100 kHz(broadlythis means that the bandwidth of the transmittedsignal will be around 100 kHz). The carrier frequencyis not a relevant factor incalculatingISI. (The symbol rate can be assumedto be equal to the bit rate for the purposes of this discussion.) If the reflectionis delayedby 1/100 kHz, namely10 s, then the reflected signal will arrive during the next symbol periodcausingISI. In 10 s, radio signals travel 3 km. Hence, if the reflectionoccurs from anobstacle 1.5 km past the receiver, ISI will occur. As the symbol rate increases, the distance requiredfor ISIto occur reduces proportionally. Withabandwidth of 1 MHz, the distance falls to 300m;and at 10 MHz, down to only 30m. Some of the wirelesslocalloop (WLL) systems discussedinSection5.3 have bandwidths in excess of 1 MHz, and hence ISI might be expectedto be problematic. To further explainISI, consider Figures2.8 to 2.10. Ineachcase the figures showthree lines. The topline is the signal receiveddirectlybythe mobile, where in this case a010101 datastream has beensent. The sec- ond line is the signal receivedwithsome delay, although here it has been assumedthat it has been receivedwiththe same signal strength. The bot- tom line is the compositefrom the firsttwo signals, that is, the signal that the mobile radio actuallyreceives. In Figure 2.8, the receivedsignal in the case that the delay is only one-fourthof abit periodis shown. The resultant signal is still clearlydis- tinguishable as a 010101 transmission;however, the lengthof time that the trace spends at the level correspondingto 1, or to 0, is reduced, increasingthe importance of accurate timingwithinthe receiver. In this case, the interference wouldbe judged to be Rayleighrather than ISI because the delayedsignal was delayedby less thana bit period. In Figure 2.9, the delay has been increasedto three-fourthsof abit period. At this point, the interferenceis verging upon becomingISI. Here, the
42. 42. 30 The Complete Wireless Communications Professional Figure 2.8 Receivedsignals inthe case that the delay is only one-fourthof abit period. Figure 2.9 Receivedsignals inthe case that the delay is only three-fourthsof abit period.
43. 43. Thebasics of mobile radio 31 Figure 2.10 Receivedsignals inthe case that the delay is a complete bit period. 010101 signal couldbe recovered, but onlywith a perfectlysynchro- nized receiver. Finally, Figure 2.10 illustratesthe case where the delayhas become exactlya bit period. In this case, it has clearlybecomeimpossibleto distin- guish the transmittedinformationunless someadditional intelligenceis employedinthe receiver. Nowimagine the case that the transmitteddata signal were 0011001100 . In this situation, it would actuallybe possible to distinguishthe transmitteddatauntil the delay was equal to two bit periods. Hence, ISI has the effect ofrenderingsome transmitteddata sequences unreadable, but not others. For this reasonit is sometimes called frequency selectivefading because if the transmitteddatacontaineda range of frequencies andthe frequencycontent of the receiveddata was examined(e.g., using a spectrum analyzer or conductinga Fourier trans- form), thenthe receivedsignal would appear to be missingcertainfre- quencies (suchas the frequencyof the 101010 patterninour example). ISI can be correctedthroughthe use of equalizers, whichare discussedin more detail inSection2.4.4.
44. 44. 32 The Complete Wireless Communications Professional 2.3 Radio spectrum utilization Radio spectrum is the one fundamental requirement for cellular and wireless communications. All radio communicationsrequireradio spec- trum, the amount requiredbeingapproximatelyproportional to the bandwidth of the informationto be transmitted. It was once said, Spec- trum is like real estatetheyjust dont make it any more.This isquite an apt description. Spectrum is quite like landthereis onlyalimitedsupply of it and some parts are more valuable than others. For example, certain parts of the radio spectrum have characteristics that particularlysuit cel- lular radio, and these parts of the spectrum are inmuch demand from all the companies who would like to be cellular radio operators. Radio spectrum is usuallymanaged by the government or their agen- ciesfor example, the FCC inthe UnitedStates. It is their role to ensure that the rights to use the spectrum are given out fairlyto all those who needit and to make sure that no two people are given the same bits of spectrum. Giving out radio spectrum fairlyis adifficult task. It is a little like tryingto fairlydistribute the welfare budget;everyone seems to have a valid claim and there is not enough to go around. Recently, some gov- ernments have resortedto sellingthe rights to the spectrum onthe basis that the personwho is preparedto pay the most must be the personwho needs the spectrum most badly. When enoughcellular spectrum for around sixcellular operatorswas auctionedin the UnitedStates in1997, 1 the government received$20B inrevenue. This shows just how scarce the spectrum is and how much people are preparedto pay for accessto it and is a topic to whichwe will returnto in more detail inChapter 12. What has tendedto happen in most countries is that certainparts of the radio spectrum, suchas the 900- and1,800-MHzbands, have beenset aside for cellular. This, and some other parts of the spectrum, have then beendivided typicallybetweenthree or four differentcellular operators, with the net result that eachoperator has beengiven somethinglike 25 MHz of radio spectrum each, normallypartitionedinto anuplink and a downlink band, and so the assignment is writtenas 2 12.5 MHz. The shortage of radio spectrum impacts uponmost of the designissuesof both 1. Actually, to be more accurate,it received pledges for over$20B. Subsequently,some of the largest bidders have defaulted on their payments with the result that the amount actually received is much less.We return to this issue in Section 12.3.
45. 45. Thebasics of mobile radio 33 the mobile radio technologyand of the networkand is a critical point tobe understoodbythe complete wirelessprofessional. To understand whether there is sufficient spectrum it is important to understand how the capacity of a cellular system is calculated. We first discuss the underlyingtheory. The capacityof a cell interms of the number of radio channels is given by m B= t BK (2.4)c where B is the total bandwidth (spectrum)assignedto the operator, B ist c the bandwidth requiredper call, and K is the cluster size or reusefactor. Assuming that the operator has beengiven a fixedspectrum assignment and that theyhave selectedatechnologywitha fixedbandwidth per call (as is typicallythe case), the remainingvariable is the cluster size. The concept of acluster is best understoodwiththe aid of Figure 2.11. First a few words about this figure. You will note the cells are shownas hexa- gons. This is somethingof ajoke in the cellular industrywhere nobody has yet seena hexagonal cell inreal life. Cells inreal life are moreor less circular (becauseof the way radio waves propagate, as explained E M A I F N B J G O C K H P D L Figure 2.11 A cluster of cells.
46. 46. 34 The Complete Wireless Communications Professional 2 previously), but unfortunatelycircular cellsdo not tessellate. A hexagon is quite close inshape to a circle anddoes tessellate, andso most cellsare drawn as hexagons to ease understanding. In Figure 2.11 cellsare namedA to P. Now imagine that it was decided to use frequency1 in cell A. The same frequencycertainlycannot be used in cells B, F, or E because theyadjoincell A, and so there would always be interference at the edge. Typically, it cannot be used in cells C, G, J, or I either because the interferencetherewill still be too great. However, it couldbe usedagain in cells D, H, K, O, N, or M. If we assign our frequen- cies onthis basis, the result might start to looklike Figure 2.12. Because onlyseven frequenciesare required, it is saidthat there is acluster size of seven. Like the cluster size, the other important concept is to understandthe number of radio channels requiredina particular cell. Typically, the mar- ketingdepartment will determine howmany subscribers theyexpectand the number of call minutes per busy hour that theyexpect the subscribers to make. Based onthis information, the engineers needto determinehow many radio channels will be neededto meet this demand. This concept is discussedinmore detail inChapter 7. Traffic levels are turnedinto radio channels via the Erlang formula. Erlang was a Swedish engineer workingfor the Swedish Post andTele- communicationsorganizationwho workedon traffic levels infixednet- works, but his work is equally applicable to mobile networks. Erlang developedtwo formulae relevant to mobile radio systems. The Erlang B formulais given by A NN !P = B N A nn ! (2.5)n= 0 where P is the probabilityof blocking, A is the offeredtrafficinErlangs,B and N is the number of traffic channels available. One Erlang is one con- tinuouslyusedtraffic channel, so duringany given period(sayone hour) if a user talks for half the time theywould be said to generation0.5E. If there are 10 users all talkingfor half the time, the total traffic loadwould be 5E. The Erlang B formulaapplies to nonqueuing, or blocking, systems. 2. Shapes are said to tessellate when they can be placed togetheron a flat surface without either overlapping or leaving any gaps.
47. 47. Thebasics of mobile radio 35 6 6 5 7 5 7 1 1 4 2 4 2 3 6 3 6 5 7 6 5 7 1 5 7 1 4 2 1 4 2 3 4 2 3 6 3 6 5 7 5 7 Figure 2.12 The cluster repeatedmanytimes. That is, when a user wants to make a call, if there are no channels avail- able, thentheir request to make a call is blocked. This much is true of cel- lular systems. However, the typical blockeduser will immediatelypress the send buttonagain in the hope of gettinga radio channel this time, and this retrybehavior is not allowed by Erlang B. Despite this, ErlangB pro- vides results that are very close to real life for mobile radio systemsandis widely used. It is normal to apply the Erlang formulato the traffic levels experiencedduringthe busiest hour of the day, or busy hour,in order to size the networkfor the worst case. The number of traffic channels, N, requiredfor agiven A and P canB onlybe solvediterativelyby substitutingvalues of N in the preceding equationuntil the desired P is reached. As a result, tables of Erlangfor-B mulae are available (a useful engineeringtool is aspreadsheetmacro per- formingErlang B calculations that canbe quite simplywrittenanExcel macro to perform this calculationis providedinAppendix A). A graph showing the variation of radio channels requiredwith Erlangs of traffic
48. 48. 36 The Complete Wireless Communications Professional and different blockingprobabilitiesis shownin Figure 2.13. Those wish- ing to know more about the Erlang formulashouldconsult [5]. There are a number of features to note about this figure. The first, unsurprisingfact is that as the requiredblockingprobabilitygets lower, more radio channels are neededto handle the same amount of traffic. For a typical mobile radio system, blocking probabilitiesof around2% are consideredacceptable, whereas for WLL systems blockingprobabilities below 1% are necessary. The secondis the slight steplikenature of these curves that is caused by the fact that onlyan integer number (i.e., whole number) of channels is possible. The thirdis that the number of channels requiredtends toward the number of Erlangs for high levels of traffic (e.g., usingthe 2% blockingcommoninmobile radio systems, for 1 Erlang, 5 channels are requiredanefficiencyof 20%;whereas for 19 Erlangs, 25 chan- nels are requiredanefficiencyof 76%). This is known as trunking gainthe more trafficthat canbe trunkedtogether, the more efficient the system. Erlang also lookedat the case where calls couldbe queued, as can happen in some fixednetworks. In this case, he derivedthe Erlang C equation, given by N ()()P pN pdelay = - 0 ()1! (2.6)pN 35 30 25 Pb 0.1%= 20 Pb 1%= 15 Pb 2%= 10 5 0 1 357 9 11 13 15 17 19 Traffic (E) Figure 2.13 Channels versus Erlangs for a range of blocking probabilities.
49. 49. Thebasics of mobile radio 37 where p is the utilizationper trunk, given by the percentage of time the trunk will be utilizedduring the busy hour; N is the number of trunks; and p , the probabilitythat there are no calls inthe queue waiting to be served0 is given by () () - 1 n Np Np1N N- 1 + - np!! 1 (2.7) Nn =0 Note here that insteadof the probabilityof blocking, all calls are queued until they are served, so there is a probabilityof delay. In such queuing networks, the grade of service is normallyspecifiedinthe form 95% of all calls must experienceadelay of less than1 sec.Erlang C does not describe mobileradio systems well because typicallycall requests are not queued. To date, there has not beena well-publicizedformulathat exactlydescribesthe mobile radio system takinginto account repeated pressingof the sendbutton; hence, ErlangB remains the standardfor- mula in use. 2.4 Basic systemdesign 2.4.1 System overview In this section, the general designphilosophies ofradio systems are explained. Each of the areas coveredis highly complexand there will typicallybe a number of specializedbooks andjournal articlescovering the relevant issues. Here, anoverview of the different areas is provided, along with details of where further informationcanbe discovered. Figure 2.14 shows a blockdiagram of a stylized digital radio system. Note that the orderingof blocks is important: speechcodingmust be performed first, modulationandmedium access last, and cipheringafter error cor- rection. Analogradio systems broadlydo not have any of these steps and are sufficientlysimple that theydo warrant further discussionhere. Each of the stages is now describedinmore detail. 2.4.2 Voice encoding Any waveform, whether it is speechor from adifferent source, canbe convertedfrom analogto digital by simply usingan analog-to-digital
50. 50. 38 The Complete Wireless Communications Professional Information source Information sink Source encoder Source decoder (speech encoder) (speech decoder) Error coder Error decoder Interleaver Deinterleaver Burst formatting Reformatting Ciphering Deciphering ModulatorRadio channel Demodulator Figure 2.14 Blockdiagram of a radio system. converter. All suchconverters sample the incominganalogsignal at peri- odic intervals and quantify the strengthof the signal at that point. The quality of signal representationinan analog-to-digital converteris meas- uredby the number of samples per secondandthe number of bits used to quantify the signal at that pointthe higher eachof these values, the more accuratelythe signal will be represented. However, equally, the higher both of these parameters, the greater the informationgenerated and the more bandwidth requiredto transmit the signal. Hence, abalance needs to be struckbetweenquality of representationof signal and band- width requiredfor transmission. In all cases, there will be an error whenthe speechis recreateddue to a differencebetweenthe digital waveform and the original analogwave- form. This difference is knownas the quantization errorand isshownin Figure 2.15.
51. 51. Thebasics of mobile radio 39 Sample points Speech waveform Quantization error Recreated digital waveform Figure 2.15 The speechsamplingprocess. The difficultyresidesinknowing what quantization error canbe tol- erated. The hearing process is complexand the ear is very sensitive to some errors while verytolerant of others. Mathematical measuresof errors suchas the root mean square (RMS) value of the quantization error tend to be poor predictors of the speechqualityas perceivedby the user. Instead, subjective listeningtestsare used, where a panel of users is asked to rate the relative quality of a number of different speechcoders, nor- mally ona scale of 1 to 5. The average score is usedto rate the speech coder and is known as the mean opinion score (MOS). Hence, the processof the design of speechcoders becomes one ofdesigningnewcoders and then subjectingthem to panel judgment. In the case of the simple coder describedpreviously, the samplingrate shouldbe set usingthe Nyquist theorem. This is a theorem that shows if a waveform is sampledat twice its highest frequencysine wave compo- nent, thenusing a knowledge of the amplitude and frequencyof the vari- ous components of the signal it wouldbe possible to rebuildthe original signal. Speechcontains frequencies upto around15 kHz depending on the speaker;however, most of these higher frequencycomponentsform part of unvoicedconsonants and can be dropped with onlylimitedloss of intelligibility. Indeed, the key components ofaspeechwaveform reside below 4 kHz. Hence, a samplingrate of 8 kHz is typicallysufficient for an acceptable voice quality(at least equivalent to existingtelephone systems).
52. 52. 40 The Complete Wireless Communications Professional Early speechcodersusedanagreedformat. They tookan analog-to- digital converter set at a samplingrate of 8 kbps and sampled8 bits each time, resultingina total data flowof 64 kbps. The quantization levels for eachsample were set ina slightlynonlinear fashionwhere theywere compressedat lowsignal levels and expanded for higher signal levels because experimentationhas shown that the ear is more subjective to errors at lowsignal strengththan at high signal strength. This speech- codingtechnique is known as -law PCM.PCM speechcodingis widely usedin fixedtelephone networks where bandwidth is not so critically constrainedas for mobilenetworks andwhere high quality is considered essential. Incidentally, this implies that there is generallylittle pointin producinga voice coder of higher qualitythan PCM since most calls will pass througha fixed networkat some point where their voice quality will be reducedto PCM levels. PCM is not an efficient way to encode speech. Speechwaveforms when consideredover a short durationof time (e.g., 20 ms) are highly repetitive andcan be clearlyseento be composedof the superpositionof a number of sine waves. By taking account of the predictabilityandperio- dicityin short samples of speech, near-identical qualitycanbe achievedto PCM codingbut at much lower levels of transmittedinformation. The most advanced speechcoderstryto model the manner inwhich the speechis generated. The speechcoders usedbydigital cellular systems fol- low this route to agreater or lesser extent andare known as parametric encoders becausetheyencode the basic parameters ofthe speechwave- form [1, 6]. In order to model humanspeechit is necessaryto understandsome- thing about the manner in which it is generated. Human speechcanbe consideredas consistingof two different types of soundvoicedsounds (vowels) with a regular, periodic structure andunvoicedsounds (conso- nants) with a more noiselike characteristicthat is less predictable. Exam- ples of voicedand unvoicedwaveforms are shown inFigures 2.16 and 2.17. The first taskof a parametric encoder is to decide whether each segment of speechis voicedor unvoiced. It then needs to determine: The equivalent filter coefficients that characterize the vocal tract of the speaker; The loudness of the speech; If voiced, the pitchinformationfor the speech.
53. 53. Thebasics of mobile radio 41 4,000 3,000 2,000 1,000 0 - 1,000 - 2,000 - 3,000 - 4,000 0 5 10 15 20 25 30 Time (ms) Figure 2.16 An example of voicedspeech. 800 600 400 200 0 - 200 - 400 - 600 - 800 0 5 10 15 20 25 30 Time (ms) Figure 2.17 An example of unvoicedspeech. The speechdecoderthenrecreates the speechdependingonwhether it was voicedor unvoiced in the followingmanner:
54. 54. 42 The Complete Wireless Communications Professional In the case of voicedspeech, aperiodic waveform is createdwiththe same pitchand loudness as the original and passedthrougha filter with the appropriate coefficients. In the case of unvoicedspeech, random noise withappropriate loudness is input into a filter withthe appropriate coefficients. In the basic form described, parametric coders are veryefficient at compressingspeechto as little as 2.4 kbps. However, there are two prob- lems that result inthe speechqualitybeing relativelypoor: Because speechis determinedto be either voicedor unvoiced, in the transitionperiodbetweenthe different typesof speechthe qual- ity will be poor. Interactions betweenthe soundsourceandthe vocal tract are ignored, resultingina poor characterizationof the speechinsome cases. A solutionto theseproblems is to use hybridcoders, also known as analysisby synthesis (AbS) coders. Insteadof simplyassumingthat the exci- tationis either aperiodic waveform or white noise, theytransmit more detailedinformationabout the excitationeitherusinga model of the pulse shape, size, and spacingor by selectingone of anumber of excita- tionmodels from acode book. Byselectingthe best possibleexcitation sequence, the aim is to minimize the difference betweenthe encodedand the decodedsignals. A generic blockdiagram of an AbS speechcoder is shown in Figure 2.18. AbS codecs divide the input speechinto frames, typicallyabout 20-ms long. For eachframe, parameters are determinedforasynthesis filter, which attempts to synthesizethe vocal tract, and then the excitationto this filter that most closelyreproduces the speechsignal is determined. This is achievedby determiningthe excitationsignal that when passed into the determinedsynthesisfilterminimizesthe error between the input speechand the reconstructedspeech. Hence the name analysis-by-synthesisthe encoder analyzes the input speechbysynthesiz- ing many different approximations to it. Once the analysis of the frame is completed, the encoder transmitsinformationrepresentingthe synthesis filter parameters andthe excitationto the decoder. At the decoder the given excitationis passedthroughthe synthesis filter setwiththe
55. 55. Thebasics of mobile radio 43 Input speech s(n) ^u(n) s(n) e(n)Excitation Synthesis -generation filter e (n) ErrorError w weightingminimization Encoder ^u(n) s(n) ReproducedExcitation Synthesis speechgeneration filter Decoder Figure 2.18 Generic model of anAbS encoder. transmittedparameters to give the reconstructedspeech. The synthesis model may also include a long-term predictor (LTP) that models the pitchin voicedspeech, allowingthe filter to shape the excitationsignal more accurately. The error-weightingblockshownin Figure 2.18 is usedto shape the spectrum of the difference betweenthe input speechandthe synthesized speechsignal in order to reduce the perceivedeffect of this error. This can be achieved by notingthat in frequencyregions where the speechhas high energy any error will be partiallymaskedby the speech, whereas in the frequencyregions where there is lowenergyany error will be highly noticeable to the listener. The error-weightingfilter emphasizes the noise in the frequencyregions where the speechcontent is lowprior to the error minimizationfunction, concentratingany errorsinto frequency regions where theywill be least noticeable. Suchweighting has been found to produce asignificant improvement inthe subjective qualityof the reconstructedspeechfor AbScodecs.
56. 56. 44 The Complete Wireless Communications Professional There are many different types of AbScodecs dependingonhow the excitationwaveform for the synthesis filter is chosen. The perfect AbS codec wouldpass everypossible waveform throughthe filter to deter- mine the excitationsequenceproducingthe best possible matchto the input speech. However the numerical complexityinvolved in passing every possible excitationsignal throughthe synthesis filteris typically intractable. For most codecs, somemeans of reducingthis complex- ity, without compromisingthe performance of the codecsignificantly, is used. There are broadlythree differentclassesof AbS codecs knownas mul- tipulseexcited(MPE), regular pulseexcited (RPE), and codeexcitedlinearpredic- tion (CELP). The differences betweenMPE, RPE, and CELP codecs arise in the manner in which the excitationsignal is generated. WithMPE codecs, the excitationsequence is givenby a fixednumber of pulses for everyframe of speech. The positions ofthese pulses within the frame and their amplitudes are determinedbythe encoder and trans- mittedto the decoder. Determiningthe optimal positionand amplitude for eachpulse is not computationallytractable;hence, asuboptimal methodof findingthe pulse positionsandamplitudes is used. Typically, about 4 pulses per 5 ms are used, leading to good-qualityreconstructed speechat a bit rate of around 10 kbps. The RPE codec also uses anumber of pulses to characterizethe excita- tionsignal. However, with RPE codecs the pulses are regularlyspacedat some fixedinterval and the encoder needs onlyto determine the position of the first pulse and the amplitude of all the pulses. Therefore, less infor- mationneeds to be transmittedabout pulse positions;so for agiven bit rate, the RPE codec canuse more pulses thanMPE codecs. For example, at a bit rate of about 10 kbps, around 10 pulses per 5 ms can be used in RPE codecs, comparedto 4 pulses for MPE codecs. This allows RPE codecs to give slightlybetter qualityreconstructedspeechqualitythan MPE codecs. However, theyalso tend to be more complex. Although MPE and RPE codecs canprovide good-qualityspeechat rates of around10 kbps and higher, they are not suitable for rates much below this due to the large amount of informationthat must be transmit- tedabout the excitationpulses. Inorder to go belowthis bit rate it is necessaryto use aCELP coder. WithCELP, a large number of poten- tial excitationwaveforms are prestoredinboththe encoder and the decoder. The encoder trieseachof the possibleexcitationsequencesand
57. 57. Thebasics of mobile radio 45 determines whichhas the lowest error. It thensends the number describ- ing the positionof this codewordinthe code bookalongwith a power level. Typically the code bookindexis representedwithabout 10 bits (to give a code booksize of 1,024entries) andthe gain is codedwith about 5 bits. Thus, the bit rate necessaryto transmit the excitationinformation is greatlyreducedaround15 bits comparedto the 47 bits used, for example, in the GSM RPE codec. Considerable researchis underwayto determine means to reduce the number of excitationsequences that need to be triedfor eachsample of speechso as to simplifythe designof CELP coders. For example, it has beenfound that the CELP codec structure can be improvedand used at rates below4.8 kbps by classifyingspeech segments into one of anumber of types (e.g., voiced, unvoiced, and tran- sitionframes). The different speechsegment types are thencodeddiffer- entlywith a speciallydesignedencoder foreachtype. The GSM full-rate speechcodec is anexample of a RPE codec operat- ing at 13 kbps. The input speechis split upinto 20-ms-longframes, andfor eachframe a set of eight short-term predictorcoefficients are found. Each frame is thenfurther split into four 5-mssubframes, andfor eachsub- frame the encoder finds adelay and a gain for the LTP. Finally, the resid- ual signal after bothshort- andlong-term filteringis quantifiedfor each subframe. This residual signal is decimatedinto three possibleexcitation sequences, each13 samples long. The sequence withthe highest energyis chosenas the best representationof the excitationsequence, and each pulse in the sequence has its amplitude quantifiedwith three bits. At the decoder the reconstructedexcitationsignal is fedthroughthe longterm and then the short-term synthesisfilters to give the reconstructedspeech. A postfilter is usedto improve the perceptual qualityof this reconstructed speech. An area of recent progresshas beenin variable-rate speechcoders. Such coders generatemore informationwhen the user is speakingand less informationwhenhe or she is not. Advanced variable rate coders typicallyhave around eight different ratesthat are selecteddependingon the amount of informationrequiredto characterize the speech. A20-ms sectionof speechis sampled, the amount of voice activitymeasured, and an appropriate coder rate set forthat section. The first of this type of coder was developed by Qualcomm and calledPureVoice. More recentlythe GSM standards committeesare workingonan advancedmul- tirate coder (AMR coder). Suchspeechcodersare extremelycomplex
58. 58. 46 The Complete Wireless Communications Professional and beyond the scope of this book;suitable references for further reading would be [1, 6]. Variable rate coders have the advantage that less interfer- ence is generatedwhen the user is generatinglessinformation;the reason why this adds more capacityis detailedinSection9.4. The rate at which codecs will improve infuture years is far from clear. At the time that GSM standardizationwas taking place in the late 1980s, 13-kbps voice coderswere available and the view was that soon6.5-kbps voice coders of comparable qualitywould become available. By the late 1990s, suchcoderswere stillnot available and the trend of rapidreduc- tionin codingrate seemedto have come to ahalt. However, there is still, in principle, muchreductionincodingrates that canbe achieved, and despite the current slowdownin progress, it seems likelythat further advances in speechcodingwill be found. 2.4.3 Secure transmission Most analogcellular systemshadlittle inthe way of encryption. Anybody with a scanner able to scanthe cellular frequencybands was typicallyable to pickup a cellular conversation. Althoughnot initiallya problem, even- tually there were anumber of scandals resultingfrom politicians having their calls intercepted, culminatinginan interceptedphone call from PrincessDiana, the so-calledsquidgy episode. Analogphone systems also startedto experiencefraudproblems. The earliest ofthes