GSM and UMTS (P12)

Chapter 12: Radio AspectsSection 1: The Early Years from 1982 to 1995 Didier Verhulst1 12.1.1 From Analogue Car Telephone to Digital Pocket Phone CEPT took a very much forward looking de

Chapter 12: Radio Aspects

Section 1: The Early Years from 1982 to 1995

Didier Verhulst1

12.1.1 From Analogue Car Telephone to Digital Pocket Phone

CEPT took a very much forward looking decision when it decided to create, as early as 1982,the GSM Group with the mandate to define a second generation harmonised cellular system inEurope At that time, the true market potential for mobile systems was not known Also manytechnologies, which became key to the GSM radio design, were just emerging This is trueparticularly of cellular networking, digital signal processing and real-time computing

In fact, GSM work started when the telecom industry was experiencing a fundamental shiftbetween the ‘‘circuit switched analogue’’ world and the ‘‘packet switched digital’’ world.Microprocessors had just been introduced a few years before and it was the time when the first

PC was created The PTT administrations were introducing digital switches in their telephonenetwork to replace mechanical switches, and they were developing their first packet switcheddata networks We know today that this ‘‘digital’’ revolution ultimately lead to the Internet as

we know it today, but this was not at all clear at the time As we shall see, even the decision toselect a digital rather than analogue modulation was not obvious and it took almost 5 years to

be settled!

This technological ‘‘turning point’’ was also a wonderful opportunity for a young tion of engineers who had just learned in school the beauty of digital transmission and packetswitching, and had therefore the opportunity to contribute actively to the creation of a newstandard

genera-12.1.1.1 Marketing Requirements

In the early 1980s, the market for a second generation cellular system was perceived asprimarily radiotelephone in vehicle In a study called ‘‘Future mobile CommunicationServices in Europe’’, prepared for the Eurodata foundation in September 1981, PACTELintroduced the concept of ‘‘Personal Service’’ as opposed to ‘‘Mobile Service’’ and explainedthat the total market for vehicular mobile could be as high as 20 million in Europe while thedemand for low-cost hand-held service could ultimately reach 50% of the European popula-

1 The views expressed in this section are those of the author and do not necessarily reflect the views of his affiliation entity.

GSM and UMTS: The Creation of Global Mobile Communication

Edited by Friedhelm Hillebrand Copyright q 2001 John Wiley & Sons Ltd ISBNs: 0-470-84322-5 (Hardback); 0-470-845546 (Electronic)

tion But it was also concluded in the same report that a single system could not realisticallyserve both vehicle-mounted and portable terminals, because vehicle terminals would requirehigh power to insure continuous coverage and complex control function to ensure seamlesshandover at all speeds, while low-power hand-held terminals would use a network of non-contiguous small cells and should not be considered as truly mobile.

The work of GSM started initially with the objective of providing service primarily tovehicles, but it was recognised in the process that there should also be a proportion of portabledevices as these started to appear even in first generation systems Ultimately, the radiointerface selected in 1987 by GSM turned out to be efficient enough to allow a true personalservice, with continuous service anywhere, and a number of users largely exceeding the mostoptimistic early market projections The first commercial GSM terminals in 1992 werevehicle-mounted or bulky transportable terminals, but the terminals in use a decade laterare almost exclusively very compact hand-held devices!

12.1.1.2 Technical Background

The cellular concept was first described in the 1970s by the Bell Labs, and the first operational cellular network was launched in 1979 in Chicago In Europe, the NMT systemstarted operation in the Nordic countries in 1981 The key radio features of a cellular network,i.e its seamless handover between base stations and the reuse of frequencies between distantcells were being implemented for the very first time in commercial networks when GSMwork started We were therefore to design a second generation when the true performances ofthe first generation were not yet known!

pre-Around 1980, we were just seeing the first practical implementations of digital processing

in commercial domains such as microwave transmission and digital switching but, while theprinciples of digital encoding and modulation were already well known, there was still somedoubts about the amount of processing which could be implemented in cellular base stationsand, particularly, in mobile terminals

When the first descriptions of the GSM radio interface were published around 1987,showing the mobile stations monitoring in parallel a large number of logical traffic andsignalling channels multiplexed in time, we heard sometimes the comments that this wasfar too complicated and could never be implemented in low cost terminals Retrospectively, it

is amazing to see the amount of software found today in low cost devices such as toys and also

to realize that current GSM hand-held terminals already have more processing power thanmost micro-computers produced only a few years ago!

12.1.1.3 Building a New System

As research engineers, we faced the ideal situation whereby we had to define the mostadvanced system possible with a very limited number of constraints It was really the perfect

‘‘blank page’’ exercise We were given access to a completely new 900 MHz spectrum, up to

25 MHz in each direction, without any requirement to ensure upward compatibility with thefirst generation In fact, because there was already several incompatible analogue systems inpreparation in various countries of Europe, it became clear instead that only a truly innovativeand more efficient system could be adopted by all administrations

In comparison, the US mobile industry tried in the late 1980s to define their own digital

second generation system, with a constraint of upward compatibility, in terms of channelspacing and signalling protocols, with their analogue first generation ‘‘AMPS’’ This was seen

as a key advantage to allow the production of dual-mode analogue/digital mobiles, thusallowing the operators to digitalise their network progressively according to traffic demandwhile maintaining continuous coverage

As it turned out, this compatibility constraint delayed the introduction, in the ‘‘digitalAMPS’’ standard, of advanced features such as detailed measurements by mobile terminals,advanced handover control, and flexibility for innovative frequency allocations schemeswhile these features were in GSM from day one As a consequence it took several releases– and many years – for the American second generation cellular standard to seriouslycompete with GSM in terms of performance In fact, the digital AMPS standard never became

a real threat to GSM in the world market, and it was even challenged in its own market by theIS-95 CDMA proposal in the early 1990s

12.1.2 GSM Initial Work on Radio Specifications

At the beginning of GSM, some essential decisions had to be made concerning radio meters, including the choice between analogue and digital modulation In the case of a digitalsystem, there was also a number of key specifications to be produced concerning the trans-mission bit rate, the type of modulation, the multiple access principles, the source and channelcoding schemes, as well as the frame structure and the detailed mechanisms to handover fromone cell to another

para-12.1.2.1 Establishing ‘‘Working Party 2’’ (WP2) on Radio Aspects

At the GSM 03 meeting in Rome in early 1984, it was decided to set up three specific workingparties to progress on key technical subjects: services, radio and networking The secondworking party, WP2, was mandated to investigate radio transmission aspects I was asked tochair that group and we worked on a general model of the radio transmission channelapplicable to a digital mobile system The activity of WP2 continued in 1984-1985 duringspecific sessions in parallel to the GSM plenaries, and focused on early comparisons betweenvarious digital multiple access options Early 1985, I was leaving the French Administrationand I handed over WP2 responsibility to Alain Maloberti

During 1985, it was decided by GSM that due to the increasing amount of work required,the Working Parties would hold dedicated meeting every three months during the intervalbetween GSM plenaries The years 1985-1987 were very important for WP2 as they allowedthe selection of key parameters for the radio subsystem This process was lead by radioexperts from various PTT administrations involved in WP2, and it was supported by experi-mental programs involving manufacturers from various European countries In February

1987, when the main options had been decided and the work was focusing on the finalization

of the first release of GSM recommendations, ETSI allowed manufacturers and researchinstitutes to contribute also directly to the work of GSM plenary as well as working parties.The production of the GSM standard was therefore a truly European effort involving allconcerned parties of the industry

12.1.2.2 Analogue Versus Digital

On the comparison between analogue and digital options, it was by no means obvious at thattime that the quality of encoded speech could be equivalent – not to mention better – thanplain analogue FM when considering (i) the limitation in terms of bit rate to accommodate anaverage spectrum utilization of about 25 kHz per carrier as in analogue systems and (ii) thefact that gross bit error rate over the fading mobile radio carrier could be as high as 1022.Also, while we were evaluating digital options, analogue systems such as NMT were evenable to improve their capacity with channel spacing reduced from 25 to 12.5 kHz whilemaintaining a good quality of speech

From the early days, we did work however with the ‘‘working assumption’’ that the GSMsystem would be digital, but this assumption was only formally confirmed in 1987 when wecould prove, including with field trials, that a digital system would really outperform allanalogue systems

12.1.3 The Choice of the Multiple Access Scheme

One interesting advantage of digital transmission is that there exists a variety of methods tomultiplex several users over the same radio carrier, namely ‘‘Frequency Division MultipleAccess’’ (FDMA), ‘‘Time Division Multiple Access’’ (TDMA) and ‘‘Code Division MultipleAccess’’ (CDMA) In comparison analogue systems are restricted to the ‘‘one carrier peractive user’’ FDMA scheme I remember some meetings during the early years where thebasics of TDMA had to be explained to experienced radio engineers who had always thought

of radio resources in terms of ‘‘frequency carriers’’ and never in terms of ‘‘time slots’’ Thefact that several mobiles could coexist without interference on the same frequency connected

to the same base station was truly intriguing for a number of delegates!

During 1984, WP2 had already identified the three multiple access options FDMA, TDMA(narrowband or wideband) and CDMA (frequency hopping or direct sequence) which would

be the object of much debate until 1987 when the ‘‘narrowband TDMA/frequency hopping’’solution was selected Interestingly enough, several years later, quite similar discussions tookplace to compare wideband TDMA with CDMA in the context of UMTS

12.1.3.1 The Selection Process

GSM decided to launch a series of experimental digital systems to facilitate the selection ofthe radio transmission and multiple access scheme This trial activity was initially supported

by the French and the German administrations who decided to collaborate in the definition of

an harmonized system, and agreed in 1985 to focus their efforts towards the selection of adigital second generation system Accordingly, manufacturers were selected from both coun-tries to develop prototype systems implementing different digital radio subsystem concepts.Soon after, the Nordic administrations also decided to join and additional proposals weresubmitted to the experimentation program which took place in Paris from October 1986 toJanuary 1987, at the France Telecom research centre, CNET, under the auspices of the GSMpermanent nucleus

Eight different system proposals, corresponding to nine different radio subsystem solutions(one system proposal having two different multiple access solutions for mobile-terminated

and mobile-originated links) were proposed for the Paris trial Sorted out by multiple accesstype, they were the following:

GSM needed to compare on the same basis all these different radio subsystems tative measurements were therefore performed with identical environmental conditionscreated with propagation simulators, designed according to the specifications agreed by theCOST 207 Working Group on propagation The work of this group have been very importantsince all the previous mobile channels propagation models could be simplified assumingnarrowband transmission, while a more general model and simulators, applying as well forwideband transmission had to be elaborated In order to crosscheck the behaviour of the radiosubsystems with the propagation simulator and in a real environment, qualitative fieldmeasurements were also organised in Paris around the CNET

Quanti-Before the experimental program took place, GSM had decided that any new system wouldhave to satisfy five minimum requirements and that the comparison between multiple accessoptions would be based on eight additional comparison criteria

FDMA

† MATS-D (mobile to base), by Philips/TeKaDe, Germany

‘‘Narrowband TDMA’’

† S900-D, by ANT/BOSCH, Germany

† MAX II, by Televerket, Sweden

† SFH900, by LCT (now Nortel Matra Cellular), France

† MOBIRA, by Mobira, Finland

† DMS90, by Ericsson, Sweden

† ADPM, by ELAB, Norway

‘‘Wideband’’ TDMA (combined with CDMA)

† MATS-D (base to mobile), by Philips/TeKaDe, Germany

† CD900, by Alcatel SEL 1 ATR and AEG and SAT, Germany and France

Minimum requirements

1 Quality: the average speech quality must be equal to that of first generationcompounded FM analogue systems

2 Peak traffic density: the system must accommodate a uniform traffic density of 25 Erl/

km2, with a base station separation equal or greater than 3.5 km

3 Hand-held stations: the system shall be able to accommodate hand-held stations

4 Maximum bandwidth: the maximum contiguous bandwidth occupied by one mentable part shall be less than or equal to 5 MHz

imple-5 Cost: the cost of the system, when established, shall not be greater than that of any wellestablished public analogue system

The results of the Paris trial lead to two fundamental conclusions (for more details refer toDoc GSM 21/87 and GSM 22/87):

1 A digital system could satisfy all the minimum requirements set by GSM, and in fact adigital system would do better than any analogue system for all five criteria;

2 With respect to the comparison criteria, the radio experts of WP2 agreed on the followingcomparison table (Table 12.1.1) regarding the three main ‘‘broad avenues’’ of systemoptions

The conclusions of the WP2 work were:

1 A digital system can exceed the minimum requirements compared with an analoguesystem;

2 TDMA has advantages over FDMA;

3 Narrowband TDMA is preferred to wideband TDMA although both can meet the mum requirements

mini-There was a majority of countries supporting the choice for narrowband TDMA But it wasalso apparent that wideband TDMA was a viable option and, as recalled by Thomas Haug in

System comparison criteria2

1 Speech quality

2 Spectrum efficiency

3 Infrastructure cost

4 Subscriber equipment cost

5 Hand portable viability

6 Flexibility to support new services

7 Spectrum management and coexistence

8 The risk associated with their timely implementation

Table 12.1.1 Comparison results for the 3 main system options

2

GSM required that, for any selected digital scheme, performance with respect to criteria 1–6 would have to be at least equal to that of analogue systems and would be significantly better in at least one criteria Candidate systems were also compared with respect to criteria 7 and 8.

Chapter 3, a lot of additional technical and political discussions took place before GSM couldfinalize in May 1987 its choice for narrowband TDMA with frequency hopping.

12.1.4 Tuning the Details

Undoubtedly, the ability of GSM to agree in 1987 on the ‘‘narrowband TDMA’’ broad avenuewas a very important achievement But it was by no means the end of our efforts: rather it wasthe beginning of very intense activity which lead to the finalisation of the details to bespecified in the GSM radio interface The exact definition of the physical layer of the radiointerface by WP 2 was also a prerequisite before the functional specifications and the detailprotocol design of the logical layers could be progressed by WP 3

It was decided that the key radio aspects would be documented in the 05.xx series of

‘‘Recommendations’’ (later called ‘‘Technical Specifications’’) describing the different nel structure, the channel coding scheme, the modulation, the transmitter and receiver char-acteristics, the measurement and handover principles, the synchronisation requirements,etc.… In comparison to many other recommendations produced by GSM, the 05.xx servicesmay appear pretty thin: in total it was less than 200 pages! But each parameter specified wasoften the result of very thorough analysis and had to be supported by detailed simulations andexperimental measurements

chan-12.1.4.1 Channels Structure

It was defined that the ‘‘physical layer’’ would support a variety of traffic and signallingchannels Hence a large number of new acronyms were created: TCH/FR, TCH/HR, BCCH,CCCH, SDCCH, SACCH, FACCH, AGCH, PCH, RACH, etc.,… To make things moreconfusing, it was decided also that control channel BCCH/CCCH multi-frames would bemade of 51 frames (of 4.615 ms) while the traffic channel multi-frames would be made of 26frames (these two numbers being chosen as ‘‘prime’’ to allow a mobile in traffic having an idletime slot every 26 frames to ‘‘slide’’ across the complete BCCH multi-frame every 51 £ 26 ¼

1326 frames) The exact structure of each frame, composed of eight time slots of 148 bitseach, was also decided together with the allocation of each bit, including those for trainingsequences and those for actual data, with a specific channel coding and interleaving schemefor each type of traffic and signalling channel

12.1.4.2 Modulation and Channel Coding

The experimental program performed in Paris, and the additional analysis performed byWP2, had shown that particular care had to be given to the selection of the modulationand channel coding schemes It was clear also that the performance of the radio link,under mobile propagation conditions characterized by severe multi-path conditions, wouldalso depend strongly on the equalisation algorithms implemented in mobile station and basestation receivers In fact, as reported by Thomas Haug in Chapter 2, Section 1, the final choice

of the modulation scheme was difficult to reach and the initial preference of WP2 for ADPMwas finally changed to GMSK The reason for this choice was that the latter modulationmethod did not include any redundancy; therefore all the redundancy could be used forchannel coding, which was much more efficient by taking advantage of interleaving and

frequency hopping schemes, particularly for slowly moving terminals It was also decided byWP2 that the equalisation method should not be specified in the standard, thus leaving free-dom of implementation in the base station and mobile receivers As it turned out, for allmanufacturers, we saw significant improvements in terms of receiver performance betweenearly versions and more stabilised versions of their products (which ultimately performedbetter in terms of sensitivity than the minimum requirements set in the recommendations).GSM was right not to over-specify and to simply define minimum performances rather thendecide on exact implementation.

In addition to the choice of channel coding for full-rate speech 13 kbit/s, a substantial effortwas dedicated very early to defining several types of data traffic channels, including full-rate

at speeds ranging from 2.4 up to 9.6 kbit/s, and also half-rate from 2.4 up to 4.8 kbit/s withdifferent levels of error protection Retrospectively, we probably defined too many types oflow bit rate data channels as the demand for high speed became quickly dominant and there istoday very few mobile data applications with speed less than 9.6 kbit/s A few years later, aspart of the GSM phase 21 program, WP2 efforts would in fact be redirected towardsenhanced data coding at speeds of 14.4 kbit/s instead of 9.6 kbit/s, associated with the use

of multiple time slots (HSCSD and GPRS) More recently, the ‘‘Edge’’ modulation was alsoproposed to increase the bit rate over a single carrier above 300 kbit/s! It is quite remarkablethat a channel structure initially designed to squeeze many low bit rate data circuits on asingle carrier could be adapted later, without too many difficulties, to allow bursty traffic to betransmitted at a much higher bit rate

Concerning speech, GSM decided in 1993 to introduce half-rate coding as an option toincrease capacity In more recent versions of the standard, ‘‘Enhanced Full-Rate’’ (EFR) and

‘‘Adaptive Multi-Rate’’ (AMR) speech coding options were also defined to provide othertrade-off’s in terms of quality versus spectrum efficiency These various options are allcompatible with the initial definition of the GSM radio channels, and provide a good demon-stration of the flexibility we obtained with our digital foundation

12.1.4.3 Handover Mechanisms

In first generation analogue cellular systems, the decision to handover from one base station

to another is a central process made by the network based on ‘‘uplink’’ signal strengthreceived at base stations The second generation GSM system, being digital and time divisionmultiplexed, had the flexibility to introduce innovative schemes for handover In particularGSM Mobile Stations (MS), which are not transmitting or receiving all the time, have thecapability to (i) perform measurements of the ‘‘downlink’’ signals received from the serving

as well as the neighbouring Base Transceiver Stations (BTS) and (ii) report these ments regularly to the network The handover decision algorithm, implemented in the BaseStation Controller (BSC), utilizes both ‘‘uplink’’ measurements performed by the BTS and

measure-‘‘downlink’’ measurements reported by the MS This technique is referred to as ‘‘mobileassisted handover’’ which proved to be a very efficient and future proof feature of GSM.GSM being digital, the measurements of radio transmission performance could be basednot only on signal strength but also on estimates of the Bit Error Rate (BER) and FrameErasure Rate (FER) for speech WP2 dedicated big efforts to the definition of the precise radiomeasurements to be performed by MS and BTS, and to the detailed mechanism for thedecision to handover There was at that time a considerable debate on whether we needed

to specify the complete handover algorithm, but the GSM group took the wise decision todefine only the radio measurements, and not to specify the handover algorithm itself whichwas to become a proprietary implementation of each base station system manufacturer Indoing so, GSM left a lot of freedom for competitive innovation and, indeed, we saw a lot ofnew ideas introduced in the 1990s when the GSM networks had to cope with ever increasingtraffic demands It turned out that the initial radio subsystem specifications, and therefore allthe mobiles produced during the early years, could support advanced radio mechanismsintroduced later such as ‘‘concentric’’ cells and multiple layer ‘‘micro cell\umbrella cell’’handover.

12.1.4.4 Spectrum Efficiency Features

While selecting the parameters of the GSM radio subsystem, priority was given to overallnetwork spectrum efficiency as opposed to the efficiency of a single base station GSM hadindeed derived precise modelling of the maximum number of users per cell as a function ofvarious parameters such as the carrier spacing, the voice activity radio, the availability ofvarious diversity schemes, etc.,… It was concluded during the early definition stages that agood system, be it FDMA or TDMA, should always include a good level of inter-cellinterference rejection even if it would be with added coding redundancy and therefore lesscarriers per cell

Also GSM took the decision to introduce from day 1 Slow Frequency Hopping (SFH) as amandatory feature for terminals: this feature added some initial complexity but it turned out

to be very useful many years later when GSM operators were able to implement high-capacitycellular reuse strategies taking into account the interference diversity effects provided bySFH Examples of such innovative strategies include the utilisation of ‘‘fractional’’ reuseclusters whereby GSM cellular planning with SFH is based on a minimum reuse distancewhich is non uniform for all frequencies, or the option to use all the hopping frequencies inevery cell, controlling the interference level by the load of the cells

With features such as voice activity detection, interleaving, channel coding and frequencyhopping, GSM had introduced very early in its TDMA design some advanced functionalitieswhich differentiate GSM from other more traditional FDMA or TDMA systems Theseadvanced features also allowed GSM to compete well with the IS-95 CDMA standardwhen it was proposed in the early 1990s by American industry

12.1.4.5 New Frequency Bands

In the early 1990s the ‘‘PCN’’ initiative was promoted by the UK Administration to extend thespectrum utilisation of GSM from the 900 to the 1800 MHz band Accordingly, someadaptations of the radio subsystem were made to utilise 75 MHz of additional spectrum inthe so-called ‘‘DCS 1800’’ band New types of mobile stations were defined, with reducedpower to allow easier implementation of hand-held devices in such ‘‘Personal’’ Communica-tion Networks As it turned out, the majority of GSM terminals produced today are in factdual-band as many cellular networks have increased their capacity by combining the utilisa-tion of both 900 and 1800 MHz spectrum

The flexibility of the GSM standard to adapt to new spectrum was very attractive, and theexercise was reproduced again later to accommodate, in the US, the so-called PCS spectrum

at 1900 MHz (accordingly some terminals became tri-band 900-1800-1900) Other extensionbands have also been studied, including specific frequencies allocated in the 900 MHz bandfor railways applications, as well as more recently extensions of GSM also in the 450 MHzand the 800 MHz bands.

12.1.5 Group work towards a single standard

It would be difficult to name only a few people as the main contributors to the definition of theGSM radio interface

During these early days of GSM, there was a truly open collaboration between manyEuropean organisations, originally limited to PTT’s, then rapidly extended to their industrialpartners as discussed above The discussions were very open and in a true collaborative spirit

At that time, in comparison with more recent practices in standardisation forums, we werealso less concerned by the need to protect the Intellectual Property of our technical contribu-tions and, as a result, we were able to exchange rapidly and openly a lot of new ideas betweenmany contributors In that respect I am not sure that, today, the creation of an innovativestandard like GSM could be organised again so efficiently

In the radio interface definition, we could probably isolate contributions from many fic GSM participants But the more remarkable result is that, even with a large number ofinputs, the group was able to converge quite rapidly towards a consistent, well optimised andfuture proof foundation for its radio interface

speci-Chapter 12: Radio Aspects

Section 2: The Development from 1995 to 2000

1995 I was nominated and elected as the vice-chairman of the group, and I kept the positionuntil SMG2 closed its activities in 2000 However, Niels and I continued to serve in similarpositions in the 3GPP GERAN structure

When I was involved in the phase 2 activities of SMG2, a manager suggested to me, thatwith the finalisation of phase 2 the standardisation work could be stopped The estimation wasthat there would be no future enhancements, and it would be sufficient to produce a product.This prediction was not entirely correct The growing success of GSM inspired people to getnew functions and features for the concept Instead of a dry out of the work, a multitude ofideas where launched Looking at my notes during this time, I found it hard to describe thework in a comprehensive way along a time line, because of the multitude of activities.Therefore, I selected work items of the period, to structure the chapter I also consideredissues, which failed to find their way into a product, and also the work items, which amounted

to big leaps of the concept In 1995 the basic definitions of the GSM concept were 10 yearsold, however the system concept showed a remarkable flexibility of integrating new functionsand features never considered 10 years ago The key element of backward compatibilityrequires sometimes painful work around solutions, but finally enables mobiles from phase

1 to still work in networks which may have a release 99 functional content

1 The views expressed in this section are those of the author and do not necessarily reflect the views of his affiliation entity.

GSM and UMTS: The Creation of Global Mobile Communication

Edited by Friedhelm Hillebrand Copyright q 2001 John Wiley & Sons Ltd ISBNs: 0-470-84322-5 (Hardback); 0-470-845546 (Electronic)

A not insignificant part of the work was influenced by the UTRA selection process and thework on UTRA throughout 1998 before the work was continued in the 3GPP structure.But even after the work transition of UMTS to 3GPP, SMG2 did important work related tothe third generation For the conclusion of the 3GPP Release 99, UMTS contained a featurehandover from UMTS to GSM To have a useful functionality, this process has to work inboth directions SMG2 started therefore in 1999 to specify the conditions for a GSM toUMTS handover This feature was available on time when also the UMTS standard achieved

a level of completeness As an addendum in the 3GPP Release 4 the GSM to narrowbandTDD handover was incorporated

Other ETSI groups steadily worked on improvements for the speech coders The largestleap in the late 1990s was the Adaptive Multirate Codec (AMR) This codec has the property

to align its coding rate or resource usage according to the link quality The coder itself is atask of another group, however SMG2 was in charge of defining the needed channel codingsand the performance thresholds The AMR is part of Release 98, although a part of the workpenetrated into 1999

Not all activities were entirely covered by the SMG2 Sometimes other standardisationbodies took the lead for special needs In that sense the US T1P1 took care of the frequencyadaptation to 1900 MHz Later this work was integrated to the ETSI core specification Thesame applies for the location services LCS The initial requirement was set up by the require-ments of the US FCC authority T1P1 took the lead work to develop the concepts for GSMpositioning It was not a simple task, because from its properties, the GSM system is not inany sense prepared for such a feature Several positioning methods were allowed in theconcept, and after drafting the required CRs the discussion in SMG2 started together withthe people from T1P1 This review process was fruitful in that it catered for aspects ofcompatibility and removed ambiguities LCS based on the A-interface is part of the Release

98 functional content, although parts of the work spread largely in 1999

For Release 5, enhancements of the LCS are planned, to allow operation via the Gbinterface and the Iu interface This work is still ongoing

12.2.2 Working Methods

The working methods changed significantly in the second half of the 1990s, and I think it is anaspect worth mentioning It also reflected the technical progress at this time, and the increas-ing demand for data service capabilities This may illustrate the peak of the iceberg of serviceexpectations for the wider consumer market of the future

When I started my participation in GSM standardisation, just the secretary had a laptopcomputer A kind of 286 PC in the shape of a portable sewing machine, and likely of the sameweight All documents where copied in large quantities, to have sufficient copies for alldelegates More than one meeting followed the rule of copy availability, rather than theagenda Hosts were terrified by copy machine break downs, or late submission of inputdocuments A part of the meeting room or parts of the corridors were filled with tablescarrying piles of documents The delegates rushed off at every break to collect new availabledocuments Experts had little notepad sheets with a matrix for the document numbers tickingoff the numbers of the documents successfully collected

In the beginning revisions of documents resulted in a re-submission for the next meeting

Or the document was so handsome, that a handmade version could be generated at the

meeting An example of such a handmade document can be found on the CD-ROM.2BengtPersson, Franco Pattini and I drafted this document, and the final copy was done by me.Over the years more and more delegates used notebook computers, mainly to draft theirindividual meeting reports online, to make revisions of input documents, and to draft liaisonstatements The capability to generate new and revised documents during the meeting accel-erated the working speed of the committee However, this is one reason that the number ofdocuments in a meeting starts to increase Further, more participants showed up, thereforealso more potential contributors Another aspect, the split in working parties, especially in theplenary sessions, resulted in all the approved change requests being published a second time.The consequence was that the delegate carried huge amounts of paper as additional travelluggage Meetings in 1995 had about 100 documents for a week, today it is usual to deal withabout 500 documents in 1 week.

The introduction of e-mail exploders allowed a convenient submission of the documents;there was no need to collect a hardcopy Over a certain period, this was at least an improve-ment for the delegates, who just collected the papers only available as hardcopy For the host

of the meeting, there was no choice Each document of the meeting had to be copied,irrespective of the availability by the reflector It turned out that the copy cost was taking

up a significant part of a meeting budget In a meeting hosted by Siemens in 1997, with a largeplenary with 150 delegates dealing with GSM and UMTS, 80 000 pages were printed in 1week In 1999 the last paper meeting took place Now all meetings are electronic meetings,and CD-ROM or LAN does the document distribution For March 2002 it is scheduled tohave all meetings as wireless LAN meetings

Another characteristic of the work is the travel the delegates do In a faked ETSI tion document (see CD-ROM3) which was available at the SMG2 meeting in Madrid 1994,the term ETSI was defined as European Touring and Sightseeing Institute People outside thestandardisation work see the delegates mainly travelling Usually people not involved over-estimate the ‘‘pleasure’’ travelling In principle the work could also be done by having themeeting always in the same place But the ancient working procedures were introduced toallow the involved parties take care of the hosting of the meetings I can personally identifyone advantage of this principle In my memory I can connect discussions and documents todedicated meeting places and dates Therefore I find it quite helpful in retrieving data from

abbrevia-my mind

Personal relations are a key element of the work, and should not be underestimated in itsworth to building consensus decisions Although all delegates are in charge to represent theircompanies’ interests, they are also committed to a higher goal, to build a good standard Thiswork of intensive discussions, in coffee breaks, side meetings and evening sessions conductedover years helps to build up good relations with the other long-term delegates These relationsmay not apply for the quality of friendship, but it is far more than just working together

12.2.3 System Scenarios

The GSM system scenarios were a result of DCS 1800 activities in 1990 But in a lot of theactivities in the second half of the 1990s it paid off to have the system scenarios available Atthe GSM2 meeting in May 1990, the creation of specifications for DCS 1800 was decided

2 TDoc GSM 2 189/91.

3 Draft prETS 800 069-2 ‘‘Christmas Edition’’.

One evening after the plenary session, a subgroup stayed together to define system scenarios

to be used for the definition of the DCS 1800 radio parameters Some experienced people,having suffered long-winded discussions in phase 1 suggested this What we mainly did, was

to define coupling losses and the procedures on how to calculate the effects between MS/BTS;MS/MS; BTS/BTS As a rookie in this business, I was on the one hand fascinated by theability of these folks to define the scenarios; on the other hand I was puzzled how this wouldhelp in defining a 05.05 radio specification for DCS 1800 But it was becoming quite clearerfor me during the discussions in the first DCS 1800 ad-hoc meeting which took place inBorehamwood (UK) From the input documents I got a picture of how to use the scenarios,and in the following ad-hoc meeting in Bristol, I was able to give my first contribution based

on system scenarios The importance of the parameter definition is simply to enable dinated system scenarios, i.e two operators can co-exist in the same geographical area,without any co-ordination of spectrum, TX power, etc Of course, a little guard band isneeded between the operator band slots, however, GSM is also quite efficient in this view.With the conclusion of the work the documents were filed the usual way in PT 12 In theupcoming discussion about micro BTS a new need for scenario calculations was given Some

uncoor-of the initial documents were based on the original structure, where others started with theirown considerations In the discussions it was found most useful to base all new considerations

on the known structure Further, the idea was born, to collect all relevant background papers

in an ETR This was the start of the 05.50, which was compiled from a number of usefulbackground papers Over the years, with other frequency extensions, Pico BTS discussions,compact EDGE and mixed mode operations, the 05.50 was stepwise enhanced The 05.50 is avery valuable paper for people trying to understand how the parameters in the 05.05 werecreated However, the paper is somewhat difficult, due to the fact that the basic papers werenot updated This means that changes of parameters that for some reasons happened later, arenot reflected in the original calculations The best solution is to use the 05.50 for the calcula-tion principle, the calculated parameters will differ between 05.05 and 05.50 In merging theDCS 1800 and GSM phase 1 specification in the GSM phase 2, it was realised that somefigures did not fit together A re-calculation of the GSM figures according to the systemscenarios took place Some figures were aligned, others not, because the operators were afraidthat this would cause harmful relaxations to the system By experience it is best to use theDCS 1800 05.05 parameters and the 05.50 to understand how the principles work

The value of a system scenario document is common sense in the standardisation worktoday The GSM scenarios may be questioned in some assumptions, but today it has beenproved that they enabled a stable system operation

When the activities on UMTS radio, i.e UTRA started it was immediately decided to set

up a scenario paper Of course for wideband systems such as UMTS, the straightforwardworst case calculations are not suitable If you calculated UTRA system scenarios in such amanner, you would simply find that the system would not work Therefore the systemscenarios for UTRA are based on simulations, describing a defined loss of capacity which

is acceptable

12.2.4 Frequency Extensions

GSM railway was the frequency extension we had to cater in 1995 In the US T1P1 hadstarted activities to adopt GSM for the US market and this was going along with a GSM

definition for the PCS frequency band in the 1900 MHz range The initially independent workfrom T1P1 concluded in a reuse of channel numbers for 1900 MHz, which had already adefinition in the 1800 MHz band This is troublesome for multi-band mobile stations compris-ing 1800 and 1900 MHz The ARFCN is not an unique definition in this case.

Later the US version was integrated into the GSM core specification In 1998 a 450 MHzversion was defined, inspired by the idea of having a replacement technology for the NMT

450 system In 1999 the US 850 MHz band was defined, mainly for EDGE purposes Theavailability of additional spectrum in the 700 MHz domain required a further extension Withthe last frequency enhancement, it was becoming obvious that the channel numbering struc-ture of 1024 channel numbers was short Therefore activities were started, to generate a newbackward compatible channel numbering system, to overcome the limits of the actualscheme The principle currently under discussion, is based on the UMTS principle Instead

of an absolute number, an absolute start value is defined, and then the channel numbers asrelative information are used This concept which is not yet concluded, will remove allcurrent borders in the channel numbering system

12.2.5 Micro/Pico BTS

The growing success of the GSM concept called for solutions to handle the increasing traffic,especially in the cities There is of course quite a variety of spectrum efficiency improvingmethods But methods where changes of the mobile station are necessary have little impact in

a market with a high penetration of existing mobile stations After a short attempt to createmicro cell operation with special low power mobile stations, the common sense approach wasthat the existing mobile station should be suitable for all kinds of cells, macro, micro and picocells This of course called for the introduction of the micro cell base station definition Byuse of the system scenario concept the micro BTS TX power and the receiver dynamic rangewas adapted to the operational scenario needs Three proximity scenarios were used in thecalculations, and resulted in three micro BTS classes In 1997 the idea to adapt the BTS to aspecial scenario was brought up again as the pico BTS concept In contrast to the micro thepico is intended for indoor use only This calls for the consideration of a much tighter MS toBTS proximity, which was subsequently used for the calculations For the indoor use also thefrequency stability requirements for the BTS were to be reconsidered, due to the fact thatindoor significant Doppler effects cannot be expected

In UTRA we find the same strategy now Specifying a multi-purpose mobile station andthen defining different BTS classes

12.2.6 Mobile Station Power Classes

The GSM standard consisted of four power classes, later enhanced to five classes Class I wasthen removed from the standard, due to the consideration that a 20 W class is not state-of-the-art anymore The discussion about the impact of radio waves on the human body made itnecessary to remove this class, which was quite simple, because no MS of this class was everbuilt The class V (0.8 W), which was added later, is also a relic of the idea, that for specialscenarios low power MS may be needed Class II and III (8 W and 5 W) mobiles wherearound, especially as vehicle mounted devices Also car modules or adapters used this power