Tài liệu GSM and UMTS (P4) doc

After the testing of pre-operational phase 1 networks in 1991, followed by their commercial opening in 1992, the work of the GSM group focused on the development of the phase 2 specifica

Chapter 4: Consolidating GSM Phase 1 and Evolving the

Services and Features to GSM Phase 2 in ETSI SMG

(1992–1995)

Philippe Dupuis1

4.1 General

When the agreement on the selection of the GSM technology had been achieved some people thought that the rest would be easy We knew however that producing a set of specifications that would ensure the interpretability of mobile stations produced by any mobile terminal manufacturer, and network infrastructure produced by any manufacturer of network infra-structure, or the interoperability of different elements of the network infrastructure produced

by different manufacturers, would be a formidable task, particularly in the relatively short time frame agreed It is thus not surprising that, when the first set of specifications was released in 1990 with a total volume of about 5000 pages, it was incomplete and contained many imperfections or errors It was thus decided to call it phase 1 and to immediately start working on a second release called phase 2 After the testing of pre-operational phase 1 networks in 1991, followed by their commercial opening in 1992, the work of the GSM group focused on the development of the phase 2 specifications which actually lasted until 1995 Several new specifications had to be written, but more importantly, all existing specifica-tions had to be reviewed and checked for completeness, correctness and consistency as well

as their ability to evolve in the future The accent was on the management of the documenta-tion Precise procedures had to be put in force This work was possibly less glamorous than the work done during the previous period but it was certainly equally important to the success

of GSM, particularly outside of Europe Australia had been one of the first non-European countries to adopt GSM and some Australian colleagues were sitting at our meetings to make sure that we would deliver a workable product The volume of the work on phase 2 was formidable In addition we invented a concept for further evolution, known as phase 2 1 , and

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

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

had been put in charge of early work on UMTS The work which took place in parallel with these two other areas is reported in the following chapters Needless to say there was never a dull moment at our meetings

At the same time the GSM group had become an ETSI technical committee, it was renamed SMG and I took over from Thomas Haug as chairman in spring 1992 SMG had

to cope with the new ETSI environment This was an additional challenge While it entailed some bureaucratic burden the integration within ETSI has had some very positive effects as it enabled us to have fruitful exchanges with colleagues in other fields on the specification and testing methodology and opened new channels for mobilising the necessary resources The old Permanent Nucleus (PN) for instance had been transferred to ETSI headquarters as a

‘‘project team’’ It became PT12 and was funded by the ETSI budget, through which some financial support from the CEC was made available The various ETSI forums also provided opportunities to advertise our work and to attract contributions from companies represented

in ETSI

This section shows how a second GSM battle was won thanks to the excellence of the work done in the various sub-technical committees (STCs) of SMG, where hundreds of engineers from the European industry contributed their expertise, time and efforts This whole process was supported by PT12 They integrated all the contributions and checked them for consis-tency

4.2 SMG Within ETSI

ETSI had been created in 1988 The initiative came from the CEPT as an answer to the wishes

of both the CEC and the European Free Trade Association (EFTA) who considered European standardisation in telecommunications as an important step towards the establishment of a single market

ETSI pioneered the concept of involving in the creation of standards all concerned parties

In 1993 there were 316 full members from 25 countries comprising administrations, national standards organisations, public network operators, manufacturers, research bodies, service providers, users, etc It was thus a large organisation and had to be ruled with well defined procedures It was governed by a General Assembly (GA), a Technical Assembly (TA) and a Director General assisted by a permanent secretariat based at ETSI’s headquarters in Sophia Antipolis, in the South of France

The SMG chairman had to report to the TA The meetings of the TA, usually in Nice, were actually prepared by meetings of the Technical Committee Chairmen (TCC) which took place in Sophia Antipolis Thomas Haug, because he had been in charge long before ETSI existed, could take the liberty not to attend all of these meetings One factor which made things easier for him and later for me was that in the early years of ETSI the chairman of the

TA was Stephen Temple who was of course convinced that ETSI had to support effectively the work of GSM/SMG The Director General, D Gagliardi, as former chairman of the CCH

in CEPT also had some sympathy

It was obvious however that this situation would not last for ever and that I had to attend the TCC and TA meetings on a regular basis, and ‘‘play the game’’.2One issue in the TA was the approval of our work programme It was very difficult for SMG to comply with the proce-2

TDoc SMG 31/94, TDoc SMG 179/94 and TDoc SMG 508/94, chairman’s reports to SMG describe typical issues discussed at Technical Assembly meetings.

dures as our work programme included a large number of items and was constantly updated I was often criticised for this Some of the ETSI members, mostly manufacturers, were indeed concerned that SMG was going too fast and that it was difficult for them to follow in developing the products Because of this they would have liked to be able to control our work programme at the TA level They were typically those who had not initially believed in the success of GSM and were then behind schedule in the development of GSM products Another important issue at TCC and TA meetings was the funding of the ‘‘Project Teams’’, the teams of experts who worked fulltime at Sophia Antipolis to assist the Technical Commit-tees in the development of specifications We had been able to keep the same level of support that we had before with the CEPT Permanent Nucleus With a total at some times of up to ten full-time experts SMG was among the largest spenders in ETSI and I was also criticised for this Another criticism was that we were using our PTs for the programme management, while the rule was that they should deal only with the drafting of specifications The running cost of the PTs was paid by the ETSI budget itself funded by the members fees, or by special financial contributions from ETSI members interested in a particular project The major part

of such voluntary funding came from the Commission of the European Communities (CEC) interested in specifications needed for mobile station type approval testing At the most difficult time the GSM MoU and some manufacturers also contributed financially to make sure that PT12 would stay alive

The funding of PTs working in the SMG area was thus an annual headache for the chair-man of SMG and the co-ordinator of PT12, without mentioning the experts themselves who often lived in Sophia Antipolis with their family, and had to know in advance whether their contract would be renewed

It led to endless and passionate discussions with Peter Hamelberg who succeeded Stephen Temple as TA chairman and with the new director general, Karl-Heinz Rosenbrock With the success of GSM both in Europe and outside of Europe they realised the importance of our work and things were easier However, at the beginning I had to fight hard I still remember an argument with Peter Hamelberg in which he had said that we were possibly overdoing it when seeking a full European harmonisation, as he thought that mobile users were mostly using the service in their home country I told him, using words that Thomas Haug had often used in the past, that he had yet to ‘‘see the light’’ A few months later he started using a GSM handset and called me to say that he had ‘‘seen the light’’, which made our collaboration much easier Another problem we had with the ETSI rules was that, when GSM was adopted in many countries outside of CEPT, these countries wanted to send delegates to the SMG meetings This had started with Australia Later representatives of the American PCS 1900 operators became regular participants, and contributors, in the SMG meetings The objective of this participation was of course to fully align the GSM and PCS 1900 specifications, with a view

to the later introduction of tri-band operation We also had occasional delegates from Russia and Hong Kong ETSI meetings were in principle restricted to ETSI full members, belonging

to CEPT countries, and therefore, for each of them, I had to negotiate exceptions with the ETSI secretariat ETSI at large was however proud of the GSM world-wide success and tried

to become more open to non-European participation

4.3 Co-operation with the GSM MoU

SMG also had a close relationship to the GSM MoU Association The chairman of SMG

presented a report at each MoU plenary meeting.3Having chaired the GSM MoU in 1988 I was even invited at one time to sit in a small GSM MoU steering group including the past chairmen A representative of the GSM MoU also took part in SMG meetings Initially the MoU chairman attended our meetings, or was represented by John Moran, the MoU Secretary based in Dublin Later Fred Hillebrand became the full-time GSM MoU technical executive and was put in charge of the co-operation with SMG

This operation included several areas Firstly, the operators in the GSM MoU co-ordinated their views on the introduction in GSM of new services or features and the GSM MoU thus provided a most important input to SMG on what to include in the work programme The GSM MoU was also in charge of some ‘‘administrative’’ matters primarily concerning the operators, such as allocating various network or base stations codes, as well as managing the circulation of the security algorithm The GSM MoU had also accepted to fund the development of the type approval test equipment, which required a strong co-ordination with SMG itself in charge of the elaboration of the test specifications This co-ordination is described in more details below The GSM MoU was also in charge of developing specifica-tions of interest to them only, such as the one concerning the exchange of charging informa-tion between networks for roaming users, but this was clearly a separate area

During this period the GSM MoU grew from a European to a world-wide organisation but the excellent co-operation spirit remained what it had been in the early days

4.4 SMG Working Structure in 1993

4.4.1 Working Structure

Like the 11 other TCs of ETSI, SMG had set up some subcommittees (STCs), actually six of them They are represented in Figure 4.1 below In addition to them, two expert groups dealt, respectively, with test specifications and speech coding All groups, except SMG5 were contributing to the GSM standard, SMG5 being dedicated to the co-ordination of the work

on UMTS

4.4.2 PT12

Possibly the most important element of the working structure was PT12, the successor of the

PN who had moved from Paris to ETSI headquarters in Sophia Antipolis in the South of France It used to be known in SMG as PT12 although in ETSI’s terminology it was a set of Project Teams (PTs) contributing to the work programme of SMG, each one being assigned a specific task, as represented in Figure 4.1

This aggregation of PTs consisted of seven experts seconded to ETSI by ETSI members Each of them was in charge of assisting one STC, organising the meetings, keeping track of the work programme and documentation One of the PT12 experts acted as co-ordinator The first co-ordinator had been Eike Haase, who had joined the initial PN in Paris as deputy to Bernard Mallinder When he left in 1992, Jonas Twingler of Sweden took over until 1995, and then Ansgar Bergmann from Germany The role of the co-ordinator was fundamental He was

in the best position to propose improvements to the working methodology He had at the same

3 TDoc SMG 278/93, TDoc SMG 115/94 and Tdoc SMG 125/96 are examples of such reports.

time the responsibility of selecting the individuals proposed for the SMG PTs and maintain a good team spirit For the period from 1992 to 1995 which is considered in this section one must particularly acknowledge the work done by Jonas Twingler on GSM phase 2

4.4.3 Working Procedures

In SMG, like in other TCs of ETSI the detailed work was done in the STCs SMG itself was thus mostly co-ordinating the work of STCs, defining their work programme, approving their deliverables and deciding about policy issues

The GSM standard consisted then of 140 specifications and ten reports The primary responsibility of each of these deliverables was allocated to a particular STC and to a rapporteur Once the draft had reached a sufficient state of completeness and stability it was submitted to TC SMG for approval From then on, other STCs based their work on this draft This implied that any change must be brought to the attention of, and approved by,

TC SMG Proposed changes were presented in documents called Change Requests (CRs) complying with a precise format defined by PT12

To give an impression of the amount of work, Figure 4.2 indicates the number of CRs approved by TC SMG at its Meeting Nos 4–8 One notes the steady decrease of the number

of CRs relative to GSM phase 1, corresponding to a maintenance phase in which they were mostly of an editorial nature and related to testing For GSM phase 2 the number of CRs reached almost 300 at the SMG Meeting No 8 at the end of 1993 Later it fluctuated around

200 CRs per meeting Of course not each of them was actually presented and discussed at the meeting The fact that, despite this, any participant could object to it was considered as part of the ‘‘democratic’’ process as it allowed a small organisation not capable of sending repre-sentatives to all the STC meetings to have a say in the final decision

4.5 The Phased Approach Concept

The concept of the phased approach had been introduced during the course of the develop-ment of the GSM specifications In 1990 it became apparent that all the intended features could not be specified in time for the commercial opening of the first GSM networks in 1991

It was also recognised that there was still room for improvement in many areas of the signalling and procedures, as well as weaknesses to be corrected and enhancements to be included for the future proofing of the system

It was then decided to finalise a self contained version of the specifications supporting only

a subset of the planned services and called GSM phase 1, with phase 2, the complete, and at this time one thought final, version of the GSM specifications, intended for release some time later

4.5.1 Cross Phase Compatibility

A major subject for debate was of course upward compatibility The main requirement was that phase 1 services should remain available to phase 1 mobile stations accessing a phase 2 infrastructure The obvious way to achieve this was for the phase 2 infrastructure to be able to identify a phase 1 MS and, once it had done so, to revert to phase 1 signalling

Such a solution was of course feasible because added complexity on the infrastructure side was not considered to be an obstacle On the MS side on the contrary all efforts were then made to limit complexity and such an approach could not be used It was therefore agreed that the reverse upward compatibility objective, i.e that the phase 1 services be available to a phase 2 MS in a phase 1 infrastructure, could necessitate some upgrading of such infrastruc-ture

4.5.2 Transition

In the networks the transition from phase 1 to phase 2 was expected to proceed in a gradual way and indeed started very early as most of the products designed after 1992 were already compliant with some parts of the phase 2 specifications

On the MS side however, a clear distinction between a phase 1 mobile station and a phase 2

MS was needed as the infrastructure had to give a different treatment to both categories There has thus been a stepwise transition at the date at which the first phase 2 mobile station entered a GSM PLMN This of course was closely linked to the initiation of the phase 2 type approval and took place only in early 1996

For the reverse case of a phase 2 MS operating in a phase 1 network, report GSM 09.90 released in March 1993 had given guidance to the early PLMN operators as to the modifica-tions which should be introduced in their networks to remove possible incompatibilities This report was remarkably short showing that it was not a difficult issue

4.6 The Work on the Core Specifications

4.6.1 Phase 1 Maintenance

Before going into the work on phase 2 let us recall that SMG had also the responsibility of the maintenance of the phase 1 specifications, e.g to correct errors or ambiguities that would be discovered, normally this concerned the test specifications One interesting issue came at the time Slow Frequency Hopping (SFH) was activated in the networks Although all mobile equipment has been tested for SFH it was discovered that certain units did not behave well When it was discovered that a first mobile equipment manufacturer had put faulty units on the market, the other manufacturers were rather inflexible and did not want to seek a solution that would avoid recalling such mobile units Very soon after it was discovered that most units on the market had problems Then with a remarkable co-operation spirit, mobile equipment manufacturers, base station manufacturers and operators agreed to jointly identify solutions which could be implemented in the base stations to minimise the inconvenience Report GSM 09.94 was presented on this subject at SMG#15 in July 1995 That competitors could work together to jointly correct their respective mistakes is another demonstration of the value of open standardisation forums

4.6.2 Phase 2 Additional Services

It is often thought that the GSM phase 1 specifications contained a drastically reduced set of services This is not true The phase 1 specifications indeed included the basic voice services, almost all data services initially planned, fax services, a set of supplementary services, such

as call forwarding and call barring, and even Slow Frequency Hopping (SFH) The support of multiple encryption algorithms, to suit the requirements of the law enforcing agencies, was even added later The most likely reason why fax and data services were offered, and features such as SFH activated much later, is that operators and manufacturers attempted to spread the huge development efforts and costs required by the introduction of GSM over time They also had to concentrate all their available R&D resources on the development and debugging of the basic telephony service

The inclusion of additional services was thus not actually the main objective New phase 2 services, however, included now popular supplementary services such as calling line identi-fication, call waiting,as well as multiparty communicationsand closed user group It is also worth noting that many new or improved functions were defined for the Short Message Service (SMS) as well as for the Subscriber Identity Module (SIM)

4.6.3 Optimisation

Optimisation aimed at the correction of errors or limitations discovered in the phase 1 specifications, as well as the improvement of some key functions such as cell selection and re-selection for a Mobile Station (MS) in idle mode, and handover It also included the merging of the GSM and DCS 1800 When DCS 1800 had been introduced in the UK the specifications were urgently needed and for this reason were managed separately from GSM phase 1, leading to divergence in some areas In phase 2 it was decided to fully merge both, thus paving the way for the future dual band operation

4.6.4 Future Proofing

This concerned in particular the addition of hooksintended to allow the later introduction of new elements, in the same way the introduction of multiple encryption algorithms has been made possible

4.6.5 Specification Methodology

Such a huge amount of work could not have been achieved without a clear methodology SMG 3, the STC in charge of network aspects, particularly under Michel Mouly’s chairman-ship was more particularly involved in this

Initially there had been a debate as to whether GSM would follow the OSI layered model Some argued that it would make the signalling more complex but eventually GSM followed the OSI model More apparent at our meetings was the use of the Stage 1, 2 and 3 methodol-ogy To illustrate it in simple terms, if SMG was asked to specify a new service, the work would start with the drafting of a precise functional specification, which was called Stage 1, normally under the responsibility of SMG 1 On the basis of this specification one would then identify the required information exchanges between network elements, which were the subject of the Stage 2 specification Finally SMG 3 would produce the detailed specification

of the corresponding signalling messages known as the Stage 3 specification This was the basis on which SMG and its STCs worked as a well-oiled machine

4.6.6 The Half-Rate Speech Codec

When the speech codec, later known as ‘‘full-rate speech codec’’, was adopted in 1987 one believed that a more efficient speech codec with a much lower bit rate would soon be feasible Radio channels were thus designed in a way which would allow the later introduction of a half-rate speech codec, thus doubling the capacity of the networks, as well as the spectrum efficiency of the system It was thought that half-rate speech was needed to make GSM competitive on the world market

The rationale behind this plan was that the progress in the state-of-the-art of speech coding would produce more sophisticated coding algorithms achieving a greater data compression Such algorithms would of course require more computing power and memory space but the progress of microelectronics allowed the packing of more of this into a GSM handset Because this evolution did not proceed as fast as expected, or the GSM requirements were

to difficult to meet, workable solutions emerged only in 1992 A pre-evaluation was conducted which in early 1993 concluded that the two best proposals originated from Motor-ola and ANT Both companies then proceeded to fully develop their proposal and delivered a hardware implementation as well as a C-code description of their algorithm for the final round

of tests Both candidates had a complexity of 3–4 times the complexity of the full-rate speech codec It should be noted in this respect that SMG had decided that half-rate mobile stations should have the full-rate capability, and the codec had thus in fact to be ‘‘dual-rate’’

An argument arose at SMG#8 in October 1993 as Motorola had been late in delivering these elements but they were forgiven The big argument came at the next meeting, SMG#9 in Nice in January 1994.4The results of the subjective tests indicated that the Motorola codec was better that the ANT one but it did not meet all the agreed requirements Both candidates had also poor performances in high acoustical background noise and in tandeming conditions, i.e on mobile-to-mobile calls It was thus not possible to make a decision based on the previously agreed rule of the competition There was a strong Motorola delegation led by

Dr Mikulski from Motorola headquarters in Schaumburg, where I remember having met him

as early as 1982 There was certainly among us a willingness to reach a conclusion on a matter which had been on the agenda for too long On this basis we could possibly have forced a decision to adopt the Motorola codec, but Dr Mikulski demonstrated his co-operation spirit in agreeing with Dr Guels of ANT that both companies would work together to develop

an improved version of the Motorola codec Eventually its specifications were approved by correspondence in February 1995

What happened later is another story To my knowledge there has thus been no large scale implementation of the half-rate speech codec The market decided that there was more straightforward solutions to increase the capacity of the networks, in particular the introduc-tion of micro-cells One reason may also have been that a high quality full-rate codec was adopted by SMG soon after as we will see in Chapter 5

4.7 The Work on the Test Specifications

4.7.1 Back on the Type Approval of Phase 1 Mobile Equipment

The GSM phase 1 specifications were frozen in 1990, in order for the first networks to start service in mid-1991 Commercial operation was, however, postponed until spring 1992, because of the delay in the development of the mobile stations and test tools for their type approval Eventually type approval started in early 1992, but it was considered only as an interim type approval A new European type approval regulatory regime had indeed been defined, to be enforced starting on the 1 January 1994 GSM phase 1 full type approval had thus to be based on the European Directive on telecommunication terminal equipment (Directive 91/263/EEC) and at the same time to include a more comprehensive set of tests

4 TDoc SMG 524/93 deals with the decision making procedure to be followed at this meeting.

In 1992 and 1993 SMG produced the corresponding test specifications The final version of the corresponding Technical Basis for Regulation (TBRs) was approved at an extraordinary meeting held in Paris on the 28 May 1993 These TBRs were later incorporated into the Common Technical Regulations (CTRs) 5 and 9, the first ever CTRs of the new regime, adopted on the 28 September 1993

4.7.2 The Development of the Test Tools

To ensure the full interoperability of mobile equipment produced throughout the world it was important that all type approval laboratories use the same test equipment, at least that there exists a set of reference test tools

SMG was not in charge of the development of such testing tools, be they hardware or software The responsibility of SMG ended with the delivery of the test specifications Of course, the development of the specifications had to be done in close liaison with the entities

in charge of the development of the test tools, e.g giving the higher priority to the develop-ment of the test most likely to be performed for type approval

For GSM phase 1 the initiative had come within the GSM MoU where a group of operators and administrations agreed to fund the development of the System Simulator (SS) used for type approval testing and entered a contract to this purpose with Rhode & Schwartz of Germany The corresponding phase 2 activities were again initiated in the MoU even if there was an increased involvement of other actors, such as the test houses, and the CEC who provided part of the necessary funding

4.7.3 Issues Related to Type Approval Testing

Several difficulties as well as other important issues were debated regarding type approval testing

One difficulty was that tests of a quite different nature were required The verification of the radio characteristics on one side was mostly a matter of hardware, while on the other side the tests of the signalling protocols were essentially a matter of software In GSM phase 1 it was attempted to use a unique system simulator for all tests This was a challenge in a way that the expertise on radio hardware and signalling protocol was in general located in different companies Therefore for GSM phase 2 it was decided to use several machines

One issue was the depth of testing The appropriate depth of testing for type approval is a matter of judgement as it is known that no finite set of tests can guarantee the full confor-mance of the implementation of a complex protocol Operators were in favour of a large number of tests as they expected that it would save them future trouble in the operation of their networks Manufacturers were not far from the same position as they were afraid to have

to recall later large quantities of mobile stations needing a software correction But regulators were just interested in the verification of ‘‘essential requirements’’, such as user safety, electromagnetic compatibility, effective use of the spectrum, etc This was the subject of endless discussions leading to thinking that in the future it would be necessary to create an

‘‘industry certification’’ complementing the regulatory type approval

Another issue concerned the extent to which the development of the test specification should be based on the use of the Tree and Tabular Combined Notation (TTCN) test descrip-tion language TTCN had been developed in ETSI with the support of the CEC and its use was