GSM and UMTS (P16)

Chapter 16: Short Message and Data ServicesSection 1: The Early Years from mid-1982 up to the Completion of the First Set of Specifications for Tendering in March 1988 Friedhelm Hillebra

Chapter 16: Short Message and Data Services

Section 1: The Early Years from mid-1982 up to the Completion of the First Set of Specifications for Tendering in March 1988

Friedhelm Hillebrand1

16.1.1 The Mandate by CEPT and the First Action Plan of 1982

The mandate given to the Technical Committee GSM by CEPT in mid-1982 requested the

‘‘harmonisation of a public mobile communication system in the 900 MHz band’’.2Thisdecision took place during the very hot promotion phase for ISDN Therefore it is remarkable

to note that GSM was not defined as a ‘‘mobile ISDN’’ Instead the decision leaves the nature

of the GSM open, but requests the study of the interconnection with ISDN

The first action plan for the group GSM elaborated by the Nordic and Dutch PTTs wasapproved at the meeting GSM#1 in December 1982 It mentioned basic requirements for dataservices3

It is expected that in addition to normal telephone traffic, other types of traffic (non-speech) will

be required in the system However since such predictions concerning the user requirements will contain a great amount of uncertainty, a modular system structure allowing for a maximum

of flexibility will be necessary

The services offered in the public switched telephone networks and in the public data networks atthe relevant period of time should be available in the mobile system The system may also offeradditional facilities

These basic requirements regarding data services were very far-sighted They were nearlyforgotten for long periods due to priorities for telephony

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

2 GSM Doc 1/82.

3 GSM Doc 2/82.

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)

16.1.2 Discussions on Data Services from the Beginning of 1983 to the End

of 1984

The focus of the GSM work in the period from the end of 1982 to the end of 1984 was onstrategic questions and the requirements for the basic telephony oriented system, e.g therelationship between GSM and emerging analogue 900 MHz interim systems, radio aspects,speech coding and hand-held viability There was also a lack of data communication exper-tise in GSM Therefore the progress was slow in this area during this period Results were notsaved in permanent documents and were often forgotten after some time

During GSM#2 meeting in February/March 1983 a general discussion on ISDN and OSIand their applicability took place without firm conclusions

During the GSM#4 meeting November 1983 regarding ISDN it was clarified, that it wasnot possible to provide the full capacity of B- and D-channels but only the functions.GSM#6 in November 1984 received a report from a working party on network aspectswhich had met during the GSM meeting.4It proposed to use the ISDN concepts of terminaladaptors and interworking units for data services in the GSM system

16.1.3 The First Concept for Data Services Agreed in February/March

1985 (GSM#7)

During GSM#7 the Working Party 1 ‘‘Services’’ (WP1) elaborated a document on ‘‘servicesand facilities of the GSM system’’ which was endorsed by GSM.5This was based in the dataservices part mainly on an input submitted by Germany and France.6The output documentcontained a reference model for data services, definitions of tele-, bearer and supplementaryservices, network connections and types of mobile stations The reference model introducedterminal adaptor functions at the mobile station and an interworking unit between the mobileand the fixed network (the diagram is shown in Chapter 10, Section 1, Fig 10.1.1).The annexes contained lists of teleservices including the Short Message Service (SMS).SMS had three services: mobile originated, mobile terminated and point to multipoint Itforesaw a maximum message length of, e.g 128 octets, and an interworking with a messagehandling system Other non-voice teleservices mentioned were, e.g access to message hand-ling systems and to videotex, facsimile, and transmission of still pictures

A comprehensive range of circuit switched bearer services was listed:

† asynchronous duplex up to 9600 bit/s end-to-end

† synchronous duplex up to 9600 bit/s end-to-end

† asynchronous PAD access up to 9600 bit/s

The necessary connection types to support the mentioned services were defined by a set ofattributes

Another important prerequisite for the work on data services was the emerging consensus

on network functions and architecture of the basic GSM system as needed for telephony7.This allowed to start work on the reference configuration for data services8

16.1.4 Work in the Period from March 1985 to February 1987, when the Radio and Speech Coding Technologies Were Chosen

In the following meetings GSM and its working parties concentrated again on basic telephonyaspects GSM#12 (September/October 1986 in Madrid) discussed, whether the speech codecshould be transparent to DTMF signals It was qualified as desirable, but not as an essentialrequirement, since other solutions could be found

The priority for telephony left no work resources for data The situation was very critical inWP3 ‘‘Network Aspects’’, which was very busy with system architecture, mobility manage-ment, etc Also WP2 ‘‘Radio Aspects’’ was fully loaded with the evaluation of the differentradio techniques and had no time or capacity to work, e.g on channel coding for dataservices

16.1.5 Fundamental Decisions on Data Transmission in GSM#13

(February 1987 in Funchal)

The catalyst for some fundamental decisions on data services was the discussion of ments on the speech codecs with respect to transparency for DTMF and voice band data.During the selection decision of the voice codec at GSM#13 it became clear, that a transpar-ency requirement for DTMF and voice band data would lead to additional complexity and adeterioration of the speech quality On the other side it was realised, that data would needadditional protection, since the system would provide at cell boundaries only a bit error rate

require-of 1 error bit per 100 transmitted bits (compared to less than 1 in 1 000 000 in ISDN).GSM#13 decided that the speech codecs should be optimised for speech only, since GSMwas intended to be primarily a mobile telephony system It was further confirmed that thecapability of the system to carry DTMF and voice-band data was an essential requirement.Based on the proposal of an ad-hoc group9GSM decided that DTMF should be ‘‘transmitted

as a signalling message over the Dm channel’’ and that the DTMF tones should be ‘‘injectedinto the audio path at the receive end’’ ‘‘Data services should be supported by the GSMsystem The terminal equipment is connected to the mobile station via a terminal adaptorand the transmission is fully digital Terminal adaptors for the V-series are to be specified.’’10

As a consequence of this decision GSM agreed on the necessity of putting extra resources

on this task, since WP3 dealing with network aspects was not able to cope with it due tooverload and lack of specialised expertise

To start the work, a sub-working party under WP3 was created I was appointed as aninterim chairman, since I was one of the very few GSM members with data communicationexpertise.11 Furthermore GSM decided, that interim terms of reference should be agreedbetween the WP3 chairman and myself and that the group should meet independently ofWP3 and should not make use of resources presently used by WP3

Chapter 16: Short Message and Data Services 409

16.1.6 The Agreement on Strategic Issues and the First Set of

Specifications for Tendering from May 1987 and to March 1988

16.1.6.1 Setting Up of the New Group IDEG and Initial Discussions

The new group was called the Implementation of Data and Telematic Services Expert Group(IDEG) The first IDEG meeting on 20–22 May 1987 in Bonn was attended by 18 delegates.12Draft terms of reference and a set of working assumptions for the work as well as an actionplan were elaborated and agreed (see annexes to the meeting report)

The creation of IDEG coincided with the time when the rules for participation in CEPTwere relaxed GSM had asked the superior bodies to allow manufacturer’s participation in thedetailed technical work after the GSM radio decision Prior to that time, CEPT participationwas limited to representatives of the post and telecommunications authorities only The newrules allowed industry experts to participate in meetings by the invitation of, and as an advisor

to, a CEPT member organisation They were admitted as experts assisting the CEPTmembers, not as representatives of their companies They were also members of the delega-tion of the CEPT member they were assisting They could not submit documents in the name

of their company, but had to hand them to a CEPT member Industry participation was limited

to two experts per committee per country This restrictive situation was not sustainable Itlead later to the creation of the European Telecommunication Standards Institute (ETSI) andthe transfer of all standardisation work from CEPT to ETSI In the case of IDEG, there was amajority of delegates coming from industry from the beginning

The key tool to start the work was a set of working assumptions This concept had beenused in the decision on the basic parameters of GSM at GSM#13 in February 1987 in Funchal.The working assumptions were agreed as preliminary conclusions They were seen as open tochange, but the proponent of a change would need to prove their case This is essentially aprocess of ‘‘slowly drying cement’’ which allows the chairman of a group to find an agree-ment on a soft consensus conclusion and to start the process which leads to a firm consensusconclusion

Key working assumptions with regard to strategic issues were elaborated in the first ing:

meet-† Confirm telephony as the prime system application

† Co-use the telephony optimised system for data services to the maximum possible extent

† Ensure a high quality of service

† Avoid changes to the system architecture and support data services by add-on modules/functions, which can be implemented as options and can be dimensioned to meet the datatraffic needs

† The concept of ‘‘mobile office’’ lead to the requirement to have identical coverage andvelocity requirements from the mobile station as the telephony service

† Limit the additional complexity, since data was assumed as a small share of the systemtraffic

GSM#14 (June 1987 in Brussels) approved the proposed terms of reference for IDEG andconfirmed me as chairman

12 Report in GSM Doc 70/87.

16.1.6.2 The Agreement on Fundamental Aspects

In the second IDEG meeting (6–8 July 1987 in Heckfield, UK) a number of fundamentalaspects were discussed and brought to conclusion for the specification work

An important concept was to use existing data terminals and provide the normal interfaces

to them by a terminal adaptor function in the GSM mobile station This meant that theterminals needed to see a quality of service (bit error rate, delay) which is comparable tothe quality of service in fixed networks Quality of service is a critical issue for data services

in a radio network There are short interruptions of a connection during hand-overs or radiofades At cell boundaries 1 out of 100 bits is corrupted compared to 1 out of 1 000 000 inmodern digital fixed networks Therefore measures have to be implemented in a GSMnetwork to enhance the basic quality of service

The first measure is a powerful channel coding on the radio interface tailored to data whichadds checksums to detect and to correct errors (FEC ¼ Forward Error Correction) Thisreduces the bit error rate by several orders of magnitude

The short interruptions during handover and radio fades cannot be corrected by forwarderror correction, since they last too long if normal transmission delay values are requested.This type of problem can be tackled by protocols providing an automatic re-transmission oflost information For efficiency reasons such a mechanism can be applied only on blocks ofdata The protocol provides an additional checksum per block, which allows the detection ofwhether the received block is erroneous A sequence number allows the detection of lostblocks The receiving side confirms blocks received without detected errors and requests re-transmissions of blocks with errors or lost blocks

Such protocols are known in the fixed networks as the High Level Data Link Control(HDLC) family of protocols They cannot be used in a mobile network due to the bad basicquality of service Due to the great block size (e.g 1024 octets in fixed packet switchingnetworks) many blocks would have errors both during the first transmission and the re-transmission Therefore the throughput would fall to zero

Therefore a special protocol had to be designed which was robust to cope with the difficultradio network environment This protocol was called Radio Link Protocol (RLP) It wasmatched to the GSM transmission time slots, had a block length of 240 bits ( ¼ 30 bytes) andother means to secure the highest possible throughput

Services using only FEC were called transparent services Services using FEC and RLPwere called non-transparent services Both types of services provide a low bit error rate.Transparent services provide a constant throughput and a constant transmission delay Theyhave however interruptions of service caused by handover and fading Non-transparentservice secure the transmission of all blocks They re-transmit blocks lost or corrupted bythe interruptions caused by handover and radio fades During such activities the throughput isreduced and the transmission delay time is increased

IDEG refined, added details and agreed the set of working assumptions.13They covered allrequirements without adding complexity to the basic telephony oriented system.14 Keydetailed working assumptions were:

† Data for connection-less bearer capabilities (e.g for SMS) are transmitted on a controlchannel

Chapter 16: Short Message and Data Services 411

13

GSM Doc 83/87.

14

GSM#15 Report, Section 7d.

† DTMF information is carried on a control channel on the radio interface.

† Connection mode bearers are carried on one traffic channel, no multiplexing of low speedbearers on one traffic channel

† A reference configuration for the mobile station was agreed

† Interworking requirements and architecture were agreed

† A rate adaptation mechanism was selected

† One-way transmission delay less or equal 200 ms

† Transparent services use rate adaptation and forward error correction, no ARQ

† Non-transparent services use rate adaptation, FEC and RLP

† Bearer services must be capable of transmitting up to 9.6 kbit/s

I recall a very lively discussion on the maximum bit rate on full-rate channels Severaldelegates pleaded to limit the bit rate at 4.8.kbit/s in order to have a very good protection by a

‘‘heavy channel coding’’ They argued also that such a bit rate is more than sufficient for asingle user It was through the enduring efforts of Alan Cox and Ian Harris (both of Voda-fone), who succeeded in convincing the meeting that 9.6 kbit/s on a full-rate channel wasrequired in order to be future proof

In addition the action plan was revised The target was to complete the specification for theessential services (E2) which would be needed for a tendering purposes in early 1988 until theend of January 1988.15A list of the planned specifications was elaborated Major specifica-tions under IDEG’s prime responsibility were:

† GSM PLMN connection types

† Technical realisation of the SMS

† Transcoding for data and telematic services

† Rate adaptation (on several interfaces)

† Radio link protocol (on several interfaces)

† Terminal adapters (several specs)

† Interworking with circuit and packet switching data networks

† Service interworking

† In addition contributions to the work of several other WPs were needed:

† WP1 specifications on teleservices, bearer services and charging

† WP3 specifications on network architecture, interworking with PSTN/ISDN, numbering/routing/identification

† WP2 specifications on channel coding

GSM#15 in October 1987 confirmed the working assumptions and the action plan and

‘‘promoted’’ IDEG from a WP3 sub-group to a working party reporting directly to GSM.IDEG was renamed Working Party 4 ‘‘Data and Telematic Services’’ (WP4)

16.1.6.3 The Production of Specifications Needed for Tendering

In the following period WP4 concentrated its effort on producing draft specifications inaccordance with the action plan Two WP4 meetings were held on 26–30 October and on23–27 November 1987 in Bonn.16These meetings elaborated the first draft specifications

15 GSM Doc 84/87.

16 Report in GSM Doc 144/87.

Each meeting week had short opening and closing plenaries on Monday Friday morning,respectively From Monday afternoon to Thursday evening six drafting groups worked on

† SMS

† architecture and connection types

† terminal adaptor functions

† radio link protocol and coding

† interworking

† numbering and routing

Four specifications were completed for examination by GSM#16 in December 1987 Thesewere the most urgent ones, identified by GSM as particularly important for the tenderingactivities: 03.10, 03.41 (deleted later), 04.21, 07.02 Twelve specifications reached the status

† 04.21 rate adaptation at the MS/BS interface

† 07.01 principles on terminal adapters

† 07.02 terminal adapters for asynchronous bearer services

† 08.20 rate adaptation on the A interface

16.1.6.4 The Specification of the SMS

WP1 had produced a service description for the three SMSs:19

† Mobile originated/point to point

† Mobile terminated/point to point

† Cell broadcast

An additional input with a concept proposal for the technical realisation of the ‘‘mobileterminated/point to point’’ service came from France.20It contained a proposal for the func-tional architecture It proposed a new entity, the service centre in charge of:

† Dialogue with the user for message submission and status requests

† Handling of messages: storage, status, transmission

† Operation and maintenance functions

It proposed a layered function split between a message application subsystem and amessage transmission subsystem as well as several protocols.WP4 created a DraftingGroup Message Handling (DGMH) under the leadership of Finn Trosby (Telenor) in July

Chapter 16: Short Message and Data Services 413

1987.21This group was given the responsibility to deal with message handling access services(MHS)22and SMS and also to look for a common architecture for both groups of services.Great efforts and priority were put in the beginning to the MHS access services But overtime they eroded They were downgraded from teleservices to bearer services and completelydeleted in spring 1988, since normal bearer services were seen as providing sufficient func-tions to support this application In the background was also a development in the market, thatMHS according to CCITT X.400 were not a tremendous success.

But the group produced an initial draft specification GSM 03.40 ‘‘Technical realisation ofthe short message service’’ in November 1987.23It contained a first description of

† Service elements

† Network architecture

† Service centre functions and service centre network

† Functions in other network elements

† Routing principles for the message transfer between the mobile station and the servicecentre

† Protocols and protocol architecture

The SMS was defined between mobile stations roaming in a GSM network or roaminginternationally and a service centre, which had store and forward functions Both Point topoint services and the cell broadcast service were treated within this scope

I tried to interest the ISDN community to work with us on a compatible SMS service in theISDN This would have provided a standardised access to and from ISDN users But theinitiatives did not fall on fertile ground Therefore the SMS did not provide a standard for theaccess from and to fixed subscribers

A close co-operation with the Layer 3 Expert Group responsible for the layer 3 protocols

on the radio interface between the mobile and the base station was necessary, since the SMSrelated messages could be transmitted on normal signalling channels of the GSM system withlower priority than the signalling messages This would ensure, that short messages could betransmitted to mobile stations in idle mode or involved in a call The message length wasgiven a ceiling of 180 octets Later the final value was fixed at 160

The detailed technical work lead also to a proposal to refine the service descriptions24inFebruary 1988 This document covered also the international operation of the point to pointservices

The initial technical specification and the revised services description contained the initialresults achieved in the first year They provided a firm basis and framework for the laterdetailed work,25which is the basis of the tremendous success in the market

16.1.6.5 Achievements of the Data Group in its First Year from May 1987 to March 1988

IDEG/WP4 had despite the very late establishment after GSM#13 in February 87 met the firstimportant target: to complete the work needed for the planned tendering of ten operators on

21 Minutes of second IDEG meeting on 6–8 July 1987 in Heckfield, UK, IDEG Doc 58/87.

22 See Chapter 10, Section 1, Fig 10.1.2.

23 WP4 Doc 152/87.

24 WP4 Doc 85/88 rev 1.

25 See Chapter 16, Section 2.

29 February 1988, contributions to WP1, 2 and 3 as well as the specifications under WP4prime responsibility These documents cover the architecture and all functions needed for theasynchronous bearer services A broad stream of work in other areas was started and initialresults were reached This included SMS.26

WP4 had become a committed community of data communication experts who were new

to the GSM group WP4 had grown from 18 participants in the first meeting to more than 50participants in this period A very efficient and effective working and co-operation spirit hasbeen developed by the group

There were many valuable contributions to the work in this early period Key contributorswere, e.g.: Chritian Be´nard-Dende´ (France Telecom), Alan Clapton (BT), Alan Cox (Voda-fone), Graham Crisp (GPT), Alfons Eizenho¨fer (Philips), Ian Harris (Vodafone), MichaelKrumpe (Siemens), Thomas Schro¨der (GMD), Paul Simmons (Nortel), Finn Trosby (Tele-nor) and Hans Wozny (Alcatel SEL)

16.1.6.6 Continuation of the Work

Since I only took on the chairmanship of IDEG/WP4 on a temporary basis and had becomeresponsible for the implementation of the GSM network for Deutsche Bundespost Telekom inGermany (D1), I had to find a replacement chairman During the creation of IDEG and itsconversion to WP4, Graham Crisp had been particularly active in the development of theGSM architecture to support the wide variety of data and telematic services and interworkingscenarios that had been identified in the requirements As a result, I proposed Graham Crisp

as my successor However, the idea of an industry representative chairing a CEPT body wasnot welcomed with open arms by the GSM membership27 I, therefore, had a significant task

in convincing the members that an industry representative could be entrusted with the manship of a CEPT working party As a result, Graham was elected Chairman of WP4 byGSM#17bis in March 1988 He was the first colleague employed by a manufacturer tobecome a working party chairman in CEPT

chair-The first WP4 meeting to be chaired by Graham was held in Florence in Italy on 5–8 April

1988 Those early years of GSM data standardisation saw some significant changes in thestandardisation process, i.e from CEPT as an organisation open only to PTTs, via industryparticipation, to ETSI In 1989 the GSM work was transferred to ETSI and WP4 becameGSM4 The GSM phase 1 standards for service opening in 1991 were ‘‘frozen’’ in 1990 The

Chapter 16: Short Message and Data Services 415

26

An interesting snapshot of the state of the GSM standardisation in 1988 is contained in the proceedings of the

‘‘Third Nordic Seminar on Digital Land Mobile Radio Communication’’, Copenhagen, September 12–15 1988 During that seminar, a GSM day was held The GSM day and the rest of the seminar included a number of papers and presentations on the data and telematic aspects of GSM These included: Implementation of telematic and data services in a GSM PLMN, F Hillebrand; Architectural aspects of data and telematic services in a GSM PLMN, G Crisp, A Eizenho¨fer; The Radio Link Protocol (RLP) – a recommendation for the transmission of data in the CEPT GSM Public land mobile network, T Schroeder, I Harris, H Madadi; Rate adaptation and interworking functions for the support of data communication services by a GSM PLMN, A Clapton, C Gentile, S Thomas, G Ponte, P Simmons; Support of data transmission services in the European digital cellular 900 MHz mobile communication system, J.C Benard-Dende; Message communication within the GSM system, B Kvarnstrom, J Reidar Rornes, F Trosby.

27 Based on a proposal of GSM CEPT had allowed the participation of colleagues from industry in technical working groups But they were seen as part of a delegation of an administration (see guidelines in GSM Doc 3/87 not available on the CD ROM) This showed clearly the need to open a path for full participation, which was achieved in ETSI.

last GSM4 meeting chaired by Graham Crisp was GSM4#22 held in Vienna, Austria on 13–

17 May 1991

All quoted GSM Plenary documents can be found on the attached CD ROM The quotedGSM WP4 documents are not copied on the CD ROM They can be retrieved from the ETSIarchive

Chapter 16: Short Message and Data Services

Section 2: The Development from Mid-1988 to 2000 Kevin Holley1

16.2.1 Short Message and Data Services in 1988

In 1988, Graham Crisp of GPT took over the role of GSM WP4 chairman from FriedhelmHillebrand and continued the good work progressing towards the completion of phase 1 Istarted to attend GSM WP4 at the same time It was a small but focused group, about 25people splitting into four or five groups to do the detailed development in groups of 4–6people, but joining together at the end of the meeting week to agree results

In those days, the major focus of the group was in trying to get ISDN terminal equipment towork together with the mobile devices So the focus was more on what signalling needed to

be carried to and from the mobile across the network than trying to squash modem tones into acarrier designed for speech There were some proposals for ‘‘modem codecs’’ which wouldconvert modem tones, but it was felt that an all-digital system would give a much betterperformance and that ISDN would anyhow be the way of the 1990s by the time the systemwas installed

16.2.2 Radio Link Protocol

In order to carry data across the network efficiently, it was decided that two types of servicewould be offered, called ‘‘transparent’’ and ‘‘non-transparent’’ The ‘‘transparent’’ serviceswould be characterised by low delay but would suffer from data loss when the signal qualitywas low The ‘‘non-transparent’’ services would be characterised by high data integrity buthigher delay when the signal quality was low The ‘‘transparent’’ services were thus protectedonly by ‘‘forward error correction’’, by sending the important bits more than once, and alsousing predictive Viterbi techniques just as for the speech encoding

For the ‘‘non-transparent’’ services it was necessary to detect errors in the transmissionpath and retransmit packets which were received in error

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

In 1988 two separate simulation studies were undertaken to show how well the sion protocol, called Radio Link Protocol (RLP), performed RLP numbers each data packetand makes sure that each is received intact, and in the correct order When something goeswrong, the receiver has to tell the transmitter to go back a few data packets and retransmitfrom there These studies, reviewed by WP4, showed that the extra complexity required topick out one data packet and retransmit that one only, contributed significantly to the overallthroughput Whilst of course this made the RLP implementation more complex, it was felt to

retransmis-be retransmis-beneficial to make the most of the limited data capacity available

16.2.3 Cell Broadcast

During 1988 interest also increased in the cell broadcast service The basic concept was tomake text available to all phones in a particular area when they are idle (i.e not in any kind ofcall) The text could be general information or ‘‘teasers’’ which encourage users to makerevenue-generating phone calls The basic parameters for this service were set this year Anoperator could remove one signalling channel (SDCCH) and use this for Cell Broadcasts(CBCH) instead Each message would be transmitted with a header showing the intendedpurpose of the message, a serial number and a page number (to allow longer messages to betransmitted) Due to the capacity of an SDCCH it was only possible to send messages of 93characters of 7 bits, and one of these messages approximately every 1.88 seconds With thistechnical capability, it was envisaged that a whole range of information could be madeavailable to users, however as the service proceeded to market it hit problems with handsetimplementation and investment by network operators

16.2.4 Facsimile

In 1989 the attention of WP4 was drawn to the development of a reliable way to transmitfacsimile images over a mobile network The existing CCITT recommendation T.30 was theworld standard for facsimile machines, but this was designed for consistent data channels,which performed at the same level throughout the duration of the call With a mobile dataconnection the channel quality can vary dramatically within a few seconds It was thusnecessary to find a way to re-code the facsimile transmission so that the mobile channelinconsistencies were hidden from standard fax machines One of the major problems with thisapproach was the so-called Non-Standard Facilities (NSF) capability Fax manufacturers hadbeen able to detect the presence of a remote machine of the same manufacturer, and throughthe NSF negotiation in T.30 decide to use a completely different protocol, different modemtransmission rates or a variety of other techniques to make ‘‘like’’ fax machines exchangedocuments more effectively Whilst this offered the user some advantages on fixed lines,unfortunately it was impossible to support these features over a mobile channel

Even without NSF, fax machines use two speeds during transmission, one (higher speed)for the actual picture data and another (lower speed) for negotiation between each page All ofthis meant a detailed picking apart of the way fax machines work in order to re-create thisreliably at either end of the mobile channel

In 1990 it was discovered that, whilst the decomposition of the fax protocols for carryingover GSM was sound, the transport mechanism chosen to carry this over the radio was flawed