Railways 2

111

GSM-R GLOBAL SYSTEM FOR MOBILE RADIO FOR RAILWAYS C0MMUNICAl"S

Trans-European Trunked Radio (TETRA) system. Both are digital mobile radio standards issued by the European Telecommunications Standards Institute (ETSI).

H. Hofestadt

In the view of railway requirements both have their merits and shortcomings. Essentially, GSM is well established and has already proven technical and economical viability in various countries all over the world, whereas TETRA IS still in the standardisation process. On the other hand, the TETRA standard offers PMR functionality which ISessential for railway operation. It is mainly because of the availability of the GSM system that the UIC decided in 1993 to use GSM as the future mobile radio standard for railways.

Siemens Germany

Transportation

Systems

Group,

ABSTRACT

The paper describes general requirements of European railways for a common universal mobile radio system. The reasons for selecting the Global System for Mobile Communications (GSM) as a basis are discussed, as well as necessary extensions to this standard to meet the railway requirements. The resulting GSM-R system offers new services comparable to those of Private Mobile Radio (PMR) INTRODUCTION

Currently, European railways use different mobile radio systems for various applications like train radio for mobile track-to-train communications, radio systems for shunting operations, and for maintenance purposes. Moreover, even internationally standardised applications like train radio, which has been standardised by the intemational railway organisation the "Union International des Chemins de Fer" (UIC), are oflen incompatible in different countries.

Time scales for introducing the new radio will probably differ between the European countries. German railways, for example, urgently need to replace their train radio system for obsolescence reasons alone. The most powerful driving force, however, is the cost reduction potential offered by radio based signalling. In 1990, therefore, German railways initiated the national project Diensteinfegrierender BahnmobiFununk (DIBMOF, Mobile Radio for Railway Integrated Services) to investigate options and approaches to consolidating many communication functions within one digital radio system The present paper is organised as follows. First, the railway requirements are discussed in some detail. Then the necessary extensions of GSM to a Global System for Mobile Communications for Railways (GSM-R) are presented. Finally, a possible way towards this eventual goal is outlined. RAILWAY REQUIREMENTS

Having in mind the EU directive on harmonisation of European High-speed Railway Networks it becomes clear that a unified mobile radio system (with respect to both different applications and countries) is needed. Such a new radio standard must not only integrate current applications but must also allow for more advanced services and railway applications, like Automatic Train Control (ATC), for example, currently being standardised in the UIC project European Train Control Systems (ETCS). Railway authorities have realised that proprietary systems are becoming increasingly expensive, especially when there are only a few suppliers for a country-specific radio system. Therefore, in order to induce international competition, a new radio system has to be based on a suitable, well established and widely used standard. Possible candidates have been identified to be the Global System for Mobile Communications (GSM) and the

As stated earlier a general requirementfor the new radio standard is the ability to integrate all or most of the current mobile radio applications of railways. Instead of listing all the specific application requirements we present and discuss them in the following more generalised form: vital railway applications, e.g.: ATP, ATC non-vital railway applications, e.g.: train radio shunting maintenance diagnosis positioning systems on-line passenger information on-line reservation

Electric Railways in a lltiited Europe, 27-30 March 1995 Conference Publication No. 405, @ IEE 1995.

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 24, 2009 at 04:29 from IEEE Xplore. Restrictions apply.

112 private passenger communications, e.g voice communications mobile data communications FAX transmission Requirements for vital applications

Controlling a fast moving train remotely from a wayside control centre is obviously an example for a vital application. It is generally a difficult task to get approval for such a complex system. The necessary level of confidence depends on the specific railway authorities in the individual countries. It was realised early on that the classical %Me box" approach to proving the vital operation of a train control system is not applicable in case of radio control links between trains and control centres. Essentially, in this approach the integrity of each bit of control information is proved along the whole way from a control centre to a train, for example.

Instead, the opposite approach is taken. The whole communication path between a train and a control centre including land lines, communication nodes, and radio links is regarded as a "black box" with no safety restrictions at all (1). The safety requirements are met in this approach by suitable end-to-end procedures based on cryptographical methods for authentication and integrity of data. These end-to-end procedures are supposed to run in safe environments, e.g., voting computer systems. This approach is currently being standardised through the "Comitee Europeean de Normalisationen Electrotechniclue" (CENELEC). To summarise, the requirement placed on radio due to vital applications is to provide a point-topoint data link from a wayside control centre to moving trains at speeds up to 500 kmlh. Both the availability of such a link and its quality have to be sufficiently high in order to prevent trains from unwanted stops.

Requirements for non-vital applications

A major requirement for many non-vital applications is to provide point-to-point voice and data links between mobile objects like trains and cars, for example, and wayside controllers/computers

In addition to point-to-point voice and data communications there IS a requirement for voice group calls and emergency broadcast calls. These

are well-known features of Private Mobile Radio (PMR) systems used by emergency services and police for example. The essential difference between an emergency call and a group call is that the former only allows the originator to speak. It generally requires shorter call set-up times and a priority scheme to pre-empt (terminate) on-going calls if no free channel is available. The call setup time must be such that at least an indication that an emergency call has been initiated be provided within a short period of time (approximately one second). Group calls are currently used by shunting and maintenance teams, for example. Moreover, there are railway-specific addressing requirements. Concerning the down-link case from a wayside control centre to a moving train the train registration data is used to set up a call by dialling the running number of the train independent of the locomotive which is currently pulling the train. At the same time it must be possible to call the locomotive of this particular train by its stock number. Concerning up-link calls, which are set up from a moving train to wayside control centre, for example, there is a general requirement for a location-dependent addressing mechanism. In case of an emergency, for example, the driver of a train hits an emergency call button of his train radio control panel which causes an automatic call set-up to the local wayside controller (e.g. signal man) in charge of that particular section. Requirements for passenger communications

As more and more people get used to their private GSM handsets, they will probably want to use it in trains as well. This includes voice, fax, and mobile data applications. Therefore, it is a requirement to extend the coverage of public mobile networks to trains. THE GSM-R APPROACH

In this section the above railway requirements are compared with available GSM services (including Phase 2 services). If there is no such service available suitable extensions to the GSM standard are introduced and discussed.

It is, of course, preferable to introduce as few modifications to GSM as possible. Moreover, all necessary extensions have to be formulated as generally as possible in order to open a broad range of possible new applications to GSM not

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 24, 2009 at 04:29 from IEEE Xplore. Restrictions apply.

113 necessarily confined to railway environments (e.g. PMR features for fleet management applications). This is the only practical approach to keep the costs of modifications low and thereby make them attractive to the GSM service providers. The opposite approach of leaving GSM as it is and introducing necessary features around the GSM system has turned out to be only practicable as an intermediate step towards an extended standard. Necessary extensions to GSM which have been identified as essential are currently being discussed and eventually standardised via ETSl in GSM phase 2+. The aim is to have the final GSMR standard agreed by 1995. Most of the extensions will be located in the Mobile Services Switching Center (MSC), the Base Station Controller (BSC), and the Mobile Stations (MS). Figure 1 shows the principle structure of a GSM-R system including wayside control centres for the two applications Train Radio (TR) and Automatic Train Control (ATC).

i'

1

A further problem due to high-speed operation is the performance of a handover between cells. Using standard algorithms, a large overlap between cells would be necessary. These problems can be overcome by introducing dedicated handover algorithms which take into account that the path of a train is predetermined. Automatic Train Control based on GSM

In case of high-speed trains, tracks are usually almost a straight line; i.e. it is just a matter of sufficient infrastructure (base stations) to realise a Rician environment. This forms the basis for providing highly available GSM radio links to trains. As stated earlier the radio link does not need to be regarded as a vital component German railway authorities have decided to use a standard Bearer Service (BS24) providing '2400 Bit per second for automatic train control purposes. For each train a pennanent radio link is established between the train and a wayside control centre. For efficiency reasons half-rate channels will be used for this purpose as soon as available. In order to guarantee that a radio channel for ATC is available in the next radio cell during handover either sufficient network capacity or a priority scheme has to be provided, and will be described in a later section.

Figure 1: GSM-R System Structure Non-vital applications based on GSM High-speed Performance of GSM

As GSM is essentially made for point-to-point GSM is specified for speeds up to 250 km/h. Fortunately, this holds under multipath propagation conditions as described by the Rayleigh model. Various measurements performed in the DIBMOF project have shown, however, that in typical railway environments the direct path between transceiver and receiver is dominating, i.e. these environments are strongly Rician with typical Rice parameters in the order of 10 dB to 20 dB in tunnels (5),(6).

communications all categories of point-to-point railway communication requirements can be realised in a straight forward manner. Useful services are, for example, the teleservices Telephony, various Short Message Services, and fax group 3 services. For data communications various Bearer Services with data speeds between 2400 and 9600 Bit per second are available (8).Additionally, in GSM phase 2+ a General Packet Radio Service (GPRS) will be available which is more efficient for many applications.

Based on these measurements detailed link srmulations have been performed (7) including diversity schemes. A major conclusion is that receiver diversity gives a significant gain and is most effective in the Rician case. A trackside signal to noise ratio of approximately 10 dB without diversity should provide for adequate GSM link performance up to 500 kmlh.

However, the missing PMR functionality of GSM causes problems with voice group calls and emergency calls for example. In principle, a group call could be emulated by the GSM Multiparty Service. This is, however, a time consuming method: one must first establish a point-to-point communication, put it on hold to establish a second call, and then set up a conference. The process can be repeated

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 24, 2009 at 04:29 from IEEE Xplore. Restrictions apply.

114 so that up to 5 communications can be merged by the originator (8). It is immediately clear that a more efficient method is required.

Within the ad hoc working group on UIC issues in ETSl a general group call extension to GSM is being discussed (4). Essentially, for a group call a single traffic channel per cell is dynamically allocated. Paging messages containing the actual channel number for the particular group call are sent to all mobiles. Handovers have to be initiated by the mobile stations in contrast to standard GSM where this is done by the wayside GSM infrastructure. As mentioned earlier a priority scheme is required to be invoked at call set-up and handover but, unfortunately, is not available in standard GSM. Its purpose is to pre-empt (terminate) ongoing calls of lower priority to make resources available for a higher priority call. It is intended to provide 8 different priority levels in GSM-R. Most of these extensions will be located in the MSC and BSC. Concerning the fast call set-up requirement, standard GSM has to be improved as well. Even if the authentication procedure is switched off, call set-up times are still too long. Therefore, a direct assignment procedure for mobile terminated calls has been suggested to ETSl which speeds up the assignment process for this kind of calls. The principle is to allocate a down-link channel in all cells were the mobile might reside followed by a paging message to instruct the mobile stations accordingly. Meeting the addressing requirements

For calling a train under its running number there are two promising solutions. The first solution is to enter dummy entries for all running numbers into the Home Location Register (HLR). The HLR essentially is a huge database holding subscriber information like current location, for example. These dummy entries point to real Mobile Station ISDN numbers (MSISDN) of the mobile stations on the corresponding trains. The pointers can be set remately from a mobile station via remote operations through the Operation and Maintenance Centre (OMC) of the MSC. Alternatively, the redirection in the HLR can be done using the Unstructured Supplementary Service Data (USSD) service of GSM. Moderate changes to the MSC software are necessary for this purpose. The second solution is to use the "follow me"

feature of Intelligent Network (IN) systems which could be provided externally to the GSM system. This solution would not require any changes to the GSM system. The location dependent addressing of a wayside terminal (e.g. calling the local signalman) from a moving train can be reaiised in various ways. The most promising one is to use a cell-specific routing depending on the identifier of the cell where the train is currently located. This feature is already available in some MSCs. Depending on the cell identifier either the routing to the final destination can be performed automatically, if it is unique, or a set of possible destinations can be retrieved from an MSC database and transmitted to the mobile station on the train. Meeting passenger communication requirements

The easiest way to meet the passenger communications requirements is to make public mobile networks available by equipping trains with repeaters. Measurements in trains at 200 kmh have shown that this is a possible simple solution. GSM-R frequency band allocation

The "Conference Europeene des Administrations des Postes et des Telecommuniactions" (CEPT) has assigned a dedicated frequency band for UIC GSM-R applications. This UIC band ranges from 870 - 874 MHz (up-link) and 915 - 919 (downlink). However, this specific allocation causes significant problems because in this case there is no guard band between the standard GSM up-link band (890 - 915 MHz) and the UIC down-link band. Therefore, it has been suggested to move the UIC band by 6 MHz to higher frequencies (876 880 MHz up-link, and 921 925 MHz down-link) which would give the desired guard band. An additional advantage with this allocation is that the UIC band would be adjacent to the GSM extension band (880 - 890 MHz and 925 935 MHz). In Germany for trials and pilot projects this frequency band will probably be used.

-

-

-

The way towards GSM-R systems

As mentioned earlier the aim is to have the GSM-R standard agreed upon in 1995. It then remains to implement these features which will probably take

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 24, 2009 at 04:29 from IEEE Xplore. Restrictions apply.

115 at least one or two years. However, since many applications, e.g. all point-to-point applications can be realised using standard GSM several trials and pilot projects will take place before the full GSM-R standard is available. For the DIBMOF trial in Germany, for example, outfitting of the trackside from Stuttgart to Bruchsal with GSM infrastructure will start at the end of 1994. The length of the track is approximately 60 km. Major points of investigation will be the train radio and the automatic train control applications. The DIBMOF trial will be carried out in close contact with other trials planned by the UIC. A prerequisite for introducingthe new radio system is a smooth transition from current analogue train radio systems to the new digital standard. The most suitable solution to this problem seems to be the concept of dual mode terminals in the trains; i.e. two separate mobile stations (an analogue and a GSM mobile station) under control of a single control panel. This concept is similar to the "chameleon" radio approach taken for the channel tunnel project (9).Trains from Britain, Belgium and France running through the channel tunnel are equipped with a train radio terminal capable of operating on the different radio systems. CONCLUSIONS

Manor Research for their contributions to the presented work. The collaboration with the Deutsche Bahn AG and other DIBMOF partners is gratefully acknowledged as is the partial funding of the presented work by the German Ministry for Research and Technology (BMFT, Bundesministerium fur Forschung und Technologie) and the City Administration of Berlin (Berliner Senat).

REFERENCES

K. Lennartz, "Signaltechnisch sichere Datenubertragung im Rahmen von ClRnet", Signal + Draht, Vol. 85(4),p ~96-101 . (1993).

M. Zeilhofer, F. Kollmansberger, "Projekte fur eine europaische Betriebsleittechnik: DIBMOF, DEUFRAKO-M und ETCS", Eisenbahntechnische Rundschau (ETR), Vol. 42(1/2),pp. 43-46 (1993). R. Knewitz, "DIBMOF Das zukunftige Mobilfunksystem der Bahn", Eisenbahningenieur, Vol. 44(8), pp. 528-535 (1993). W.T. Webb, R.D. Shenton, "Pan-European railway communications: where PMR and /E€ Electronics& cellular meet", Communication Engineering Joumal, August 1994, pp. 195-202 (1 993).

M. Goller, K.-D. Masur, "MeRergebnisse und In this paper common requirements of European railways have been discussed The impacts of the decision to base the future radio standard on GSM have been analysed and the necessary extensions of the standard to adapt GSM to railways have been identified The option for a significant cost reduction due to radio based signalling and automatic train control has been detailed. Moreover, the possibility of opening a broad range of PMR applications to GSM due to the railway extensions has been addressed.

lo realise these advantages it remains to finalize the new GSM-R standard through ETSl and to demonstrate its capabilities through trials and pilot projects, The above mentioned problem with the UlC frequency band allocation urgently needs to be clarified.

Parameter zur Modellierung von Bahnim 9OO-MHz-6and', Mobilfunkkanalen Nachrichtentech. Elekfmn., Vol. 43(6), pp. 290-295 (1 993). M. Gbller, "Radio Channel Measurements on Lines of the German Federal Railways in the COST237, TD(92)20, 900-MHz-Band", Vienna 1992.

D. Lefebvre, "GSM Radio Link Simulation Study: Final Report", internal Siemens Transportation Systems Ltd. report, August 1994. M. Mouty, M.-B. Pautet, "The GSM System for Mobile Communications", published by the authors, ISBN: 2-9507190-0-7, 1992

R. Parris, "Ground to train radio - Multistandard radio equipment The chameleon radio system", ITTG '93, Lille, France, September 1993.

ACKNOWLEDGEMENTS

I would like to thank Siemens Transportation Systems Ltd., Siemens Mobile Networks and Roke

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 24, 2009 at 04:29 from IEEE Xplore. Restrictions apply.

-