Significant developments of GSM-R in France

Many railway operators decided to implement ground-to-train radios on their networks during the fourth quarter of the 20th century, which, for most of them, was to use variants of UIC specific analogue technology.

In order to anticipate the upcoming obsolescence of this existing radio and having in mind the objective to improve interoperability of railway operations all over Europe, the UIC took action in the early 1990s to determine which new technology could be specified and promoted.

The MORANE initiative was defined for that purpose, and the result of its technical evaluation was to select an evolution of GSM specifications, named GSM-R, with the ‘R’ standing for ‘Railway’. The rationale for such a choice was the following:

• GSM-R is based on the widely known GSM specifications. It is a standard and by no means a new additional specific choice
• The ‘R’ of GSM-R covers the fact that;
• The radio frequencies used are close to the GSM ones but dedicated to railway operators. Therefore, GSM-R products are globally inherited from GSM products with a slight adaptation to support the specific 4 Mhz wide frequency band
• Some PMR-like features are added to GSM specifications in order to deal with the railway functional requirements. In terms of products, this means that some software variants are required

GSM-R is digital technology which can support any data applications in addition to voice applications. By the end of the 1990s, a first set of GSM-R specifications was completed to form a standard named ‘EIRENE’, which is one of the numerous chapters of the Technical Specifications for Interoperability (TSIs).

As for GSM, the European Union requires GSM-R to be implemented all over Europe, which means that:

• Railway infrastructures have to be upgraded to GSM-R within a certain timeframe. This brings some constraints:
• The speed of deployment is undoubtedly different from one country to another
• Within one country, until deployment is completed, transitions between technologies must be managed
• When ground to train radio is mandatory, one of the two following migration schemes must be implemented:
• Mobile-oriented: All engines are to be equipped with mobiles – CAB radios – which can support both GSM-R and the old analogue technology
• Network-oriented: CAB radios operate in single mode, GSM-R or analogue. GSM-R and analogue networks are operated in parallel

GSM-R services

GSM-R services have been designed to fulfil the functional requirements of the dispatchers, whose role is to manage the railway traffic and of the drivers, whose role is to perform their trip in a secure way. The services include quite conventional mobile telecom features and some features targeting the railway needs are added, such as:

• Group calls including mobiles and dispatchers;
• A specific group call is the REC (Railway Emergency Call)
• Priorities and Pre-emption: EIRENE defines a priority to call types, which leads to the pre-emption of one call of lower priority when a new incoming call cannot be served due to lack of resources. Such a mechanism guarantees that calls of the highest priorities, usually those related to the safety of operations, are always completed
• Functional Numbering: a specific addressing plan is defined which allows users to designate targets by their functional number, such as a train number, instead of their mobile 10-digit telephone number
• Location Dependant Addressing: for a driver located at a given geographical point, the functional need is to reach in the quickest way ‘the local dispatcher’ or ‘the regional dispatcher’ without having to look for the 10-digit telephone number

The EIRENE specifications in their current version define the set of mandatory features that must be implemented to ensure interoperability. However, the efficiency of railway operations via the ground-to-train radios, coupled to some national rules – mainly the REC usage – may lead to the need of additional work in order to evolve the standards through the international standardisation groups.

In addition, GSM-R has been chosen by standardisation groups to be the transmission layer for ETCS (European Train Control System) the new railway signalling technology promoted by the EU. The application layer, split between the infrastructure equipment (the RBCs) and the in-train equipment (the EVCs), uses the GSM-R radio to exchange information through data calls. Currently switched circuit data calls are used. Some study is underway to check whether GPRS (packet-switched data exchanges over the air) could be an opportunity to enhance the feature.

GSM-R in France

Context and assumptions

Réseau Ferré de France (RFF) was created in 1997 to own and develop the railway network.

Regarding the GSM-R project, RFF is the operator and the project owner. It defines objectives, gets the fundings from Public Authorities and defines the organisation in charge of the network implementation.

SNCF, in addition to be a railway carrier, has the RFF delegation for network operation and maintenance.

Regarding the GSM-R project, SNCF is in charge of the network Design (Radio, core network, transmission, railway telephony), network deployment, End to End validation and Operation and Maintenance.

Around 14,000km of lines, equipped in the early 2000s with ground-to-train analogue radio, must be upgraded to GSM-R. This includes 1,800km of high speed lines (HSL) and 12,200km of conventional lines.

Around 8,300 engines equipped with analogue CAB radio must be upgraded to GSM-R in parallel with the network upgrade.

Following a preliminary study, the ‘single infrastructure-double mobile’ strategy was chosen for security reasons:

• ‘Single infrastructure’ means that at a given time, for a given line, only one radio technology will be used for dispatchers
• ‘double mobile’ means that as trains travel over regions that may or may not be GSM-R equipped, their mobiles must support both GSM-R and old analogue radio technologies

RFF launched the GSM-R project mid-2003 and decided to split the country into zones that would be equipped one-by-one. In its current organisation, the project is supposed to last from 2003 up to 2015. This is due to the fact that, mainly for budget reasons, RFF cannot deploy GSM-R simultaneously in all regions. The map on page 59 displays this plan.

Challenges and constraints of GSM-R in France

One consequence of the deployment strategy and program is that transitions between lines equipped with GSM-R and lines still equipped with analogue radio will have to be managed until deployment completion. The safety constraints that must be followed in France have lead us to design some ad’hoc equipment, mobile based, which allows to transfer a REC from one technology to the other.

Opening the ground-to-train service on a given line implies the implementation of the following:

• The GSM-R radio layer, which requires;
• At least partially the common GSM-R Core Network, including the usual GSM-like NSS, MSC and HLR, but also additional servers that deliver the railway features
• BSS infrastructure, including the BSC and TCU equipment, linked to the NSS, dedicated to the radio sites
• Radio sites delivering the radio coverage suited to the functional requirements of the railway traffic management
• Data network supporting Operation and Maintenance servers with tools, allowing configuration and operation of the NSS and BSS
• Transmission network between radio sites and BSCs
• Railway telephony, linked to the GSM-R NSS, in order to provide the dispatchers with the fixed terminals they need
• Bi-mode GSM-R / analogue CAB radio in all engines that will have to run through the area

Even though GSM-R is a standard, implementation and operating rules differ from one country to another. On the French Railway Network, ground-to-train radio availability is mandatory in order to run a normal mode of operation.

The unavailability of the radio service would lead to the application of very restrictive operation rules, including drastic speed reduction. One can understand that the global GSM-R architecture has to be highly reliable in order to ensure the right level of security for train operations.

Particular attention has to be paid to Core Network, BSC and transmission backbone availability, whose failure could impact the railway traffic over a very large region, if not the entire country.

Therefore, design of the network architecture minimises the impact of any single failure on the behaviour of the GSM-R services. For example, any BTS is connected to a BSC via two independent paths (loop configuration). Site design includes replication of key elements such as air cooling, power supply and transmission adduction. For key equipment which could not offer automatic redundancy features, efficient disaster recovery plans have been set-up in order to restore the service on dedicated backup equipment.

The GSM-R network also differs significantly from a classical GSM network on the fact that the radio traffic is extremely low and sporadic. Therefore, classical tools commonly used by a GSM operator to monitor the Quality of Service, which are based on a statistical approach, do not apply and specific GSM-R tools have to be studied.

Current status

By October 2007, the status of the project is as follows:

• The Core network has been built and is operational. Selected technical architecture consists in two ‘very secure’ Core sites which include the NSS equipment (MSC, HLR, core platforms and servers such as SCP/IN, SMSC, OTA). These two sites are twin sites running in active/stand-by mode. Should a disaster happen on the active Core site, and a ‘hand-over’ to the backup site could be performed technically in less than one hour owing to the transmission backhaul implementation.
• Currently two high capacity bi-BSC-TCU sites have been built. Once again, should a BSC –TCU be damaged, transmission backhaul would allow to re-route the BTS transmission links towards an operational backup BSC fairly quickly.
• Owing to an appropriate data network, an Operation and Maintenance Centre gathers all servers and terminals which allow operations of NSS elements, BSS elements, and transmission gears: configuration, administration, maintenance,
• The new East European High Speed Line has been built from 2003 up to 2007 and is equipped with GSM-R. The GSM-R design on this line is very innovative and has been the theatre of very interesting technical experiments in the second quarter of 2007.
• A 300km-long test line named ‘the Pilot Line’ has been implemented in order to address most of the technical challenges. Conventional line from Paris towards Bar Le Duc has been selected for this purpose, and has been put into service in two steps, one in March 2006, the complete line in January 2007. Its characteristics are:
• A dense urban area in the suburbs of Paris, including long tunnels
• Many specific zones requiring some advanced radio engineering studies
• A maximum speed of 160km/h
• Three tunnels outside the suburbs
• Around 1,250 traffic runs a day
• Closeness to the East European High Speed Line, also GSM-R equipped.
• 2,000km of lines in the Eastern part of France, that is a small quarter of the whole country, including the regions named Champagne, Ardennes, Lorraine, Alsace and part of Picardie is currently under construction under the name ‘Tranche 1’. This will lead to;
• Construction of around 320 radio sites
• 25 tunnels to be equipped
• Eight sets of lines of around 250km each to be put into service one-by-one from March 2008 up to mid-2009.
• More than 2,000 engines have been equipped with a new bi-mode CAB radio, those which are involved in the traffic runs that cross the GSM-R geographical zone.

Much work has been undertaken in 2007 to interconnect the French GSM-R core network to core networks of neighbouring railway operators. Currently the following interconnections are technically available:

• Germany
• Belgium
• Netherlands
• Switzerland

Lessons learned from operations

Being a standard, one might think that GSM-R and overall ground-to-train radios built upon it should work 100% at switch-on. Obviously that was not the case and reality offered more contrast.

First, RFF and SNCF have decided to build a captive GSM-R test bed, close to Paris which is a small replication in a lab of a complete network. It includes:

• A set of all type of NSS equipment and core platforms
• Two BSC-TCUs and three BTSs
• A test line of nine BTSs, linked to the lab system
• One Railway PABX and all types of dispatcher fixed terminals
• GSM-R handhelds of different types and CAB radio

This test system is configured, operated and maintained in order to:

• Validate hardware and software of all network elements, before deployment in a live network
• Perform end-to-end system tests which ensure compliancy of services
• Investigate technical issues seen in the field to find the root causes, get the fixes and validate solutions
• Design and validate OA&M procedures before implementation on the live network
• Train people from engineering and operational organisations

Despite intense activity performed in our system lab from mid-2004, which has been fruitful in terms of number of defects discovered and fixed by suppliers, we have been facing some issues since GSM-R was first put into operational service. First was the area of mobile products. Surprising problems appeared mainly in the domain of radio network reselection, which seem to be specific to the usage of mobiles in the context of GSM-R application. Secondly, radio difficulties were encountered in dense urban areas. The number of radio frequencies granted to the GSM-R band is small (4 Mhz of band) which leads to complexity in frequency planning and the proximity of some high capacity GSM-operators sites transmitting at fairly high power can generate blocking at the mobile level. Thirdly, ground-to-train radios built on GSM-R brings new procedures and changes in the habits of both drivers and dispatchers. A significant effort must be made to train and support the users when switching to the new technology.

Tools and procedures have been implemented in order to monitor network quality of service. This monitoring actually covers different aspects including:

• Technical quality of service of the GSM-R network itself, which encompasses the behaviour of the telecommunication equipment, both GSM-R and transmission
• Functional quality of service of end-to-end ground-to-train radios, which includes the railway telephony, mobiles, and mainly deals with availability of overall service for railway operations
• Perceived end users quality of service which takes into account the capability of the technology to deliver first class audio quality and ease of use of terminals, fixed or mobile.

Finally, it is a well-known issue that building a telecommunication network while operating part of it at the same time induces difficulties and can penalise the quality of service, especially in the context of the French GSM-R network which core network is highly centralised. In order to cope with this issue, both engineering and operational organisations have put in place processes through which any set of technical operations on the network is traced and checked before execution in order to evaluate potential impacts.

The East European High Speed Line (EE HSL)

On 10 June 2007, commercial operations of the EE HSL were launched. This was the result of many years of efforts targeting the development of high speed railway from Paris towards Germany and eastern Europe.

With a length of 300km and commercial operations at 320km/h, the EE HSL, at its current stage, dramatically reduces journey times between Paris and many cities such as Reims, Metz and Nancy in France, Luxemburg, Francfort, Ulm and Münich in Germany.

This project has included many innovations in different areas such as the platform, civil works, and GSM-R coverage. Test plans have been designed in order to validate operations at a speed of 360km/h. Even more, a specific sub-project named V150 has been designed by RFF, SNCF and Alstom, with the objective to set a new world railway speed record. The objective was fulfilled on 3 April 2007, with a train that achieved the incredible speed of 574,8km/h!

GSM-R on the EE HSL

This V150 project has enabled technical teams to perform some unique experiments. This has been the case for the engineering team responsible for GSM-R radio design and implementation.

Before delivering some of the results that were gathered through high-speed testing, let us describe quickly the GSM-R radio network architecture, which is quite innovative. The GSM-R technology has been implemented on the EE HSL line in order to provide:

• Ground-to-train radio services for drivers and dispatchers
• Radio transmission for the ETCS signalling application

To meet radio service availability requirements specified by ETCS, the project team has decided to implement two layers of GSM-R radio coverage, each layer of cells, belonging for safety reasons, to a different BSC.

Such architecture happens to be quite efficient, although not easy to validate and tune. A lack of GSM-R frequencies and proximity of conventional lines also equipped with GSM-R induced some difficulties which had to be overcome.

GSM-R at very high speed

Key to the V150 project was a customised train dedicated to very high speed, in which, it has been possible to install GSM-R antennas, CAB Radio and appropriate test tools for experimentation. These experiments were performed with RFF GSM-R infrastructure of the EE HSL and with SNCF Cab Radio.

GSM-R calls were hold up to the record speed, which means that the technology was able to cope with the impacts of distortion mainly due to the Doppler Effect induced by speed and with the high frequency of hand-overs.
To sum up the results:

• Voice quality is quite acceptable up to 500km/h
• Data quality at a bit rate of 4800 bps – rate that will support the ERTMS operation – is good up to 350km/h for the uplink and 450km/h for the downlink

It must be noticed that the experiments were performed with usual GSM-R technology, which is without customised parameters or dedicated development.

These results provided us with a better understanding of the GSM-R system behaviour under extreme conditions, for both infrastructure and mobile sides. In the future, RFF and SNCF will be in a better position to develop efficient engineering rules for upcoming High Speed Lines, and especially for those equipped with ERTMS technology. It will also provide cooperation with suppliers in order to improve the technology towards additional performance.

Conclusion and next steps

This article highlights that GSM-R in France is well advanced. Overall architecture is specified and key decisions have been made, especially those related to security of operations. The core network is built and ‘up and running’ and first experiments of ground-to-train radios built upon GSM-R have been successful. 600km of lines are presently being operated with a good feedback and GSM-R is ready on one high-speed line, not only for ground-to-train radios at a speed of 320km/h, but also for ETCS validation. Plans are also in place for further deployment of GSM-R radio.

The current success of GSM-R in France will continue with the following challenges:

• On time completion of the ‘Tranche 1’ program which is the deployment of GSM-R over the eastern part of France during the next 18 months. Current status of this plan shows that some functional and technical points such as border crossing or Enhanced Railway Emergency Calls still need either clarification or further improvement from the standards. We are confident that objectives will be met.
• Validation, by performing ERTMS end-to-end testing, that the GSM-R architecture and performance on the East European HSL fulfils application requirements, the last step before starting ERTMS operation.
• Quality of service consolidation. As the network grows, efficiency of tools and associated procedures will be improved in order to keep controlling and improving the quality of service for all applications relying on GSM-R.
• Evaluation of options to accelerate the program. RFF is currently investigating the opportunity to design and implement a Public Private Partnership (PPP) for GSM-R in France. Such an evolution of the project might speed up completion of GSM-R deployment, and a significant amount of work is required within the next months in order to define potential evolutions of overall project organisation and to specify the responsibilities of the potential new partners.