GSM-R (Global System for Mobile – Railways) is now well established as the radio system of choice for track- to-train voice communication in Europe and many countries beyond. A comprehensive description of the system was given in issue 48 of The Rail Engineer (October 2008). It is also a constituent part of ERTMS (European Rail Traffic Management System), being the transmission link between the control centre and the ETCS (European Train Control System) equipment on the train. Its use in that role is much less understood, mainly because ERTMS deployment is still somewhat limited.

GSM-R is, however, an old technology by current standards and, unlike its public network equivalent (GSM-P or 2G), the railway radio system has not been developed into 3G or 4G offerings. So how long can GSM-R exist and what are the implications for the longer term? This subject was extensively discussed at the recent RailTel conference in Vienna with some illuminating facts emerging.

GSM as a standard

Whilst newer technologies have emerged, GSM-P or 2G systems continue to be used by a significant proportion of the population worldwide. There can be no escape from this and the European Radio Licensing authorities, under instruction from the EU, have to keep the technology supported until 2025. The mobile radio supply industry must equally continue to supply products for this period, meaning that these radio sets M3391M-1012will continue to be made and sold. The same situation should exist for GSM-R where there remain two suppliers of infrastructure (Siemens and Kapsch, the latter acquiring the erstwhile Nortel business) plus sufficient makers of train radio equipment.

The UK GSM-R network is now fully operational in the south of the country and has replaced the National Radio Network (NRN) between the Wash and the Severn. It will be extended to the whole country shortly and, whilst the Cab Secure Radio systems operating in the South East will remain for the time being, eventually these too will transfer to GSM-R. Other countries are also moving to nationwide networks so a lot of capital has been invested. Railway management would not be impressed if it were to be announced that GSM had to be switched off within, say, five years!


GSM was designed primarily for voice communication, this being foreseen as the prime need at the time – early 1990s. Its capacity for sending data has always been somewhat limited and the growth of texting and internet connections meant that some improved data capacity would be needed. Within the GSM standard has thus been developed the General Packet Radio Service (GPRS) which, instead of establishing a connection to a mobile and holding it, sends data in small packets to different mobiles near-simultaneously. The GPRS service has been available on the public networks for some time, much use of it being made by people needing internet connectivity.

The ERTMS programme has similarly encountered data capacity problems with its GSM-R bearer and only a circuit switched connection (i.e. the connection to a train is held continuously) is currently approved for ETCS operation. This has meant that, in busy traffic areas such as terminal stations and complex junctions, there is insufficient capacity to have all the trains connected to the system. As such, railways have resorted to conventional lineside signalling in these localities. So why not use GPRS? Tests done so far might have indicated some evidence of packet loss but this is being disputed. Signal engineers, being a conservative breed, thereby declared GPRS as ‘not proven’ and have insisted on keeping circuit switching, a decision that is unsustainable.

Mindful of the problem, the ERTMS Users Group based in Brussels has commissioned a new set of trials that have progressed beyond the laboratory testing stage and are now progressing to field tests. These will be conducted firstly on a typical suburban line, secondly on a high speed line and finally on a line that crosses a country border. The trial will include the Hertford Loop section in the UK, where a Class 313 EMU is being equipped for extensive ERTMS proving trials before a final commitment to rolling out the technology on the Great Western main line. Tests will also be done on a TGV Unit in France to prove the high speed element.

In parallel, the ETCS specifications are being modified in anticipation of GPRS operation which will impact on the delay performance for timely Tunnel Repeater [online]packet delivery, modifications to the coding criteria and a network assisted cell change.

Every indication is that GPRS will be successful. No evidence of packet loss has so far occurred. The final outcome is crucial as, without GPRS, the whole ERTMS program is at risk since circuit switching will not enable any national roll out right to the ‘buffer stops’ and a mixture of signalling technology will continue to be necessary. The data demands of ETCS are not great but they need to be continually available otherwise movement authority data will be lost. Calculations indicate that a GPRS based service will be adequate for the task within the present dedicated frequency allocation of 4MHz uplink and downlink of the GSM-R band.

Another twist in the situation is whether GSM-R networks should be migrated to IP (Internet Protocol). Existing GSM-R networks rely on TDM (Time-Division Multiplexing) infrastructure that is not IP compatible. Jochen Nowotny from Kapsch suggested that preparation for a transition to IP needs to happen since the telecom industry will progressively standardise around secured mobile IP access, a process that was conceived more than ten years ago. Above all, safety of connection must be maintained since both voice calls and the ETCS bearer have safety implications.

IP interfaces should follow the adoption of GPRS. A sensible plan will be to prepare the infrastructure first, leaving the mobile radios as the ‘last mile’ of the transition. If achieved, this will converge with the evolution of other standards used for secure communication that potentially could make GSM-R a workable system for long after the 2025 date proposed.

Beyond GSM-R

There has been considerable speculation as to what will eventually replace GSM-R with much talk, principally by some radio equipment suppliers, that LTE (Long Term Evolution) within a 4G service is the answer. Detecon, a specialist consultative firm based in Germany which has studied and advised on railway radio technology for many years, has investigated the possibilities. Certainly, LTE has a much higher data handling capability than GSM. A high peak data rate, greater than 100Mbit/sec with 150Mbit/sec maximum, will be possible. Using an IP based flat architecture, flexible bandwidth allocations of 1.25MHz, 2.5MHz and up to 20MHz will be available. High spectrum efficiency for both uplink and downlink ‘round trip’ times will be much better than GSM – typically 10msec as against 150msec.

The word ‘evolution’ is important as LTE is a mixture of well proven techniques combined with other standards and technology. Duplex operation can be either in separate frequency bands or a single time-shared band. LTE spectrum in Europe will initially be in the 0.8GHz, 1.8GHz and 2.6 GHz bands. As other mobile services are closed, additional spectrum in the 0.85GHz, 0.9GHz and 1.9GHz bands is likely. Applications such as on-board CCTV surveillance, passenger entertainment and information, and maintenance diagnostics will all be possible. Products are not yet readily available but, as the service takes off, so the price will fall. Backward compatibility with existing networks and enhanced interoperability will be a requirement.

CoreNetworkDiagram [online]There is a downside: LTE has been conceived as a data highway and voice services may be difficult. Although voice over IP (VoIP) is reasonably well established, it is not readily useable on LTE at the present level of maturity. Voice calls remain currently on the 2G and 3G networks. Since GSM-R was conceived for track to train speech, this situation is a show stopper in the short term. Group call and call priority requirements, also back to back radio operation and local call routing, are all under consideration but it will be some time before they become available. The specifications for voice services in LTE, even in the public sector, are not due to be finalised until mid-2014 with roll out unlikely before 2016.

So, whilst LTE will offer many advantages over GSM-R, migration is probably not going to be a practical proposition until 2022. This will give some comfort to railway management in terms of getting a payback on the considerable investment already spent.

So where are we?

GSM-R has taken a long time to get established. With GPRS it can do the job it was designed for, including supporting ETCS operation, and should remain in service until at least 2025. By sensitive negotiation with the licensing authorities, it may well be able to continue longer than that. As such, the lifecycle should be much more aligned to what railways expect for technology.

When change does come, there are a number of questions to answer:

» Will there need to be special features built into LTE
to satisfy railway requirements if that is the eventual chosen system, such as an LTE-R? The consensus seems to be ‘No’, since deviating from standard products means higher cost and long development times.

» Will there need to be a dedicated frequency allocation for railway use? Opinion is divided on this, although some moves are already being made to reserve spectrum for ‘public services’ operation including fire, police and ambulance. If this transpires, the railways could well be part of it.

» Will there need to be a dedicated railway radio infrastructure – masts, towers, base stations, etc? Again, opinion is divided with one vision being to build networks on a shared basis with other operators. Calculations indicate this would lead to a 30% cost reduction.

» How would a future LTE network operate for the railway? If a shared network proposition were to be established could this create an ‘IP cloud’ that would connect a network control centre to the relevant group of base stations.

» It is likely that Gateways will be designed to link LTE to earlier radio systems. How will this work with GSM-R and will this be the solution for a migratory path?

So, there is some comfort in the short term but time marches on. It only seems like yesterday that GSM-R was first thought about but it is in fact 20 years ago. In less time than that, it will be necessary to decide what will replace it, so resting on one’s laurels would be foolish. Railway telecom and radio engineers need to continue collaboration at a European (actually worldwide) level to keep abreast of what is happening and actively plan for the future.