More than 100,000km of GSM-R has been installed on European railways since the first operational system began in 2004. GSM-R currently supports voice services from trains to controllers and trackside workers, railway emergency calls (REC), some data messaging and passenger information systems.
In addition, GSM-R is used as the data bearer for the European Train Control System (ETCS). GSM-R and ETCS are the two components of the European Rail Traffic Management System (ERTMS), which was designed to provide a homogenous train protection and signalling system across Europe.
GSM–R is a 2G technology, but as Josef Doppelbauer, executive director, EU Agency for Railways (ERA), concedes: “We have to admit it is still fit for purpose, except for some limitations in capacity. We have recently upgraded the specifications to include GPRS and Edge.”
Some rail infrastructure managers are deploying GPRS and Edge to cope with these capacity issues, but that aside, it is clear that the railways are facing a larger existential crisis when it comes to their radio communications.
Existential crisis
The threat takes the form of two related issues. The first is that the rapid development of new radio standards every five to seven years is leaving the railways behind, with 5G being implemented now and 6G in discussion.
“This means that certain services that would be attractive to rail cannot be used, because we are still in the 2G world,” says Doppelbauer. “But, more importantly, the main risk is the obsolescence of GSM-R.”
GSM-R is a niche technology developed purely for use by the railways and it cannot be evolved to a modern radio standard. So, it is vital that the supply side continues to support it while a replacement is developed.
That commitment is now in place. In July 2021, the UNITEL Committee, comprising the major GSM-R railway telecom products suppliers and companies of UNIFE (European Rail Supply Industry Association), published a GSM-R Long-term Support Statement. This was to the effect that they would “provide support for GSM-R at least until 2030 on a general basis and beyond 2030 on a per contract basis”.
This commitment should not be underestimated, as suppliers rarely have any incentive to keep near-obsolete radio technologies going. Emanuele Di Liberto, head of global rail business Centre at Nokia – a key member of UNIFE and UNITEL – observes: “There are not too many GSM-R suppliers left these days. But we are committed to extending the lifecycle of GSM-R.
“It is really difficult for the [supply] industry, but I’m not sure we had a choice. It takes a lot of planning and a lot of commitment from both sides to keep something like GSM-R out there for, at the end, almost a 35-year lifecycle, which is a lot.”
In the meantime, the International Union of Railways (UIC) and ERA have been working hard to develop a strategy to replace GSM-R. Discussions began in 2010, and the successor Future Railway Mobile Communication System (FRMCS) was launched as a project in 2014.
The UIC is keen to ensure FRMCS is based on a future-proof technology capable of evolving as radio technology develops. But FRMCS also has to deliver railways-specific quality of service requirements. It is no secret that FRMCS will be based on the 3GPP 5G standard, but this means that, as with other mission-critical sectors such as public safety, railway-specific applications need to be built into 3GPP releases.
Current work is based on 3GPP Release 17, but Di Liberto says: “The general agreement is that railways will use R18 as the foundation release, although there might be a couple of things left for R19 to cover. But the crucial topic of the railway emergency call (REC) should be fully defined by R18.”
In February 2020, UIC published a mammoth document outlining over 70 possible use-cases for FRMCS. These use-cases can be broken down into three broad categories: critical, performance, and business communications.
Critical communications includes applications essential for control of train movements and assured data communications for ETCS. ETCS needs highly reliable connectivity to transmit frequent updates on position, speed and train-integrity information from train to track, and the signalling of movement authorities (MAs) from track to train, at the appropriate time to guarantee safe operations.
Other critical apps include: REC; voice and messaging between controllers and train drivers; communications to trackside workers; monitoring and control of critical infrastructure; automatic train operation (ATO); and automatic train protection (ATP).
Performance applications cover those features that help to improve the performance of the railway operation, such as train departure procedures, telemetry and IoT. Business applications include aspects such as passenger information systems and ticketing support. Passenger on-board and platform Wi-Fi also comes under this heading.
End-to-end system
FRMCS, based on 5G specifications, will be used for the complete end-to-end system, including the network and the on-board system, which covers both the terminal and the functionality implemented on board.
“In terms of architecture, we want to profit from the [3GPP] evolution and we want to separate the communication layer from the application layer. We want to have standardised interfaces between on-board and trackside, in order to be flexible for future expansion, and possibly, changes in the radio system,” explains Doppelbauer.
The UIC has identified eight key challenges FRMCS needs to solve: QoS, ETCS bearer independence, future proofing, equipment, applications, spectrum availability across countries, interoperability, cyber security, and time/cost effectiveness.
As Di Liberto points out, QoS is inherently part of 3GPP releases, so that should be built in. Ensuring ETCS bearer independence is one of the most important aspects of FRMCS, as in the current GSM-R system, the radio and the ETCS functions are tightly coupled.
5G will enable a clear separation between the communication functions and the train protection functions, as well as other functions. FRMCS will be based on the UIC’s technical specification for interoperability (TSI). Doppelbauer says that the latest 2022 TSI has introduced the modularity needed to decouple the radio and the ETCS functions.
“This is available and already submitted to the EC and it now needs the agreement of a vote by the member states,” reports Doppelbauer. “We have also specified the interfaces between these building blocks, and they are available. What is missing is some of the functionality of FRMCS itself, mainly related to the voice part and the mission-critical factors. But as far as the on-board architecture is concerned, the modularity is there.”
Basing FRMCS on 3GPP standards will give the railways the future proofing it needs in terms of an upgrade path with backwards compatibility, as well as allowing it to take advantage of the economies of scale a global radio ecosystem provides. 5G also comes with much better cyber security compared with 2G.
Sorting out spectrum is obviously central to the future success of FRMCS, and this has largely been achieved, at least as far as Europe is concerned. The EC has agreed to allocate harmonised spectrum across Europe for railway mobile radio by assigning 2x5.6MHz in 900MHz (the existing GSM-R band) and 10MHz in the 1900MHz band.
The assumption is that FRMCS will start out using 1900MHz, while 900MHz will only become fully available after the complete migration of European railways to an FRMCS-based network. Some 900MHz might be used for both GSM-R and FRMCS during the transition period, so long as co-existence solutions can be developed.
Interoperability covers not just FRMCS equipment from different vendors, but also between GSM-R and FRMCS systems. It is essential that there is a standard interworking function between the two radios during the transition period.
“The on-board equipment will have to be capable of supporting both technologies,” says Di Liberto. “The ETSI railway radio telecoms committee is working on the specific technical requirements and there will be a specific interworking standard for this.”
One potential issue is the lack of chipsets for the railway bands. The concern is that the big chipset vendors may not have much incentive to cater for these more niche mission-critical 5G bands.
Multi-application support
Naturally, replacing GSM-R is going to be expensive, but FRMCS should be cost-effective, because it will support many more applications, including ATO and ATP. Also, as Di Liberto points out: “If you compare the cost of telecoms to the overall budget of the railways, it is very, very small compared with the civil works and so on.”
In terms of timelines, Doppelbauer says the current TSI has captured as much of FRMCS as possible so far, including the interface specifications and a list of the mandatory specifications. “The full specification is not yet ready and we expect the final version in 2024, so that an update of the TSI can be published in 2024, or more realistically in 2025.”
Products can be developed on the basis of these final specifications and then the member states can decide on their deployment and FRMCS migration plans. The EC has already requested member states report on their plans regarding this.
However, Doppelbauer says it is clear from these reports that “practically no member state is planning full decommissioning of GSM-R before 2035. So, what we can say is that all member states want to keep GSM-R as long as possible. The majority of member states intend to start [migrating to FRMCS] around 2027 to 2030.
“Of course this will heavily depend on the full specification [being available] and, for us, full specification means validation. So, one of the critical steps will be that the set of FRMCS specifications will be validated, so that we are sure that when we include those specifications in the TSI that the quality and stability of FRMCS is assured.”
An FRMCS European trial, based on updated specifications and 3GPP R17 products, is targeted for 2025. Of course, all this has to be funded and Doppelbauer is hopeful that Europe’s Rail Joint Undertaking, a public-private partnership which began its activities earlier this year, might provide an appropriate funding vehicle.
What will be absolutely critical is how the migration from GSM-R is handled. The UIC has set out four key considerations that need to be addressed. These are: telecom on-board architecture (TOBA); signalling operations continuity to ensure ETCS is supported and trains continue to run safely; migration spectrum needs; and GSM-R and FRMCS co-existence.
The UIC has been working to determine the best on-board architecture to enable ETCS migration to FRMCS and to introduce the ETCS bearer independence concept. This will be followed up by a similar process for the voice cab radio. Doppelbauer says the UIC is trying not to be too detailed in its specifications.
“If we specify the on-board system in greater detail, we, on the one hand, restrict the design freedom of the suppliers, and on the other hand from the supplier’s perspective, we make it possible for a customer to buy individual bits and pieces from different suppliers.”
Doppelbauer notes that the latter is “not to the full liking of the suppliers because they usually live off the very attractive service contracts”. The advantage for end-users is that they can negotiate better contracts and escape vendor lock-in.
“But strategically, it is the only way forward to bridge the gap between the long lifetime of railway systems and the rather short lifetime of mobile communication systems,” observes Doppelbauer.
Di Liberto says that the trackside installation of FRMCS equipment will be a fairly standard radio implementation. He also states that it is important that a transport network with sufficient backbone is in place to support 5G features such as synchronisation and low latency. But this is something rail infrastructure managers can address now.
Edge computing for low-latency apps and video, along with cloud-based services, will also be needed. “Video will be one of the killer apps demanding edge computing,” points out Di Liberto. “If they want to run trains with higher levels of automation, they will need more video to know what is going on. So, you will need to build some kind of remote edge computing capability.”
The on-board installation is a rather more tricky prospect, however. Di Liberto says: “The main migration bottleneck is the rolling stock itself and that is a concern. Retrofitting trains is very expensive and takes a long time.”
A dual-mode cab radio supporting GSM-R and FRMCS is the most likely way forward during the migration period. “Clearly, what will happen is that for a very long period of time you will have two networks in parallel and then the cab radio. Depending where the trains are, they will be using one or other technology with some sort of interworking,” says Di Liberto.
“And don’t forget you need to support both frequencies, so you need antennas and coaxial cables. That will be a bit nasty to sort out, but it will happen. So, for us, the big challenge is the on-board migration timeline.”
European governments may want to keep GSM-R in operation even beyond 2035, but the rail industry is doing its best to get FRMCS prepared. “We want to provide the specification on time, so that this does not block the further downstream development,” says Doppelbauer.
