From GSM-R to FRMCS: a new dawn for rail

Just like public safety, the railway community is embracing the latest 3GPP technology, but with Europe’s countries needing to migrate in lockstep, more co-ordination is needed, as Sam Fenwick discovers

One of life’s little ironies is that the business of making things go fast is often one which is slow and methodical when it comes to change, given safety considerations and the sheer amount of players, assets and capital involved. In the world of railways, big change is ahead in the form of the impending transition from GSM-R – a standard for railway communications that was developed in the second half of the 1990s – to its replacement, Future Rail Mobile Communications System (FRMCS). This is needed, given the limited number of vendors that support GSM-R, their uncertain support for the standard beyond 2030, and the potential benefits of aligning the railway communications community with the telecommunications sector.

GSM-R currently handles voice and data for security and train tracking applications, and provides the communications that allow ETCS (European train control systems) to function. Both GSM-R and ETCS are part of the ERTMS (European rail traffic management system) standard, which is defined by the International Union of Railways (UIC) and is designed to allow interoperability between cross-border traffic.

Robert Sarfati, chair of the ETSI Technical Committee for Rail Telecommunications (TC RT) and current chairman of the UIC’s European Radio Implementers Group (ERIG), says FRMCS is a UIC project. He adds that radio standardisation is conducted within TC RT and 3GPP. UIC is the author of the User Requirements Specifications (URS) document and responsible for its subsequent updating after submitting it to a broad review by interested stakeholders. UIC is providing the use-cases, with ETSI’s TC RT subjecting them to a review analysis (ie, determining which of the use-cases’ requirements are relevant to 3GPP and conducting the initial gap analysis); followed by 3GPP working to fill the gaps in the normative domain. The latter process is still ongoing, and like the development of other mission-critical features spans multiple 3GPP releases. The 3GPP work involves rail transport users and industry. It also involves input from the public safety community, which along with other verticals will benefit from the rail transport industry’s request for 3GPP to develop in particular functional aliasing and location-based services, which will allow group and individual calls to be routed based on users’ tasks/assignments and locations.

The future-proofing that FRMCS offers the rail industry goes beyond simply being a new technology. Markus Myslivec, head of public transport solutions at Frequentis, says it is being “described and specified [to be] bearer or radio technology independent”, and that unlike GSM-R where “functionalities affect all layers of communication, even down to the physical layer”, FRMCS keeps all the railway-specific functionality in the application layer. This will free the rail community from being dependent on a single radio access technology, and as FRMCS can take advantage of the other features in 3GPP that have been (or are being) developed for 5G, they may be able to take advantage of network slicing and share parts of the system with public mobile network operators and/or public safety agencies (more on this later).

The fact that GSM-R is currently used to allow a long-distance train to travel across Europe, without encountering any communication issues, creates some logistical headaches for the transition to FRMCS. Simply put, a country cannot unilaterally switch off GSM-R and turn on FRMCS overnight – support for GSM-R must remain throughout the transition.

David Rothbaum, director of business development, mission-critical and private networks at Ericsson, highlights the time required to retrofit all the rolling stock – all the hundreds of locomotives – as a multi-year project in its own right. He also notes that many countries (such as the UK) award licences to train operating companies and it is not possible to modify a licence save at the end of the licence lease period (some licence periods in the UK are as long as seven to 10 years). He therefore expects that the transition from GSM-R to FRMCS will be much quicker in countries where the long-distance trains are still run by the main railway operator, such as France and Germany.

Frequentis’s Myslivec believes it is possible for the first transitions from GSM-R to FRMCS to be completed by 2030, and he adds that the European Union Agency for Railways (ERA) is planning to publish a new version of the Control Command and Signalling Technical Specification for Interoperability (CCS TSI) in 2022. The TSIs define the technical and operational standards which must be met by each subsystem or part of a subsystem to meet the essential requirements and ensure the interoperability of the EU’s railway system. “We recently had an interesting discussion (at UNITEL) with the ERA’s executive director Josef Doppelbauer, who of course says this will happen. I have my doubts because we are far away from completing the technical standardisation. How can FRMCS be introduced into legislation without the standards for it being completed? I don’t think these will be complete before 2021, or even 2022, then the CCS TSI can be adapted and then you will see the first pilot; but I’m still convinced that this will start by 2025, so 2030 is possible.”

According to Pierre Tane, business solutions expert, mission-critical networks – digital solutions at Kontron Transportation (formerly Kapsch CarrierCom), in some countries including Switzerland, “there is pressure from the regulators to shut down the GSM technology in the coming years, which creates more incentive to move fast”. In this scenario, he envisages that international rail corridors may be handled differently from their domestic counterparts, to maintain compliance with the TSI.

Tane adds that another possible factor that could drive FRMCS adoption is the introduction of advanced train control services, and higher levels of automation. However, like Rothbaum, he is also aware of the challenges involved. “When you look at the timescales for the introduction of new systems in railways, usually if you want to proceed and complete your migration you need to start somewhat early.” He adds that some of his railway customers say they still have analogue technology in parts of their countries despite GSM-R’s introduction more than 15 years ago. To him this implies that “large networks will have to start reasonably early to be able to complete their migration in the 2030-ish timeframe” and that the coexistence of technologies should be approached with a long duration of joint operation in mind (as compared to an overnight switchover that has sometimes been suggested by some industry stakeholders).

FRMCS isn’t expected to create major issues for control rooms but other technologies are expected to cause upheavals

The spectrum question
Myslivec says that he doesn’t think there will be big operational issues in terms of dispatching, control rooms and applications (which are his company’s traditional areas of focus), while these subsystems will still face huge challenges caused by technological changes (single core systems, virtualisation and cloud computing, etc). However, he is convinced “that the biggest challenge will be spectrum and how to handle the onboard equipment architecture in terms of migration and coexistence of FRMCS bearers and GSM-R bearers”.

Let’s look at the issue of spectrum in more detail. Currently, European railways use the 4MHz wide 876-880MHz/921-925MHz band for GSM-R, and it has been harmonised for this use by CEPT (the European Conference of Postal and Telecommunications Administrations), through its Electronic Communications Committee (ECC), and the EU. In addition, the 873-876MHz (uplink)/918-921MHz (downlink) band is not harmonised for GSM-R within CEPT, but it is used for this purpose by some CEPT countries.

As the transition from GSM-R to FRMCS will be a gradual one, an operator can’t simply refarm their existing 4MHz from GSM-R to FRMCS, and therefore additional spectrum is required during the transition. There are three candidate bands: 1.4MHz of spectrum for FDD (1.6MHz including 200kHz guard bands – 874.4-876MHz/919.4-921MHz) that sits directly below the harmonised spectrum for GSM-R; a 10MHz spectrum block for TDD in 1900-1920MHz; and a 10MHz spectrum block in the 2.3GHz band.

ECC Report 294 – Assessment of the spectrum needs for future railway communications states that the capacity provided by the 900MHz candidate band won’t be sufficient in “dense railway networks, border areas and high-density areas” during the transition, and therefore the spectrum in 1900MHz “is a prerequisite for many countries to manage the migration with dual networks operating in parallel”. However, it notes that this band is currently licensed to mobile network operators (MNOs) in many CEPT countries and that the additional spectrum used during the transition “will still be required to cover railway’s long-term needs (including critical sensing/video), border and hotspot areas”.

Rothbaum says one of the challenges with the use of the 1.4MHz ‘squeeze option’ is that currently it is not being used by MNOs or other industries for LTE. This raises the question as to “how many chipset vendors are going to step up to the plate and deliver a chipset that supports [it]?” given that the volumes required are far less than those in the public safety industry – he notes the struggle to get ProSe (proximity services – mobile broadband’s lower-range equivalent of TETRA’s direct mode operation) – into silicon for use by public safety organisations.

“There could be niche chip vendors out there, but once you do that you’re going to have a very narrow ecosystem with handsets and modem cards onboard the locomotives to cover that, so for that reason, it will be a little bit challenging. Now once they go to 5MHz spectrum [in 900MHz, the 920-925MHz band] they’re going to get a better response,” Rothbaum says.

As a brief aside (as I’ve mentioned ProSe) it’s worth noting that FRMCS has an off-network requirement referred in 3GPP terminology as a sidelink (GSM-R already has this capability). Ingo Wendler, vice chair of ETSI RT, says “off network is a challenging part” and it has been decided that on-network communications have the first priority, so that the first deployments can begin, with the sidelink to come later (to possibly act as a fallback should on-network communications be disrupted).

Returning to 900MHz, Rothbaum says another challenge with using the 1.4MHz ‘squeeze’ option is its close proximity on the downlink (919.4MHz at its lowest point) to 3GPP Band 8’s uplink (880-915MHz). He says such a small distance between an uplink and a downlink is “almost unheard of in the whole radio spectrum” given that receivers’ filters don’t “drop on a dime”, so they would still be exposed to blocking from transmitters operating in the adjacent band, hence the need to determine the maximum power that could be used without creating coexistence issues.

Rothbaum adds that this is less of an issue with the candidate spectrum in the 1900MHz band because its nearest neighbour is 3GPP Band 1 – 1920-1980MHz (uplink); and there is 10MHz of spectrum available.

Despite the issues with 900MHz that Rothbaum has highlighted, Tane says there has been recent progress in CEPT. “It has progressed from a situation where the viability of the 900MHz spectrum for railways use was questionable from a commercial and technical standpoint to a situation where operating conditions suitable for railways may be reached (including when using the lower 1.6MHz portion of their foreseen harmonised spectrum at 900MHz). We remain cautiously optimistic, knowing that the associated regulatory framework is not yet cast in stone.” He adds that the use of this band for the transition to FRMCS is “quite key”, given the large number of sites that European railways already have operating at 900MHz and the fact that the expense of building a single new site may be as much as €150,000, with most of the cost arising from civil works, passive infrastructure, redundant power supplies and backhaul.

Use of the 1.4MHz option would also allow the railways to be “more in control of their spending because the idea is not to force railways to overly densify their base station sites as soon as FRMCS arrives, but rather let them have a smooth introduction whereby they would reuse their existing sites to have a basic level of services over FRMCS in the beginning, and then with new services being brought into the picture they could either add new 900MHz sites, but only where they want it, or add the new complementary band that is foreseen at the European level, which is most probably going to be the 1900-1910MHz range.”

He adds that there are two main drivers for network densification in an FRMCS context – the desire to guarantee more throughput and the need to co-exist with other systems. “The 900MHz spectrum is located just on the outskirts of the public 900MHz band and it might on occasion [be] required to densify in certain locations to make sure that we keep interference to the public networks to a reasonable minimum”, as that could allow the output power at each site to be reduced.

Rothbaum highlights techniques such as MOCN (Multi-Operator Core Network), which allow operators to share spectrum. He says this (in combination with prioritisation techniques) could allow more efficient use of the 920-925MHz band through allowing MNOs to use the spectrum in the geographical areas where it is not being used for FRMCS, as opposed to the regulator exclusively allocating it for railway use. Rothbaum adds that this would be a “win-win solution [as] it would increase the ecosystem for the chipsets so there would be more attraction for chipsets to handle that particular band and the countries would get more revenue in allowing that spectrum to be used”. It is of course worth noting here that there are parallels with CBRS (Citizens Broadband Radio Service) in the US.

Turning to the 1900MHz band, Rothbaum notes that one of its big advantages is that it is a standard 3GPP band – Band 39 – which is predominantly used in China and is supported by today’s consumer smartphones. However, given its higher frequency than the 900MHz band currently used for GSM-R, this raises the question as to how many additional sites it may require and the cost of building these. Rothbaum says Ericsson is currently performing a study for ETSI to simulate “the edge throughput at 4km at 1900MHz”. He adds that a similar study has been performed by other companies for the extra 1.4MHz in 900MHz option.

There is a third option for FRMCS that is being promoted by Kontron Transportation to facilitate the maximal reuse of the 900MHz infrastructure (particularly sites), while allowing peaceful coexistence with the existing GSM-R system – the White Space FRMCS concept. This entails overlaying a 5G carrier over the GSM-R system and making optimal use of the spectrum available below or in between GSM-R frequency pairs.

Tane describes it as a “very convenient solution” compared with using the 1900MHz band, given the latter’s potential need for more sites. He says this approach is “gaining traction and that the proposed contents and schedule of the test campaign for it has been published in the relevant ETSI standardisation group, with the first step expected to take place at the end of October, followed by a second stage in February and a third and final phase at the end of June 2020.

Going public?
Frequentis’s Myslivec expects that with FRMCS there will still be a dedicated network handling the mission-critical/safety-relevant communications, but there will also be an additional element in the form of non-critical services such as passenger information and video transmission. When it comes to the latter, he is convinced that “the railways may not be able to afford to build their own network with sufficient capacity – they will not get the spectrum for that – [so they will] share with public MNOs or public safety authorities and use their networks”.

Rothbaum adds that Ericsson and Deutsche Bahn are carrying out network slicing trials in “the framework of FRMCS” along a 30km section of track between Nuremberg and Ingolstadt in Bavaria on a high-speed test train. “We were trying to show that we were able to isolate the network into three different slices, one carrying video surveillance [feeds], one carrying IoT [traffic], and one carrying passenger connectivity traffic over one carrier, but isolating it so that we could guarantee the throughput over any particular slice.” While he says that in the trial, there was just an IoT test device onboard, the IoT use-case being considered here is transmitting the data aggregated by the train control management system (TCMS) from the “millions of sensors onboard” to allow real-time fault reporting (currently the data is downloaded at a depot when the train is temporarily out of service). Given the amount of power available on a passenger train, this wouldn’t require the use of low-power wide-area technologies, but the same wouldn’t be true for freight wagons. Rothbaum adds that one interesting application would be to have a location update unit, like a reliable low-power IoT device at the end of a train. “Once you [have that in place] you could then introduce ETCS Level 3 across Europe, [as] one of [its] big challenges is determining reliably the end of the train’s location.”

A bit of further explanation is called for here. ETCS Level 3 supports the moving block concept – the idea that if trains’ locations are known at all times then it is possible to have trains running closer together without compromising safety, thereby increasing railways’ capacity.

Myslivec’s view that a dedicated network will still be required is supported by the current CCS TSI and shared by the UIC’s Sarfati. Wendler highlights the difference between public mobile network operators’ traditional best-effort approach to providing quality of service and the ‘very deterministic quality of service’ that the rail sector requires. He adds that while network slicing may be able to provide this from a resource management perspective, it doesn’t change the coverage requirements.

The key issue here is the way that public mobile network operators’ business model is typically concerned with providing coverage to areas of high population density, rather than unbroken coverage along railway corridors (train control systems require there to be no coverage holes whatsoever).

However, Ericsson’s Rothbaum highlights the way in which the French and German spectrum regulators are requiring MNOs as part of their spectrum licence obligations to provide rail passenger connectivity along the whole rail corridor. “Now that’s not to the same level as for critical communications,” he says, “but if they’re already making that investment towards passenger connectivity coverage as their licence requirements mandate, the [extra] investment [required] to provide adequate coverage with no coverage holes is probably small.” He also says that in two years it is likely that many railway corridors will be covered by MNOs’ 5G capabilities.

He adds that another problem with relying on MNOs to provide the connectivity for critical rail communication services is a legal one – ie, who would be criminally responsible in the event of a communications failure that resulted in a “life or death situation”?

Rothbaum also points out that given the EU’s goal is to retain the harmonisation we have today (ie, a train with GSM-R can travel across Europe with no issues), if MNOs are to provide this service, they would need to be using the same dedicated band of spectrum to do so.

There are other issues that may arise were MNOs to be used by train operators in the context of FRMCS, according to ECC Report 294 – Assessment of the spectrum needs for future railway communications. These include whether it would require certification of the MNO’s software updates and its implementation of new 3GPP releases, prior to them being deployed on the live network, which “could have [a] major impact on MNO evolutions and operations”, along with concerns over MNO lock-in, given the investments that an MNO would have to make (and which would have to be made again when switching to another MNO), together with the need for a “stable long-term relationship” between the train operator and the MNO – which creates risk given that the MNO could be bought by one of its competitors. The report adds that “when an MNO is acquired by another one, the new entity inherits the previous commitments. But a risk exists that services are terminated and existing contracts are not prolonged.”

Of course, it is not only Europe that is looking to embrace 3GPP technology for railway use. Rothbaum highlights Indian Railways’ request for proposals (RFP) for ECTS over LTE as a sign of its confidence in the technology, and adds that signalling vendors are developing ECTS products that will work over LTE in order to respond to the RFP.

In addition, Frequentis is delivering end-to-end 3GPP-compliant mission-critical communications over a private LTE network, using a public network as a fallback, for a major transport operator in the Asia Pacific region. This includes the user working positions and key technology applications such as mission-critical push-to-talk (MCPTT).

“We’re delivering all the mobile apps, the SDKs for the cab radios, all the application servers, the security measures for the applications, the dispatch system, the voice and messaging recording, the integration with the legacy system, and GIS integration,” says Myslivec.

We have seen that FRMCS has a lot to offer the railway community, its development is aiding the wider mission-critical communication sector, and the work to iron out the practicalities around the migration from GSM-R is gathering a head of steam. Given MNOs’ enthusiasm to serve many different industries, as opposed to just consumer and generic business communications, it will be interesting to see if this, coupled with networking slicing and other innovative techniques, can overcome some of the challenges around the use of public networks for mission-critical use. While the public safety world is coming to terms with using MNOs for mission-critical broadband, control of objects weighing hundreds, if not thousands, of tonnes may require even more confidence on the part of train operators.

The Finnish Transport Infrastructure Agency (FTIA) is now using TETRA via Frequentis’s URCA system, instead of GSM-R

TETRA’s railway role
Back in February, the Finnish Transport Infrastructure Agency (FTIA) deployed Frequentis’s Unified Railway Communication and Application (URCA) system, which makes use of the company’s FTS Bearer Independent Communication (BIC) solution, to allow it to migrate off GSM-R. This was driven by the GSM-R network’s operational costs and issues around interference that were being aggravated by greater use of 3G and 4G within the country. Frequentis’s Myslivec says “we had no major or even minor outages or incidents since we went live earlier this year and feedback from end-users and in particular from the management is very good – the URCA system is actually doing what it was intending to do”. He also sees little difficulty around eventually adapting the BIC system to work with FRMCS.

More recently, and as we reported in our last issue, Teltronic’s TETRA solution has been successfully integrated with the ETCS (European Train Control System) signalling application provided by Bombardier along the approximately 300km-long Zhetygen-Altynkol railway line in Kazakhstan, and has been certified by the line’s operator, Kazakhstan Temir Zholy.