The future of GSM-R?

One of the benefits of attending any major industry conference is the opportunity to network and to renew old acquaintanceships. Having been involved with the development of GSM-R since 1993, when the UIC first decided it needed an international standard digital radio system, the recent World GSM-R Conference presented just such an opportunity. It was great to meet up with former colleagues from different railways and companies. Even more so when it is so long since I’d met them that I just couldn’t put a name to the face – or perhaps that was just old age.

The UIC GSM-R Conference is held biannually to update railways and manufacturers on present issues and future developments. This year’s conference, held recently in Paris, examined the success of GSM-R in becoming a global system, just like its forebear GSM. However issues surrounding security and availability challenges, improving assessment of equipment, the problem of interference from newer 3G and 4G public mobile phone networks which use wide bandwidth radio channels, and the evolution and technology for future systems that will take over from GSM-R, are all topics that have to be resolved.

The good news for the railway industry is that the GSM-R Industry Group has agreed to support GSM-R systems until 2030, but of course this doesn’t mean that changes will not take place before then. Instead, just like public mobile networks, the mobile radios will continue to support GSM-R functionality even when a new radio bearer technology is introduced. This means that there will be mixed networks and multi-functional mobiles during the changeover, and possibly for ever.

Two networks compared

The key issue is the period anticipated for the definition of new technology and development and approval of new equipment. One of the benefits for the railways is that they will have skipped a generation in the development cycle, having missed out 3G (UMTS).

Speakers from two countries, both of which started planning their networks in 2006, made for interesting comparisons. China now has the largest GSM-R network in the world, with 33,750 route km covered, over 10 times that of Austria which has only 3,200km. Neither network is completely finished.

Christian Sagmeister from ÖBB was able to define some of the benefits following the switch from analogue networks to GSM-R – a cost reduction of 10%, improved availability, additional functionality and the support for ETCS Level 2 which is already operating in some areas.

In contrast, the China Rail speaker, Liang Yiqun, concentrated on the future. He proposed that, to change technology from GSM-R, new frequency bands for railway use should be sought from the World Radio Conference 2019.

This would be a major change. Currently Europe has a common radio frequency allocation for railways of 2x4MHz in the GSM 900 MHz band, with a further 2x3MHz available in some countries in busy areas.

Having led the successful negotiations for the original radio spectrum, I know how difficult it was to get agreement through the CEPT, and now the Electronic Communications Committee (ECC). Other countries have to negotiate separately for their spectrum allocation so that some areas of the world have finished up with a non- standard (for railways) radio spectrum – one example is New South Wales with a spectrum in the 1800MHz band.

A change to a fully global spectrum allocation would obviously improve interoperability and reduce manufacturing costs but would be extremely unlikely to be successful because of the competing requirements from other users.

Cyber security threat

The conference session on security and availability concentrated on cyber-attack on signalling and communication networks, particularly as the use of ERTMS spreads. Virginie Deniau from IFSTTAR (French institute of science and technology for transport, development and networks) and coordinator of the SECRET (security of railways against electromagnetic attacks) project, is a specialist on this very subject.

She explained that SECRET assesses the risks and consequences of electro-magnetic (EM) attacks on the rail infrastructure. It develops preventive and recovery measures as well as protection solutions to reinforce the security of a rail network subjected to intentional EM interference with the intention of disturbing command- control, communication or signalling systems.

Some EMC (Electro-Magnetic Compatibility) recommendations will also be provided to more specialised standardisation committees at hardware level within CENELEC, CEN or ETSI. Jean-Michel Evangehelou of Kapsch CC described its approach to optimising recovery from failure or attack to provide 99.9995% availability of GSM-R networks. This was based on duplicate MSCs (Mobile Switching Centres) operating continuously, rather than one being on standby, and duplicated MSC-BSC (Base Station Controller) links.

Combating interference

The RF interference into GSM-R from 3G and 4G public networks, which are replacing GSM in the 900MHz band, is a key issue affecting railways operators today. It is prevalent where the 3G and 4G transmitters are close to the railway and the GSM-R transmitter (BTS) is some distance away so that the signal strength of the GSM-R transmission is blocked by the high-level out-of-band transmissions or intermodulation in the mobile receiver.

This has been demonstrated by two reports of measurements carried out in 2013 in the UK and Germany. Whilst previously the interference from Public GSM was resolved by coordination (i.e. changing the channel in use), because 3G and 4G transmissions are wide band and have higher out-of-band transmissions, this is no longer possible. ECC Report 229 identifies the solutions available:

1. Improve the public network transmitters by fitting filters to reduce the interference levels into GSM-R cab radios;

2. Increase the power output of GSM-R BTS (Base Stations) or add additional BTSs – a costly and not always practical solution;

3. Improve the train mobile receiver signal by fitting better train antennas, though this can be of limited benefit;

4. Fit external or internal filters on the train which will resolve the issue but is costly and prevents roaming into the E-GSM band which is mandated in the GSM-R specifications;

5. Improve the mobile receiver which, depending on the design of the train radio, could mean fitting a new receiver module or alternatively a new radio unit. ETSI has produced an improved specification (TS 102 933-1 v1.3.1: RT; GSM-R improved receiver parameters) that will be referenced in the Railway GSM-R specifications for purchase of new equipment.

The Mobile Fixed Communications Networks (MFCN) viewpoint on this issue was presented by a representative of Orange France. He agreed that there were common objectives with the railway community. Nevertheless, he insisted that corrective action in their networks is less efficient than any action taken at the cab radio and that a European deadline for a transition period is necessary, during which additional mitigation measures will be required to avoid GSM-R interference.

Suppliers such as Sierra Wireless and Funkwerk have developed filters and new receiver modules and some governments (for example Sweden and the Netherlands) are providing finance to the railways to speed up the fitting process. Sweden has decided to push ahead with fitting all train units with filters, although this is unlikely to be the cheapest solution in other countries as evidenced by the emerging costs in the Netherlands of €2,000- 4,000 to supply and fit a new receiver and €2,500-5,000 for a filter.

In response to a question, Otto van Rooy of the Netherlands Ministry of Transport considered these actions would future proof its GSM-R system for the foreseeable future.

Short-term enhancements

The conference proceeded to examine the move from the current GSM-R to IP-based next-generation systems, a development which has already been defined by the European Railway Agency (ERA), the UIC and the GSM-R Industry Group.

Frequentis explained that GPRS, the first level GSM Packet Switching system, is being implemented on most GSM-R core networks to Release 4, so enabling packet switched ETCS to be implemented.

Klaus Mindel, representing UNIFE/UNISIG, noted that GPRS has already been included in the ETCS SRS 3.5.0. He suggested that a radio bearer independent structure for ETCS is a priority for suppliers so that, in the future, flexibility will exist in its implementation around the world. Whilst that was clear for a future implementation, it didn’t explain how this would support interoperability in Europe today.

GPRS has already been shown to give a major capacity boost for carrying ETCS Level 2 data from trials on the SNCF high speed line. Transmission delays for movement authorities averaged around one second for the current circuit switched implementation (with one user per channel) and 800ms for GPRS with four users per channel. However, the spread was greater with GPRS, being up to 2.35 seconds as against 1.4 seconds for circuit-switched. This is why, in the longer term, a higher speed data transmission system needs to be implemented.

SBB (Swiss Railways) thinks that, in the short term, this will be EDGE, an improved version of GPRS, which will carry up to 14 parallel sessions per GSM-R radio channel rather than a maximum of seven with GPRS. So far, no tests have been carried out on the vulnerability of ETCS data using EDGE to RF interference, so there is a question mark here. SBB was also of the view that the data link supporting ETCS should operate on an unacknowledged basis, because the acknowledgement extends the time of the transmission process too much.

Longer term solutions

The next session dealt with preparing for the future. THE UIC/ERA/EC Future Railway Mobile Communication System (FRMCS) project aims to have a completed specification by 2018. This will be a mix of ETSI and railway-specific standards.

The first step is an update on the user requirements, which should be available at the end of 2015. This is being followed by the Architecture and Technology Group report in Q2 2016 and a Spectrum Group report. It is apparent that the LTE standards (Long Term Evolution or 4G) will almost certainly be adopted. It is anticipated that the railway- specific features will be totally independent of the radio communication bearer, making the next generation railway systems being defined, not as LTE-R, but LTE plus R.

However there are issues to be resolved, for example – will dedicated radio frequencies be available for the railways, particularly for ETCS support? And what about direct train-to-train mode, and group calls? Functional addressing and location-dependent addressing are nominally outside the radio network but dependent on it.

One of the related issues is the choice of frequency band the railways should be pursuing. This appears an open question currently in Europe. If the compatibility issues can be overcome, the best option is to stick with 900MHz, but with the flexibility of LTE perhaps the use of a higher frequency band (such as 2.6GHz) could be adopted in busy areas. One of the suggestions is that sharing a network with the emergency services, as is currently about to happen in Finland, could be the way forward.

Tony Gray from the TETRA and Critical Communications Association (blue light organisations) sees their future in LTE but believed that Release 12 implementations, available in 2017, would contain too many propriety interfaces. Release 13 (end of 2018) would be standardised at the right level, but would take until 2023 to be fully proven. This matched the view of Cheil Spaans of the European Railway Agency that possible deployment of LTE by the Railways should not be before 2022.

Explaining the technology options

So what is LTE? The following explanation is based on the content of the 3GPP website where more details can be found. LTE is the access part of the Evolved Packet System (EPS). The main requirements for the new access network are high spectral efficiency, high peak data rates, short round trip time as well as flexibility in frequency and bandwidth.

GSM was developed to carry real time services, in a circuit switched manner, with data services only possible over a circuit switched modem connection, with very low data rates. This has been adopted for the first implementations of ETCS Level 2. The first step towards an IP-based packet switched solution was taken with the evolution of GSM to GPRS, using the same air interface and access method, TDMA (Time Division Multiple Access).

To reach higher data rates in UMTS (Universal Mobile Terrestrial System), a new access technology named WCDMA (Wideband Code Division Multiple Access) was developed. The access network in UMTS is a hybrid as it emulates a circuit-switched connection for real-time voice services and a packet-switched connection for data services. This step in UMTS development is where the IP address is allocated to the User Equipment (UE) when a data service is established and released when the service is released.

The Evolved Packet System (EPS) is purely IP-based. Both voice and data services will be carried by the IP protocol. The IP address is allocated when the mobile is switched on and released when switched off.

The new access solution, LTE, is based on OFDMA (Orthogonal Frequency Division Multiple Access) and in combination with higher order modulation, large bandwidths (up to 20 MHz) and spatial multiplexing in the downlink (up to 4×4) high data rates can be achieved. The highest theoretical peak data rate on the transport channel is 75 Mbps in the uplink, and in the downlink, using spatial multiplexing, the rate can be as high as 300 Mbps.

The LTE access network is simpler than GSM, being based solely on a network of base stations and so generating a flat architecture. There is no centralised intelligent controller, and the eNBs (Evolved Node B, basically an intelligent base station) are normally inter-connected via the X2-interface and towards the core network by the S1-interface. The reason for distributing the intelligence amongst the eNBs in LTE is to speed up the connection set-up and reduce the time required for a handover. The railway special requirements, such as functional addressing, will have to take place outside the radio network.

LTE in 2023, or 5G?

The conference provided a useful update on the current GSM-R technology with pragmatic options for overcoming some short-term difficulties. There is a solution for the current interference problems, albeit at a cost, and the financing of this has to be a national issue. Packet-based transmission is able to support ETCS and the implementation needs to be progressed with a degree of urgency.

The changeover to LTE is unlikely to start until 2023, although the flexibility of the ongoing programme will mean that current GSM-R features will be supported up to 2030. The railways need to work together to support negotiations for frequency allocations as this is still a major issue for the future.

The assumption that LTE will be the successor to GSM-R may be challenged by some as, by the time the specification and planning activities have begun to happen, the 5G specification will be finalised and early deployments implemented. This may be a better solution and no doubt many will voice opinions as time progresses.

Written by Les Giles