How to get reliable wireless coverage on trains

For rail passengers in the UK getting a reliable cellular or Wi-Fi signal on a train is a patchy and frustrating experience. What’s even more aggravating is that in the rest of Europe many countries have had reliable wireless coverage on trains for years.

Norway, Sweden, Germany and Poland are now on their second generation in-train coverage refresh, while France, Portugal and Latvia are starting to roll out their solutions. The UK Government has committed to having Wi-Fi on 90% of trains by 2018, but there are still big question marks over how this might be achieved.

In an effort to address the problem, Cobham Wireless and RFS (Radio Frequency Systems) held a seminar in London on 25 May to examine why the UK lacks the equivalent reliable coverage and to look at what can be done about it.

The Government View

Gavin James, Programme Manager, Digital & Telecommunications Services, Rail Digital Services at the UK Department for Transport (DfT) opened the proceedings by noting that a survey of passengers 18 months ago put access to free Wi-Fi (or connectivity in general) in their top 10 wish list.

James said: ‘The operational side of connectivity should be higher up the agenda, but we need case studies of what has been done in terms of connectivity to help make the business case stand up for new projects. We need to prove the operational efficiencies that can be achieved.’

He reported that what Government ministers want is, ‘what passengers want’, which is: the ability to make voice calls (either native calling via digital on-board repeaters or Wi-Fi Calling, if it can be shown to be reliable); and access to the Internet.

‘Government will not get involved in anything that doesn’t serve all passengers,’ warned James, ‘although it won’t stand in the way of good commercial deals – a single operator service wouldn’t be that helpful, for example’.

He continued by saying that passengers need to be able to keep a phone call going in tunnels and cuttings and that perhaps different levels of service could be tolerated in terms of main lines and commuter lines (where even then there are still pretty high expectations). But he admitted a one size coverage service to fit all rail lines is probably not realistic.

That leaves the knotty problem of who should pay for wireless coverage. Should it be Network Rail, the train operating companies (TOCs) or the suppliers? James was in no doubt.

‘We think train operators should lead the charge,’ said James. ‘They know their customers, but there are distinct business cases. A main line train operator competes with short haul airlines, while the operator of trains in the South East has lots of shortish commuter journeys – the business cases are different.’

James said that Government is looking at what guarantees for wireless coverage it can include in new franchises coming up. ‘There is a commitment to say every new franchise must include free Wi-Fi, perhaps by 2018, but definitely by 2020.’

He pointed out that on-board coverage really applies to new rolling stock coming into service, as it makes no sense to invest the technology in old trains that will be taken out of service before too long.

‘We will leverage the Wi-Fi equipment on trains,’ said James, ‘ but what we don’t want to do as DfT is dictate how the technology could work, as we do not want to inhibit innovation. What we want is great connectivity proposals from bidders, but we don’t want to specify how they provide it and in fact that is very hard to specify in bid documents.’

He conceded that pushing train operators to be responsible for good Wi-Fi connectivity is a big ask and there is debate around what levels of coverage and service they should have to achieve and how to measure whether they do so. He invited anyone with views on how to specify this to get in touch with DfT.

Including provision for Wi-Fi in new franchise agreements is one thing, but that still leaves the question of how to support similar provisions in existing franchises. What incentive is there for operators to invest in wireless technology mid-franchise?

James explained: ‘We are looking at something called the Residual Value Mechanism. If you invest in Wi-Fi during your franchise and that value carries over into the next one, there is a mechanism to pay the franchisee back for his investment. What that does is make available to franchisees a way to get a return on their investment beyond the life of their franchise.’

He added that Government is trying to help with providing backhaul and power. ‘Network Rail gets some flak for being difficult to work with from the telecoms industry, but they operate in a difficult environment,’ he pointed out.

‘It is not easy to give people access to deploy new equipment in a live rail environment, but I think Network Rail is rising to the challenge. There are plans for how to get suppliers onto their infrastructure; how to share space on GSM-R masts, for example; what is the SLA for getting access to a tunnel; and can we do trials and pilots,’ said James.

Turning to the supply side, James said: ‘I get a sense that there is a quite a lot of nervousness about what suppliers can deliver and not just in telecoms. But come and demonstrate what your equipment does. Let’s test speeds and reliability and not just focus on cost.

‘It is about: how does it work and what risk premium needs to be built into bids, because we don’t quite know whether a supplier a can work with Network Rail and train operators to deliver a solution. But I assure you Government is really behind this,’ James concluded.

The Independent Wireless Expert View 

Oliver Bosshard, managing consultant at independent wireless experts Real Wireless, outlined the key business and technical options and challenges for successful wireless coverage on trains.

He started by pointing out that one of the challenges is the number of parties involved: mobile network operators (MNOs), Government, passengers, TOCs, ROSCOs (rolling stock operating/leasing company), infrastructure managers and the telco industry supply side.

Bosshard said that everyone struggles with the financing and business model as all parties want to benefit and, if possible, make money. Technical solutions are available but the environment is challenging with tunnels, cuttings, limited room and access for on-board equipment and too many parties to corral.

Enablers include: Government pressure and regulatory change (shared spectrum might help, for example); DfT subsidy; the new 4G Emergency Services Network (ESN); and potentially - neutral host providers.

‘The challenge for MNOs is that rail coverage is expensive and will not increase subscriber ARPU, so there isn’t a lot of incentive for them to invest in it, but they need to be involved because they have the spectrum.’

Subscribers want access everywhere and just see it as a utility, for which they do not want to pay more than they already do. The MNOs are investing huge amounts of money into network infrastructure for 3G and 4G, for which they need an RoI, but the gap between revenue and traffic continues to widen, Bossard noted.

So, who will finance public wireless for rail? The options are: the operator owned model (which is not happening); the landlord owned model, where they invest in their own infrastructure (as with ESN); or the neutral host owned model, which Bosshard sees as a growing development, and an opportunity.

The best approach involves consortia of all parties with a joint investment and upgrade strategy. However, Bosshard said: ‘Neutral host could be the enabler as that might unlock some of the deployment barriers with a slow RoI over 10-15 years.’ He noted that in the USA, 70% of telecoms infrastructure is owned by third parties, so they need to be involved.

Assessing the technical options, Bosshard said the solution involves a combination of the right technology, topology and spectrum, but it generally involves a distributed antenna system (DAS) and onboard kit (repeater, Wi-Fi or cellular small cells).

‘All solutions need a certain level of outside coverage and aim to overcome the vehicle’s signal attenuation,’ he said. ‘A train is like a sardine tube full of bodies that block signals, so a leaky feeder maybe better in this situation than on-board antennas.’

The options comprise: a passive outside in solution which has to overcome the train’s metal structure and Faraday cage effect; or the use of onboard kit to boost the outside signal – either an onboard RF repeater, which does not require a backhaul solution, or a HotSpot, which does.

In terms of spectrum, he noted that there is actually a lot of spectrum out there - MNOs have c.650MHz of licensed spectrum – and not all of that is used in rail yet. MoD spectrum is due to be auctioned later this year; and there are a further 11 bands above 6GHz, which might be used.

‘These are all fairly high frequency and it is too early to know which ones will be harmonised for IMT use,’ said Bosshard. ‘There is a lot of work happening on 5G, but it is far from being defined yet and the proposed use of millimetre wave spectrum is not what you’d want for outside-in train coverage.’

He felt that Wi-Fi is a good solution, but it has its limitations and MulteFire (one of the options for using 4G in unlicensed spectrum) is coming too, which may have some possibilities.

However, in Bosshard’s view Wi-Fi’s basic access technology, Carrier Sense Multiple Access – Collision Avoidance (CSMA-CA) is not designed for high-density environments, such as busy commuter trains in rush hour with high capacity demand caused by a large number of concurrent users.

The issue here is that when many simultaneous users in a small area are all trying to access the Wi-Fi, to avoid collisions the devices ‘sense’ if the channel is busy and back off if it is – and keep on doing so – and so the Wi-Fi access point is overloaded and its efficiency deteriorates.

Others argue the latest Wi-Fi equipment can handle high density environments and that most trains are not that crowded (although try telling that to a Southern Rail commuter).

Bosshard went on to say that cellular small cells might be a better solution than Wi-Fi because they are more efficient at handling a high number of concurrent users and high traffic — but that still doesn’t solve the backhaul challenge.

‘Performance is always limited to whatever the backhaul can achieve, which is typically 4G. Hence, if there’s no mobile coverage, the whole on-board connectivity system — whether it’s Wi-Fi or femtocell based — is useless,’ he pointed out.

Bosshard concluded that none of these technical challenges are insurmountable, and that the real challenge is finding a commercial and business model that works for all parties.

He added that tunnels bring another set of challenges with each one requiring its own solution. However, a typical rule of thumb is that large and wide tunnels with a lot of overhead space can adopt an antenna solution, but for tight tunnels the more costly option of leaky feeder solutions provides a better performance.

The Supplier View  

Anthony Sutton, Coverage Sales Director at Cobham Wireless, and David Ineson, Director at Siroda, a specialist consultancy dealing in all aspects of radio network measurement for the telecoms sector, provided an overview of their experience of deploying connectivity solutions on trains with particular reference to their work on East Midlands Line trains.

There are several reasons why trains need repeaters on board to amplify wireless signals from outside. Firstly, trains pass through coverage not-spots and while the TOC might have a deal with one mobile operator, that’s not much help for any passenger on other networks.

Secondly, as is well known, the metallic coating on trains reflects RF signals (c.25dB loss) to create a Faraday cage that interferes or even blocks signals entirely. Lastly, trains travel at high speeds so maintaining call and data sessions between base stations becomes difficult.

But everybody wants reliable connectivity. The rail travellers because that is what they are used to; the MNO because it wants to gain or maintain higher customer satisfaction through no dropped calls; and the train operator, because it provides greater customer satisfaction, increased safety, better utilisation of trains, and potentially provides a key selling point over other rail operators.

The usual solution is to take the signal from outside the train and bring it inside: do whatever is necessary to improve the signal inside the train; and then stabilise it for consistency throughout the length of the train.

This is done by installing a roof antenna, a digital onboard repeater and implementing radiating mechanisms along the train either via leakage cable or small in-train antennas, which might be discreetly fitted in the carriage ceilings, for example.

User experience research by Siroda shows that when this is done the received signal level improves inside the train; passengers get fewer dropped calls; and the signal strength is better (which makes better use of the phone’s battery, thereby prolonging its life).

Sutton and Ineson explained that the process of implementing wireless coverage on railways begins with train and track coverage surveys to see what the RF environment is. This is followed by the conceptual design, agreement from the leasing company, and the MNOs – and then the safety case is signed off, followed by deployment of the solution and then monitoring to see if it has achieved its goals.

Finding suitable locations for the in-train equipment can be problematic as space is at a premium to install the repeater(s). For their work on East Midlands Line trains, Cobham and Siroda installed the repeater in a cupboard in the catering carriage. They found that for East Midlands Line trains a centrally placed repeater covers a maximum of five carriages (this depends on the power output and type of inter-carriage connection), so for a seven carriage train they needed to install two repeaters.

The repeater is designed to overcome signal loss coming into the train. It amplifies the external MNO coverage from nearby cell sites, but it also boosts the user terminal signal back out by 70-80dB, so the mobile phone doesn’t have to run a full power to get its signal back out, thereby saving the battery.

The next task is to identify the cable routes along the carriage and to decide on the inter-carriage connections. Sometimes installers are able to use existing infrastructure (light fittings, for example) but at other times they have to install new equipment.

When it comes to monitoring and measurement, Siroda‘s Dave Ineson said: ‘We sit on the train with our equipment and measure the user experience. The equipment has to be able to cope with the high speeds and be both reliable and portable.

‘It must be able to capture everything that is data important. It needs to scan fast, so you need a good receiver and the ability to process the data fast, otherwise you only get measurements every half kilometre and that is not good enough.’

Cobham’s Anthony Sutton noted that the key aspects of any in-train coverage system are: the basic system architecture, link budget, predicted coverage within the train, finalisation of equipment locations, and managing the phased installation of equipment, so that the train is only out of service for its usual every 72-hour check.

The Mobile Network Operator View 

Mansoor Hanif, Director of Radio Access Networks, at EE outlined the operator’s aspirations for the rail industry and its desire to use innovative coverage solutions across 100% of the UK. This includes an ambition to provide coverage in railway tunnels and to infill coverage in railway cuttings.

In rural areas, EE is working on a variety of solutions including small cells harnessing parent company BT infrastructure and products; small cells with in-band backhaul, and others with satellite backhaul.

Hanif said EE plans to deploy a 4G footprint covering 95% of the UK, but for the railways, he said EE is ‘prepared to deliver contiguous service to TOCs on a commercial basis’, so that means someone else has to pay for it.

EE has surveyed and analysed 66 rail routes between 2013 and 2015 as part of its efforts optimise the Orange and T-Mobile networks following the merger of the two to form EE. The surveys are used to analyse coverage, highlight areas for improvements and shape its 4G rollout.

Hanif pointed out that one option is for an MNO to invest in a coverage solution with a TOC and then offer the system to other operators. For example, Orange fitted on-board signal amplification systems on Class 390 Pendolinos and Class 220 and 221 Voyagers beginning in 2008 and the offered the system to O2 and Vodafone.

In contrast, Vodafone has done something similar on the East Midlands Class 222 train fleet, but this has not been offered to other operators.

Hanif outlined EE’s vision for what he called ‘The roadmap to 1Gbps’. A combination of technologies such as: macro site upgrades with 5CC LTE-Advanced and 256 QAM; meshed small cells in in-band backhaul on trains with possible 4G backhaul over satellite; UAV-based track maintenance and relay; next-gen repeaters, leaky feeders and meshed small cells for tunnels; and app-based optimisation for railways e.g. VPN active management in the Cloud.

He also noted that better access to Network Rail’s trackside towers with fibre would be helpful. ‘I feel there is a willingness to address issue of access to trackside and GSM-R towers,’ said Hanif.

The final piece of the puzzle is to use mobile edge computing technology in the shape of a mobile edge server deployed on the train with on-board cacheing functionality.

In Hanif’s view it is not enough to just cover the track; instead the focus needs to be on providing an end-to-end, station-to-station experience. This might include guiding people to the train using location based apps and indoor navigation logic and hosting local maps with real time overlays.

Once on the train, besides providing reliable voice calls, a wider digital experience could be provided through high performance connectivity and locally hosted-content (via the on-board edge server cacheing facility). You could even provide augmented reality, correlating raw images with local databases.

Hanif said: ‘We have a ready to go solution with partners lined up and we are looking for support to do a pilot.’ The pilot would involve using EE’s RAN, Nokia 4G eNodeBs and Akamai edge servers. ‘We’ve tested the solution in the lab and it works, so now we want to test it in real conditions, which we may do with Chiltern Railways.’

He then outlined what EE is doing to beef up its infrastructure to create an ‘ultra-reliable network’ something it must do as the UK’s new Emergency Services Network (ESN) will be running on EE’s network – so it must be upgraded in terms of availability, robustness and resilience to meet mission critical standards.

This includes a mix of new site builds, implementation of 800MHz spectrum, carrier aggregation and Voice over LTE for coverage and capacity. Resiliency plans include: battery backup, fixed and mobile generators, resilient transmission, flood defences and rapid response vehicles.

A range of new operational processes, new 3GPP features (security, prioritisation, etc) and temporary solutions (air masts, mobile network in a box, etc) are also being implemented.

However, Hanif warned that while building out its network to meet the demands of ESN will improve it greatly, and may therefore help in some geographies, ESN should not be seen as the answer for ubiquitous rail coverage.

Hanif finished by outlining the proposals for getting broadband coverage into the London Underground and some pointers towards what 5G might bring as exemplified by early proof of concepts and field trials currently being considered or underway.

The day also included two presentations by Dr Peter Raabe, Head of Strategy and Portfolio Management at RFS, on the company’s product portfolio and some key case studies, and a presentation by Cobham Wireless of the solutions it has provided for the Oslo Metro system.

Image Credit: Sergey Nivens / Shutterstock

Desire Athow
Managing Editor, TechRadar Pro

Désiré has been musing and writing about technology during a career spanning four decades. He dabbled in website builders and web hosting when DHTML and frames were in vogue and started narrating about the impact of technology on society just before the start of the Y2K hysteria at the turn of the last millennium.