Train leasing firm demos mobile edge computing for smart trains

Rolling stock operating company (Rosco) Angel Trains is demonstrating the power of mobile edge computing over a mesh network to enable train operating companies (TOCs) to take advantage of innovative on-board applications and services without having to wait for mobile network operator-driven network upgrades.

With customers including Great Western Railway (GWR), Merseyrail, ScotRail, SouthEastern, South Western Railway and Virgin Trains, Angel Trains owns about one-third of the rolling stock in use on the UK’s rail network, and plays a vital – and virtually unnoticed – role in keeping millions of travellers moving every day.

The project, “First of a kind: demonstrating tomorrow’s trains today”, which is being funded by a grant from the Department for Transport (DfT) – is being run together with Bath-based mobile edge compute company Virtuosys, whose Edge Application Platform will enable the trial.

Virtuosys bills itself as something of a “thorn in the side” of mobile operators, according to Dave Rose, vice-president of sales and business development, because whereas operators tend to look to the cellular tower to do edge computing, Virtuosys’ goal is to bring edge computing as close to the user endpoint as physically possible.

Its platform comprises a Linux-based wireless mobile edge server running on ARM processors, which offers multiple connectivity options, including Wi-Fi, Bluetooth and ZigBee. This is coupled with management and authentication software that can be run in public or private cloud environments, or even hosted locally and controlled through a smartphone or tablet.

“With us, the software developer has the option of deploying applications in the cloud, or as close to the endpoint as possible, the choice of deploying across the entire mesh network, or just on one server,” said Rose.

“The whole point of mesh networking is it’s a distributed compute environment, so from a software programmer’s point of view, he sees it as one big lump of resource, and we distribute that resource across the network. It also means that if one server goes down, the network is self-healing.”

Rose added: “The work we’re doing with Angel Trains is a really good example of where edge computing works.”

James Brown, data and performance engineer at Angel Trains, said edge computing had come onto his radar as a way to deal with the ever-increasing demands made on trackside networks and mobile availability, which are often patchy, if not non-existent.

“We know connectivity for passengers is a key requirement, yet the problem we see is that even though a number of players are investing in mobile connectivity for railways, passenger demand always increases to match whatever capacity you provide,” said Brown.

“If we want to keep up with expectations, we can’t just look at backhaul, we need to minimise the demand for data.”

The trial will attempt to minimise this demand on the external backhaul network by using edge servers in a mesh network, combined with local storage and processing, covering the train with a wireless local area network (WLAN) to run applications that target TOC-specific use cases and would more usually require external mobile network coverage to run.

The proof of concept will use two servers per carriage on a standard trainset, enough to provide connectivity down the entire length of the train, with redundancy should one of the nodes fail.

The idea is to show that the network can operate with only an intermittent backhaul connection to the internet, while the applications running over can still provide useful functions to the train crew during travel.  

“The ones we are demonstrating as part of this grant are a CCTV access application,” said Brown. “We will fit wireless cameras to trains that will send a feed to the edge nodes to be accessed by the crew on a tablet wherever they are on the train. Trains can be up to half a kilometre long and the train manager can’t be in each carriage at once.

“We will also trial a passenger announcement application, installing wireless speakers that the train manager can control over their phone to avoid issues you get on busy trains, where they have to fight their way through a crowd to make an announcement.

“We’ve also got – not as part of this grant, but already a proof of concept – an application for local messaging for staff communications.”

What the mesh network will not do at this stage is to supply web content to passenger-owned devices, but Brown envisages some immediate use cases that passengers could take advantage of. The on-board storage could, for example, cache the TOC or National Rail Enquiries website so passengers can still see timetable updates or news of disruption when the train is “off grid”.

One day, the messaging system could also help to improve customer service, extending the staff comms system to passengers to assist travellers with disabilities, or to address complaints more quickly.

“If passengers have complaints, their typical response is to go on Twitter and talk to customer services and hope that whoever is running social media can get a message to the train manager,” said Brown. “Why shouldn’t you be able to do that locally?”

In the long term, such a system could even store media content for local distribution as an at-seat entertainment service. Virgin Trains already operates such a service, although it relies on the on-board Wi-Fi.

The demonstration will run for six months from April 2018. Brown said that by the end of that period, Angel Trains hoped to have a live environment running on an in-service trainset – beyond which the ease of installation meant the technology could potentially be rolled out very quickly across the firm’s estate without having to take rolling stock out of service for upgrades.