Selecting Rail Telecoms Networks that Are Built to Last
The telecoms network is the nervous system of any rail operation, carrying everything from safety-critical signalling and train control to passenger information and CCTV.
www.nokia.com
© Nokia
The telecoms network is the nervous system of any rail operation, carrying everything from safety-critical signalling and train control to passenger information and CCTV. With the gold standard for availability being 99.999% – better known in the industry as the ‘five nines’ – the margin for error is tiny, as this represents just five minutes of downtime per year.
Even so, procurement decisions are still too often driven by upfront hardware costs rather than the true lifetime cost of keeping a network running. The focus should instead be on building the most reliable network in order to avoid failures, as well as minimising downtime required for maintenance operations.
Achieving this comes down to three things: network design, hardware and operations.
On the design side, the goal is to ensure that any single failure is resolved by the network itself, without impacting the supported devices, applications or services. This often entails implementing hardware redundancy: duplicating systems to ensure a single point of failure won’t impact the network as a whole. In signalling for example, we use what’s known as the blue and red architecture; two devices at each location, each connected to separate switches, routers and fibre paths so that if one network fails, the application traffic will still be transported via the other network.
Preventing failure matters, but so does how fast you recover when the worst case happens. Therefore, when it comes to operations the goal is to reduce your mean time to repair (MTTR). If something does require intervention, how quickly can you fix it? The faster you can get the right person with the right part to the right place the better, which is why regionalised maintenance and spares logistics are as important as the network design itself.
The Right Equipment
Equipment choice matters as well. In core network switches, full redundancy means duplicating both the management card and the switch fabric. Some vendors only limit redundancy to the management card, so when the switch fabric fails, the network goes down. In this scenario, you could be looking at an emergency callout and an end to retaining your five nines availability, plus any CAPEX saving from choosing partial over full redundancy will be wiped out.
The same applies at the edge. Redundant power supplies that aren’t removeable will require the device to be taken offline for replacement
.
Digital generated image of abstract growing bar graph made out of glass standing on reflective surface.
© Nokia
Full redundancy also pays dividends during planned maintenance. When you need to upgrade software or swap a component the work is faster, lower risk and easier to reverse if something doesn’t go to plan.
Equipment designed for IT environments where rear access is standard and space isn’t a constraint makes this worse. In a typical railway location, what should be a two-minute swap becomes a 15-minute outage: cables off, chassis unracked, module replaced, everything rebuilt. With a unit designed specifically for a rail environment, one engineer removes the failed supply and plugs in a replacement with zero service impact.
Redundancy rarely features as a serious evaluation criterion in procurement and IT-oriented equipment turns up in critical rail environments more often than it should, and these are both things operators pay for further down the line.
The Cost of Change
Choosing the right equipment reduces unplanned downtime, but planned maintenance has its own costs, and while the pressure to keep intervention to a minimum varies between metro and mainline, the reality is it’s acute in both.
In metro operations, the engineering window is often just four hours when services pause overnight, and every change must be fully reversible if something goes wrong. Due to this complex environment and short window of time, maintenance costs are high because they require highly-skilled staff working nights.
On the mainline, any maintenance touching the signalling network requires an empty train to run the route afterwards to validate that everything is functioning correctly before passenger services resume. That’s a significant operational cost, and a strong incentive to keep network changes to a minimum.
Physical Network Segregation
The maintenance window problem is compounded by a growing tension at the heart of modern rail operations: the two very different worlds that share the same network infrastructure.
Railways run two fundamentally different categories of application. Safety-critical systems such as train control, signalling, train-to-ground communications, are highly specialised and have long lifecycles of up to or over a decade.
Non-safety systems tell a different story. CCTV, passenger displays, IP phones and trackside flood sensors are built on commercial, off-the-shelf technology that demands frequent updates and can be exposed to cyber vulnerabilities. They’re also rolled out at a higher rate as networks expand: add a new batch of sensors and you may need a new switch, which means a network change that drags the safety-critical systems into scope and requires a costly, regulator-approved maintenance window.
Physical network segregation is gaining traction as the solution. It sounds expensive, effectively running two separate networks, but the physical overlap between safety-critical and non-safety infrastructure is typically around 20%. At that level, the additional infrastructure cost is comfortably offset by the reduction in maintenance windows.
One of the best examples of a fully segregated, highly reliable network comes from Swiss Federal Railways (SBB). This operator has maintained dedicated networks for safety-critical and non-safety applications for over a decade, with redundancy built in at every level and maintenance factored into procurement from the outset. The results speak for themselves; Swiss trains are consistently the most punctual in Europe.
Getting Procurement Right
While the case for network segregation is increasingly well understood, the broader lesson is that the right infrastructure decisions need to be made long before any of these problems arise.
This is because the decisions made at the tender stage can impact years, sometimes decades, of operations. Even so, hardware price still dominates the discussion, when what actually determines total cost of ownership is product design, software quality and lifecycle.
A €100 saving per router looks reasonable until the first failure costs three times that price to resolve. The real questions to ask are how often will software upgrades be required and how much will maintenance callouts cost?
Technical teams need to evaluate redundancy requirements seriously. A fully redundant system and a partially redundant one are not equivalent products and shouldn’t be scored as such. Procurement teams need to give less weight to unit price and more to operational cost over the life of the network. The two need to be aligned, because the network underpins safety, reliability and the capacity to innovate. Just remember, the network that looks cheapest on paper may prove the most expensive to run.
www.nokia.com
