Ethernet has been proven in many industries to be a very effective technology for providing data communications backbones.
Ethernet switches featuring a range of networking protocols are now being used to build these networks on trains, allowing a state of the art TCS (train control system) to manage every aspect of the train operation.
The onboard rail environment is one of the toughest imaginable for any electronic devices. An Ethernet switch which is going to be used in a network on a train has to exceed the tough EN50155 standard which covers a wide range of environmental requirements and ensures that equipment can survive the rigours of onboard operation.
A primary requirement for the Ethernet switch will be exceptional reliability. Unscheduled maintenance work is massively expensive, not just in the cost of the repair, but also in the reduction of service time of the affected train. Reliability of equipment can be assessed using MTBF (Mean Time Between Failures) figures as defined in the MIL-HDBK-217 standard.
The challenge for the designer is to maximise this figure but still reach the requirement of the EN50155 standard. For instance, the equipment must be able to operate from a wide DC voltage input ±40% as well as surviving high transients and having a long holdup in case of power loss. These factors introduce circuitry that can dramatically reduce the calculated MTBF figure. Despite this, by using high quality components it is still possible to have designs exceeding 350,000 hours.
Even when designed to be reliable, it is important that maintenance, when required, is still straightforward as, when network components have to be replaced, it will be done by a maintenance engineer with little or no network expertise. This engineer, wanting to replace a suspect switch in a network, will wish to be able to take an un-configured spare unit off the shelf, simply unplug the old device and plug in the new.
This is possible if a copy of the configuration is stored on a removable memory device which is constantly connected to a port on the switch. As long as the operating system of the switch can support fluid configuration techniques, with the configuration downloaded from a USB stick at power up, then this is relatively easy to achieve.
Automatic Network Configuration
Carriages in train sets may be replaced from time to time, and whole sections can end up being reversed as these carriages are joined and rearranged. Any network must be able to automatically reconfigure to allow for these changes, as the personnel undertaking the change will be unable to manage the network.
For example, it is the job of a TCS to ensure that the correct doors open when at a station and that the information appears on a specified display in a particular carriage. The train control system must rely on the network to be able to report if it has changed and then automatically reconfigure to allow for that change.
A combination of networking protocols such as OSPF (Open Shortest Path First), LLDP (Line Layer Discovery Protocol) and LACP (Link Aggregation Control Protocol), allow networks to be dynamically changed as the cars and carriages are changed around without the need to call on a networking specialist to reconfigure the network. Each device on each carriage should be “plug and play”
With Ethernet connections being used for many train subsystems from the actual TCS to PIS (Passenger Information Systems) and infotainment systems, it is possible that devices could be connected to the wrong port on a switch or excess data could be generated, hence jamming the network.
Also, with so many sub-systems and devices operating on the network, there are risks that data congestion could occur if mistakes are made on connection. Video traffic requires high bandwidth and hence must not be able to interfere with critical operation data such as that controlling braking or drive.
Firewalls can be used to block all but the required data from certain sections of the network. VLANs (Virtual Local Area Networks) can also be used to segregate networks, and prioritisation of data can guarantee delivery of critical packets on time.
Switches are often required to be mounted behind wall panels or in tight spaces. The design must be compact, with the depth of the product often being the critical factor. The ability of the switch to provide firewall functionality and, for instance, secure remote access support is important as this can remove the need for another box.
There will be instances of power failure to cars within a train set, and cables may occasionally be damaged. The network must be able to cope with these points of failure to prevent service delays or even safety issues for the train operator.
Ring structures can easily be created on a carriage to provide some form of network resilience locally. This can be provided using layer 2 protocols such as RSTP (Rapid Spanning Tree Protocol) or proprietary rapid recovery systems like Westermo’s FRNT (Fast Recovery of Network Topology).
It is impossible to have separate paths between cars, so dual backbones are used to provide some resilience. Protocols such as OSPF and LACP can be used to control these links but, in the event that power is lost to a carriage, built in relays on the switches can ensure that a network path is provided through the powerless carriages.
Manufacture and testing of equipment
The EN50155 standard even goes so far as requiring mandatory functional and isolation tests on all units as part of the manufacturing process, meaning that companies wishing to build switches to meet the requirements of this market must have excellent quality control over the production process.
All in all, it is tough for an Ethernet switch to operate reliably and for many years on a train. Only companies with a deep understanding of Ethernet IP technology, with high quality production facilities and onboard expertise, can offer products that are going to be fit for purpose.
Westermo is a leading innovator within the onboard rail Ethernet switch market and the recently launched second generation of onboard switches.