An STM (Specific Transmission Module) enables a train equipped with a standard European Train Control System to communicate and operate within national train control systems; in the UK, these are AWS (Automatic Warning System) and TPWS (Train Protection and Warning System). The wayside equipment can be easily be seen between the tracks as boxes (AWS) or grids (TPWS).
In the event that the STM considers it unwise for the train to continue, possibly as a result of the train being in an unsafe position or travelling at an unsafe speed, it causes the application of the emergency brake. Decisions are made very quickly; a fast passenger train can be travelling at more than 50 meters per second, so microseconds can make a life-or-death difference to hundreds of people.
ASE’s Client, Mors Smitt UK Ltd, engaged ASE to design and develop the control unit for a UK STM to operate with UNISIG’s defined interface to Baseline 2 of Subset 35. The unit had a targeted reliability figure for a general MTBF of over 220,000 hours of operation, and had to be certified to SIL-1 under EN 50129, with the equivalent software complying with the same safety level under EN 50128.
The STM monitors an RF antenna and a Hall effect sensor device on the train, which detect transmissions from the wayside equipment. The STM’s interpretation of these inputs results in decisions with regard to whether or not the train is safe to continue on its current section of track.
The hardware selected by ASE was based upon Altera’s SoC Cyclone V chipset, which provided a dual-processor ARM solution embedded within an FPGA as a single device. The hardware choice enabled ASE to develop a single main processor board and a generic backplane, thus facilitating different interface modules for different projects; in this case, the client was looking for a track signalling system, but ASE was aware that the client also had a potential need for a power monitoring solution. ASE’s design provides the client with generic interfaces to a number of communication standards and a generic upgrade feature that can reprogram both the Linux applications and the FPGA program in order to support the different I/O board types.
The system required EMC protection, with transient burst protection providing particular and unusual challenges, as the STM must be able to identify all of the real RF inputs, which could be masked by or mistaken for transient bursts.
Initially, the train’s ETCS unit was not available, even in prototype form, meaning that it was not possible to confirm the correct operation of the STM against a real unit during development. It therefore became necessary for ASE to implement most of the ETCS side of the protocol in ASE’s own test rigs, essentially emulating the ETCS, in order to confirm correct operation of the unit ahead of the real ETCS becoming available. ASE is extremely proud of the resultant test rig, which is able to simulate all inputs and outputs (including Ethernet traffic) to the unit, and is used in extensive automated testing of the STM.
The software was implemented on a Linux kernel version 3.10, for which ASE has successfully put forward a suitability case for the use of that kernel on SIL-1 rated implementations. ASE needed to write a number of Linux drivers for the bespoke hardware solution.
The end-client’s development of the ETCS (not an ASE item), with which the STM unit had to communicate, took longer than expected, further demonstrating the benefit of ASE’s test harness. As a result of the ETCS delay, the STM project was modified to communicate directly with the train units, essentially performing some of the tasks of the ETCS, such as creating driver displays using BT’s IPTCOM protocol in order to provide a temporary signalling solution and get the trains certified for passenger use.
The initial project took approximately two years from concept to completion, but its development is ongoing. We are continually advancing the scientific understanding of the interaction between the sensors on the train and the lineside equipment. We use advanced mathematical modelling to continually improve our algorithms and test equipment; the initial work has resulted in a complete and successful system, but our research and development activities are still very active!