1999 - 2000 - 2001 - 2002 - 2003 - 2004 - 2005 - 2006 - 2007 - 2008 - 2009 - 2010

Announcement - Agenda - Paper Presentations

As part of an agreement between the Universidad de Guadalajara and the light rail system in Guadalajara, Mexico, we are developing position monitoring system to generate displays showing the position of trains, time between trains, seconds early/late for each train, and alarms. Alarms will be generated when trains are too close, unexpected events are received, or trains do not cross subsequent sensors when expected, etc. This system will run concurrently with the existing original system which only illuminates an LED on a large plastic board to indicate the segment that each trains is in. Currently, it is up to the operators to estimate times between trains by viewing the panel where lights are illuminated. Given the number of trains and the inexactness of the information, it becomes a nearly impossible task, and thus it is currently very likely that operators in central control would miss dangerous situations. The new system will automatically calculate the time between trains based on the time of the last sensor crossed. Between sensors, the position of the train will be estimated, and then corrected every time another sensor is crossed. Actually, monitoring the spacing between trains is a performance issue as well as a safety issue and spacing problems become performance problems before they become safety problems. So, there will typically be performance based alarms long before any safety related alarms are detected. This system is expected to greatly improve performance issues in addition to the safety issues. Maintaining the spacing between trains for quality of service issues is also exactly what is needed for safety.

All hardware used for this project will be off-the-shelf. Where crossing sensors are located there will be either a commercially available RTU or single board computer running GNU/Linux connected to a TCP/IP network. These remote computers will send crossing events to central control. In central control, there will be high reliability servers running GNU/Linux as well. To start, the network connecting the remote devices to central control will be a single standard TCP/IP based network that will almost certainly not have the MTBF required for a safety critical system, but will be satisfactory for testing and development. It is expected that the current network will be replaced by two redundant loop networks to achieve the MTBF required for a safety critical system. Each loop network can be cut at any single location, and continue working, with the traffic re-routed automatically. The system as a whole can continue to function as long as one of the loop networks is working. As long as there is a very active maintenance policy in place to immediately fix any detected faults this system will deliver an MTBF at the levels required. It should be noted that all servers and remote devices will have two network connections so that they can all be connected to two networks.

This system will start out as a position monitoring system, but is designed to be upgraded to a full signaling system eventually and replace the existing signaling system. We are using various programming techniques including N-Version Programming, Internal instrumentation to detect errors, design patterns, peer review, etc, to provide the level of reliability required for safety critical systems.

In accordance with the agreement between the Universidad de Guadalajara and the light rail system, the license for this project will be GPL and it will be available on sourceforge.net.