GPS on the railway: A study of the value of incorporating knowledge of the track in a train positioning solution

The current capacity of the rail network is limited by the train positioning system employed. Trains are kept further apart than the minimum safe stopping distance because their positions are not known precisely. GPS offers a potential solution to this problem and would also allow a reduction in track-side positioning infrastructure and possibly even track-side signals, resulting in cost savings. However, the high reliability requirement for such a safety-critical application means that there are many problems that must be overcome before GPS is useable. One of the most serious obstructions to the use of GPS stems from the nature of the railway; a considerable proportion of the network runs through cuttings, which reduce the number of visible of GPS satellites. In practice, there must be a minimum of five satellites visible at all times since error detection is essential for safety. This criterion is often not met in the presence of obstructions. We do however have an additional piece of information about the position of a train – we know it is located on the track. A technique is being developed in the UCL Department of Geomatic Engineering which incorporates knowledge of the track location into the positioning solution. This should improve precision and reliability as well as allowing a position with a minimum of three visible satellites. This project analyses the effectiveness of this technique by creating an Excel spreadsheet that computes both this ‘Track Known’ solution and the conventional ‘Track Unknown’ solution. This allows both the general quality of GPS positioning in a cutting and the benefits of the Track Known solution to be assessed. Analysis of the spreadsheet reveals that location within Britain does not greatly affect the quality of position obtained, but the azimuth of the cutting creates significant variation. It is shown that high precision knowledge of the track makes little difference to the Track Known improvement, allowing the potential for good results with a low-precision and easily obtainable track database such as may be obtained from existing mapping sources. Steeper cutting sides are seen to significantly increase the benefit of the Track Known solution, but positioning still becomes unreliable as the cutting sides increase above 35°