Performance assessment of GNSS pseudorange differencing for relative train positioning

Due to the limited braking performance of trains, resilient train localization is a critical feature in railway collision avoidance systems. Currently, this function relies on heavy track infrastructures, which could be replaced by lightweight Global Navigation Satellite System (GNSS). Cooperative positioning, particularly pseudorange differencing, can meet the requirements and reliability required in railway applications, thus facilitating the penetration of GNSS technology into a typically GNSS-reluctant field. This paper aims to assess the performance of this approach in relation to potential sources of errors such as multipath, range, and velocity through simulation scenarios in a railway environment. In order to isolate the effect of those factors, a statistical study is carried out where scenarios are repeatedly simulated with only one changing factor. A horizontal method and relative velocity estimation using pseudorange differencing are tested alongside conventional single and double differencing. Performance comparisons with absolute position differencing show no significant loss of accuracy in any scenario.