Performance Evaluation of a Reduced Order Kalman Filter for Running Train Localization Problem

A Kalman Filter based solution to running train localization problem from station to station (in short distance) has been explored. Signal system based solution aided with track side installed hardware is a classical approach, however, involves high operational cost due to installation, maintenance and protection against vandalism. Recent development is sought based on the principles of intelligent distributed control system, which operates on data supplied by both on board navigation sensors and satellite navigation system. Such a system is prone to failure due to transient nature of satellite data in adverse situations. Kalman filter based estimator constructed with only on-board sensor data has been proven to be at par. However, such a full order filter involves many states in its dynamics, which incurs high computational cost. Present work discusses the design and implementation of a reduced order Kalman filter based on on-board sensor data, to solve running train localization problem. Simulation results are presented in order to establish the effectiveness of the proposed filter.