Target Tracking in Sensor Networks: Criteria for Sensor Selection

This paper discusses the performance of a sensor selection scheme based on four expected performance criteria in a target tracking scenario, namely best expected heading, range, Doppler and bearing accuracy. The sensor selection algorithm is based on the Modified Riccati Equation, which is capable of handling sensors with probability of detection p_d les 1. The Particle Filtering technique have been used for target tracking with an adaptation for missed detections in the case of p_d les 1. Results are compared using multiple runs with simulated data for a single, moving target and a stationary two radar configuration. One sensor yields range, Doppler and bearing information, the other range and bearing information. Three cases of p_d´s are considered. The analysis includes the quality of the state estimate and the sensor selection strategy. The sensor selection strategy is evaluated for a multiple time step planning horizon.