A Novel Sensor Selection Algorithm for Multi-Target Tracking in Wireless Sensor Networks

Sensor selection is a very challenging issue where the choice of appropriate sensors should be made during the tracking processing. In order to improve the utilization of the wireless sensor networks, this paper presents an adaptive selection approach. The proposed method is dedicated to dynamically adjusting the number and the state of sensors involved. To avoid the lag of the sensor selection process, the partially observed Markov decision process (POMDP) is introduced. Since the multi-sensor is involved, we design an intuitive fusion process for the state estimation. Besides, the metric that maximizes the probability of detection is introduced to evaluate the optimal combination of sensors. A typical multi-target tracking scenario is studied in a wireless sensor networks. The results verify the effectiveness of algorithm in the selection of the sensors.