Performance Comparison of a Multiple-Detection Probabilistic Data Association Filter with PCRLB

Most target tracking algorithms assume that at most one measurement is generated by a target in a scan. However, there are tracking problems where this assumption is not valid. For example, multiple detections from a target can arise due to multipath propagation where different signals scattered from a target arrive at the sensor via different paths. With multiple target-originated measurements, most multitarget trackers will fail or become ineffective due to the violation of the one-to-one assumption. For example, the joint probabilistic data association (JPDA) filter is capable of using multiple measurements for a single target through weighted measurement-to-track association, but its fundamental assumption is still one-to-one. In order to rectify this shortcoming, we developed a new algorithm in our previous work, the multiple-detection probabilistic data association (MD-PDA) filter, which is capable of handling multiple detections from a target in a scan, in the presence of false alarm and probability of detection less than unity. In this paper, the performance of this MD-PDA filter is compared with the posterior Cramer-Rao lower bound (PCRLB), which is explicitly derived for the multiple-detection scenario. Furthermore, experimental results show multiple-detection pattern based probabilistic data association improves the state estimation accuracy and reduces the total number of false tracks.