Tracking Algorithms for Bistatic Sonar Systems

An active sonar system consists of a sound source activating a surveillance area and a receiver listening for echoes reflected from targets. In a bistatic setup, the source and the receiver are not collocated. The resulting measurement model is described by a non-linear function of the target state, the given bistatic geometry and environmental parameters. For target tracking it is mandatory to have an adequate description of the uncertainty on the target state that is resulting from the uncertainties in the measurements and environmental parameters. An inadequate description would decrease the performance of the fusion process. To cope with the nonlinearity in the measurement model, four different methods can easily be implemented based on the Kalman Filter framework: linear transformation of each measurement in Cartesian coordinates with tracking in the Cartesian system, unscented transformation of each measurement in Cartesian coordinates with tracking in the Cartesian system, extended Kalman filtering and unscented Kalman filtering. In this paper, we compare these methods by simulating their performance in an operationally relevant setup.