Maneuver-Adaptive Multi-Hypothesis Tracking for Active Sonar Systems

In undersea surveillance, active sonar systems are commonly used to detect submarines. These sonar systems allow high detection ranges, but the interaction of sound with the sea bottom may lead to a high number of false alarms as well, especially in shallow-water environments. Therefore, automatic detection and tracking procedures are needed to provide helpful assistance to sonar operators. The MultiHypothesis Tracking approach presented in this paper is one of these procedures. It is based on nonlinear Kalman Filtering. In Kalman Filtering the assumption on underlying target dynamics is essential and has considerable impact on the overall tracking performance. As targets usually maneuver, their dynamics are varying and hidden. To include variable target dynamics, a Multi-Hypothesis tracking algorithm is adapted to consider target maneuvers by estimating and adjusting the process-noise level in the Kalman Filter equations. The level of process noise is determined for every track hypothesis individually based on the estimated velocities of the target. The impact on the tracking result is shown by applying the presented approach to different multistatic sonar datasets and comparing it to results gained by tracking with one global level of process noise. Tracking results are quantified by several tracking-performance metrics.