No abstract

The Probabilistic Multi-Hypothesis Tracking (PMHT) algorithm proposed by Streit and Luginbuhl in 1995 is adapted here for use in active sonar applications. PMHT is a batch technique that uses the Expectation-Maximization (EM) algorithm to obtain MAP estimates of the sequence of target states. Probabilistic Multi-Hypothesis Tracking for Active Sonar (PMHTAS) modifies PMHT for detecting and tracking a maneuvering target in clutter. Tactical active sonar systems often transmit several different waveforms simultaneously, and the detection performance with individual waveforms typically varies with the target Doppler and reverberation level. PMHTAS is a multi-waveform tracking algorithm: measurements from all waveforms are used to update each batch of track state estimates in a scheme to adapt the appropriate echo amplitude models to the target Doppler and local reverberation level. PMHTAS also incorporates developments presented by Willett, Ruan, and Streit in 1998 to utilize echo amplitude information and to handle clutter and target maneuvers. A new detection test statistic based on the objective function that is optimized in the EM algorithm has also been developed along with an initialization procedure for starting new tracks. This paper describes the key components of the PMHTAS algorithm design and presents some results using active sonar sea trial data.

No abstract

No abstract