Bayesian broadband passive sonar tracking

We present a Bayesian tracker for broadband passive sonar. The Bayesian formulation in Cartesian coordinates facilitates multi-sensor fusion of a distributed field of receivers. The likelihood functions used in the tracker are formulated so as to leverage a priori knowledge about the signal-of-interest´s bandwidth and spectral mask when available. The use of a priori knowledge about the signal enables substantial gains in the tracker´s detection and localization performance, even when this knowledge is approximate.