Deterministic Branching Gauss Particles in the Passive Sonar Tracking Problem

This work is concerned with the tracking of passive sonar targets (bearing and Doppler measurements), by maximizing the probability density in the presence of noise. We exhibit a deterministic particle filtering technique that allows high performance with a reduced number of particles. Simulation results are given, including the case of unknown manoeuvres from the target, represented by a priori Poisson control inputs