Optimal detection and tracking of acoustical noise sources in a time-varying environment

Abstract

A dynamical detector estimator is derived for nonlinear nongaussian signal processes based on Bucy's representation theorem which is the fundamental result in stochastic filtering. A quantized version of two-dimensional conditional probability density functions is generated for bearing and bearing-rate as well as for frequency and frequency-rate. Results of the analysis of real data for bearing and bearing rate are presented and discussed using optimal nonlinear filtering techniques.