Model based target depth estimation using received signal statistics.

Abstract

The overall goal of our research is to develop model based sonar signal processing techniques that utilize predictions of the acoustic field while being robust to environmental variability and uncertainty. Specifically, we are estimating the depth of a target broadcasting a tonal signal in shallow water using a single passive hydrophone. Three different techniques, naive Bayes, histogram filter, and estimator correlator, are examined. Each is applied to event S5 of the SWellEX‐96 measurement that took place in shallow water off the coast of California. The data consist of sinusoidal signals transmitted simultaneously from moving sources at two different depths and received at a bottomed horizontal array. Received signal amplitudes are modeled using random access memory for the two sources at different depths, and these modeling results are used in the design of the target depth estimation algorithms. Thus, the ability of each of the techniques to estimate target depth depends on an accurate representation...