Passive sonar classification using time‐frequency domain “features”.

Abstract

Success in classifying passive sonar signals depends heavily on understanding the source characteristics and the physics of propagation and scattering in the ocean environment. Passive sonar signals are typically analyzed using bearing versus time records (BTRs) and time versus frequency displays (lofargrams). Much research has been devoted to studying the relationship between features in the time‐frequency domain (i.e., lofargrams) and the range, depth, speed, and other aspects of an acoustic source. For example, striations in time‐frequency domain are known to be related to broadband excitation of resonances in the shallow water waveguide. The slope of the striations can be predicted using the so‐called waveguide invariant. Other waveguide‐dependent resonance features can be predicted with some knowledge of the environment. While studying data from a September 2007 shallow water acoustic transmission test, performed along the continental shelf off the coast of southeast Florida, we noted a “bathtub phenomenon” in lofargrams encompassing the closest point of approach of two surface ships. The relationship between these features and the characteristics of the surface ship source is the subject of this talk. [Work supported by ONR Undersea Signal Processing.]