Impact of range dependent propagation on classification of underwater objects by their sonar backscatter

Abstract

Propagation effects, such as dispersion, absorption and multi-path, can adversely impact classification of underwater objects from their sonar backscatter. One approach to handling this problem is to extract features from the wave that are minimally affected by propagation effects, if possible. In previous work, a signal processing and feature extraction method was developed to obtain moment-like features that are invariant to dispersion and absorption. The method was developed based on linear wave propagation in range- independent environments. However, most ocean environments, especially littoral environments, exhibit range dependence. Deriving propagation invariant features for such environments remains an especially challenging task. In this paper, we explore the classification utility of the previously developed range-independent features in a range-dependent environment, via simulation of the propagation of the backscatter from two different cylinders in an ideal wedge. Our simulation results show that, while performance does drop off for increasing distances in a range dependent environment, the previously developed invariant moment features do provide better classification performance than ordinary temporal moments.