Sonar signal detection and classification using artificial neural networks

Abstract

Sonar signal processing is one of the main areas where artificial neural networks have made significant contributions in recent years, specifically to the task of sonar signal classification. This paper describes research that furthers that progress with the investigation of both the detection and classification of real passive sonar signals. Specifically, it examines the use of a finite impulse response neural network (FIRNN) for the continuous-mode detection and classification of real underwater transient sounds received by passive sonar. This builds on previous work where an FIRNN was applied to the pattern-mode classification of both simulated and real data sets.