Bispectral detection of nonstationary signals new theoretical results

Detecting weak active sonar returns in multipath propagation media is a central underwater signal processing problem. This paper presents an approach to active detection using bispectral analysis of the sonar data. The authors exploit properties of the bispectrum which yield a detection scheme requiring few a priori assumptions regarding the signal and noise processes. The statistical test obtained from the bispectrum outperforms other transient signal detection techniques, such as sliding window matched filter processing