Improved estimation of machinery tonal intensity under severe underwater channel distortions

Acoustic signals undergo random fluctuations due to time and frequency dispersive nature of the underwater channel. The principal contribution of this paper is a cepstrum-based approach for channel equalization to derive machinery tonals under severe underwater channel distortions based on homomorphic transformation. This approach takes advantage of the fact that the underwater channel occupies higher cepstral indices, while the signal at source remains at lower cepstral indices corresponding to the fundamental period of the machinery tonal. Performance evaluation has been presented to support the improvement in estimation of tonals, based on the variance of the machinery tonals.