Imaging of tonal sources on a moving platform using instantaneous frequency estimation and passive inverse synthetic aperture technique

Abstract

Monitoring of underwater radiated noise is an integral aspect of the maintenance routine of marine platforms for keeping vibrations and acoustic emissions within acceptable limits. Hence, sources of radiated noise and their origin on the platform need to be precisely localized. An algorithm is proposed for imaging of tonal sources along the length of the platform. The radiated noise of a ship is dominantly composed of several tonal frequencies corresponding to the machinery fitted at various locations along the ship's length embedded in broadband noise. The proposed algorithm determines the number of tonal sources and their individual bandwidth due to the Doppler spread. To ensure that only tonal sources present for the entire observed duration are processed, signals that do not exist for a sufficient duration are filtered out. The signal corresponding to each tonal source is band-pass filtered and its instantaneous frequency obtained. The observed IF is fitted to the expression modeling the expected non-linear variation in IF and is numerically solved for estimating the various unknown parameters: time at CPA, tonal rest frequency, speed of the platform and sensor depth corresponding to CPA. As a further check, using the estimated tonal rest frequency, a matched filter is designed for estimating time at CPA in case there happen to be multiple sources having identical tonal rest frequency. Hence imaging of all tonal sources on the platform may be carried out using estimated values of time at CPA and platform speed. The algorithm has been implemented in MATLAB and performance of the algorithm for various SNRs is presented in this paper.