Blind sound channel deconvolution via artificial time reversal

Abstract

Signals that travel through a sound channel are typically distorted when they are received by a remote listener because of interference arising from multiple propagation paths. This presentation introduces a novel passive acoustic technique for blind deconvolution of broadband signals recorded on a vertical array in an unknown multipath sound channel. The technique, called artificial time reversal or ATR, is based on artificially backpropagating the signals measured at the array to their source location using a broadband Green’s function synthesized directly from the measurements. The technique exploits generic features of modal propagation in ocean sound channels to infer a phase relationship between the Green’s function and the signal. The technique allows sound-channel spread signals to be compressed to their original length. Computational examples employing a single source are presented and compared with results obtained from oceanic measurements made in May 1997 off the west coast of Italy [Hodgkiss et al., J. Acoust. Soc. Am. 105, 1597 (1999)]. The temporal correlation between an ATR-compressed signal and the original signal approaches 100% in the computational examples and reaches 90% for the oceanic data. Potential extension of this technique to multiple sources emitting simultaneously in the same frequency band is also presented. [Work supported by ONR; oceanic data provided by Dr. Song of SIO.]