Adaptive Extraction of Signals Using Microphone Arrays Against the Background of a Set of Intense-Interference Sources

Abstract

We consider the problem of extracting an acoustic signal of an individual source in the presence of a set of spatially separated sources when reception is performed using a microphone array. In this case, it is assumed that the extracted-signal source is located in the preliminarily known limited spatial region. Such a problem is of practical value, e.g., for speech listening in crowded areas, extracting signals from individual sources in the problems of acoustic diagnostics of various mechanical systems, etc. An adaptive spatial-processing algorithm, which is based on using the interference model in the form of a set of sources with arbitrary unknown direction vectors, is proposed to independently solve the above-mentioned problem for each narrow band of the received signal. The advantage of the proposed method is its robustness to the discrepancies between the used model and the actual conditions. Such a robustness is achieved without using a priori data on the character of the errors in the used model. We present the results of the speech-signal extraction using numerical simulation and the experimental data for the scenarios which involve the presence of high-power interference sources, as well as the errors when specifying the direction vector. The obtained results demonstrate a higher efficiency of using the proposed method under the actual conditions compared with other available techniques.