Direction of Arrival Estimation of Acoustic Echoes Using Source Elimination Method

Abstract

A new method for the direction of arrival (DOA) estimation for a single narrowband acoustic source with multiple low power echoes is proposed. The mathematical relations between the signal covariance matrix, the sources steering vectors, and the power of the sources are developed in order to expose the contribution of each source. Algorithms to find each of the source's DOA and their respective power are presented. The source elimination method (SEM), based on the elimination of the contribution of each source to improve the DOA estimation, is developed. Monte Carlo simulations are presented, showing that SEM yields more accurate results than multiple signal classification (MUSIC) with forward-backward spatial smoothing (FBSS) to find the echoes' DOA with an echo-to-noise ratio between -13 and -17 dB. Experimental results show that for a small array and two sources with different power, SEM outperforms MUSIC with FBSS.