Extension of COMET algorithm to multiple diffuse source localization in azimuth and elevation

This paper presents an extension of COvariance Matching EsTimator (COMET) to tridimensional multiple diffuse source localization in azimuth and elevation. Most of works on diffuse source localization deal with only azimuth estimation. However, in realistic situations, a tridimensional source localization is needed which is the main objective of this paper. We show that the dimensionality of the COMET algorithm can be reduced by separating the estimation of the different source powers and the noise variance from that of the remaining parameters, namely the Angle Of Arrival (AOA) and the corresponding angular spread. As COMET still involves multi-dimensional nonlinear optimization, we choose, in this purpose, the alternating projection algorithm to alleviate the corresponding complexity. MUSIC algorithm is processed to initialize the so-resulted algorithm. Simulations of the proposed algorithm are carried in different contexts and compared to the theoretical study, the Cramer-Rao Bound (CRB), MUltiple SIgnal Classification algorithm (MUSIC) and DISpersed Signal PARametric Estimation (DISPARE) simulation results. Copyright (c) 2004 AEIT.