Acoustic source localization by fusing distributed microphone arrays measurements

This paper addresses the problem of localizing an acoustic source by optimally fusing the observations provided by distributed microphone arrays. We consider a two-step process. First, each array provides an estimate of the source position by measuring the TDOAs (time differences of arrival) between each pair of microphones of the array. Then, these estimates are optimally fused taking into account the geometry of the arrays, in order to provide the final estimate of the source location. We study the performance of this approach by providing a closed form expression of the Cramer-Rao bound for a few simple geometries.