Localization of the sources using blind identification of signal subspace
Bendjama Ammar, Salah Bourennane · The Journal of the Acoustical Society of America · 1999
To find the direction of arrival (DOA) of N sources impinging on an array of M sensors, actual fourth-order direction-finding methods try to solve an N-dimensional problem from the statistics of the data. Current narrow-band array processing techniques are based on the second-order statistics of the received signals. In many situations, the received signals are non-Gaussian, so that they contain valuable statistical information in their moments of order greater than two. In these circumstances, it makes sense to develop array processing techniques which are designed to use their higher-order information. Of particular interest are the algorithms based on higher-order cumulants, which are asymptotically insensitive to Gaussian noise: it is thus neither necessary to know, to model, or to estimate the noise covariance, which is a reasonable assumption in many practical situations. In this paper, an approach of direction finding is proposed, based on the first step of blind identification of source steering. As a second step, the direction-finding concept is now possible due to the development, these last years, of several blind source separation methods which allow the blind identification of source-steering vectors.