Super-resolutive technique for correlated sources identification

A new algorithm specifically developed for the localization of totally correlated sources is proposed. This algorithm uses a multidimensional search with many steering vectors, in a space dimension equal to the number of sources. Many computed propagation matrices aim at the verification of quadratic error between two vectors resulting from the partition of the composite wavefront. This wavefront is described by the cross-spectral matrix eigenvector that spans the signal subspace. An extension for the case of many groups of fully correlated sources is suggested.