Resolution of coherent signals using random array

A high-resolution algorithm for resolving coherent plane waves using random arrays is presented. The algorithm is developed based on the signal subspace of the array element covariance matrix. Thus, the rank of the signal subspace must be predetermined. Commonly used methods on rank determination are based on the multiplicity of the smallest eigenvalues, and information criteria proposed by M. Wax and T. Kailath (1984). These methods work well in situations where the ambient noise is white or the noise is weakly correlated among pairs of sensors. A method for signal subspace rank determination is also proposed. Simulation results are presented to illustrate the performance of the proposed algorithm