Separately estimating the frequency-wavenumber of coherent sources by signal eigenvector smoothing technique

The authors present a novel eigenstructure-based method for spatial-temporal estimating of coherent narrowband sources impinging on a uniform passive sensor array. Based on a decomposition of 2-D covariance matrix and by using a signal-eigenvector-smoothing technique, the directions of arrival and the center frequencies of multiple sources can be separately estimated in each dimension regardless of coherence. Thus, a great amount of computational work required in searching peaks from 2-D polynomial is saved, and the aperture can be used more effectively. Simulation results are given