Signal-subspace spatial-spectrum estimator based on frequency domain model for wide-band sources

A novel approach to estimate spatial spectrum of multiple wide-band sources is proposed in this paper, which is termed signal-subspace spatial-spectrum estimator based on frequency domain model (SSEFD). Firstly, the noise-subspace frequency invariability of wide-band signals is proved by analyzing the eigenstructure of the covariance matrix of wide-band signals. The wide-band spatial spectrum is defined by the frequency invariability of noise subspace. Then a method based on Buckley´s BASS-ALE is presented to solve frequency-direction ambiguity when applying new approach to a uniform linear array. At last, the performance of SSEFD is compared to that of Wang´s CSS by computer simulation. The results show the superior of the new approach.