Generalized phase-smoothing for DOA estimation by using vector-sensor array

The problem of the noncircularity exploitation is addressed for the improvement of acoustic vector-array based direction of arrival (DOA) estimation. The generalized phase-smoothing technique (GPS) is proposed to enhance the robustness of the subspace estimate to the perturbations in the array covariance matrix, and to upgrade the multiple signal classification method (MUSIC) for DOA estimation. This smoothing scheme applies to any type of acoustic vector-array with center-symmetric spatial geometry, does rlot require coupled subarrays, and decorrelates two coherent signals. Simulation results show that GPS preprocessing can significantly improve the precision of MUSIC based DOA estimation, in particular for difficult scenarios (low signal to noise ratio, short data length).