2-D angle estimation in beamspace featuring multidimensional multirate eigenvector processing

Zoltowski and Kautz (see IEEE Tran. on Signal Processing, May 1994 and IEEE Signal Processing Letters, Jan. 1994) developed a beamspace domain based 1D angle estimation scheme for uniform linear arrays that allows one to solve for the arrival angles of plane waves via the roots of a small order polynomial and works with any type of front end beamformer. This paper extends their approach to the more general case of 2D angle estimation with a uniform rectangular array (URA) of sensors. The approach is based on the fact that the telescoped beamspace noise eigenvectors exhibit a bandpass characteristic. Therefore, multidimensional multirate processing can be performed to ultimately yield a small dimensional signal subspace. A novel version of ESPRIT can then be applied to effect closed-form 2D angle estimation and automatically couple the two components of the source directions. Simulations demonstrating the efficacy of the approach are presented