Beamspace DOA estimation featuring multirate eigenvector processing

A novel approach to angle of arrival estimation in beamspace for a uniform linear array is presented. Beamspace noise eigenvectors may be telescoped to vectors in the element-space noise subspace. The telescoped noise eigenvectors are bandpass, facilitating multirate processing involving modulation to baseband, filtering, and decimation. As these operations are linear, a matrix transformation applied to the eigenvectors may be constructed a priori. Incorporation of the technique into either the Root-MUSIC or ESPRIT prescriptions provides a computationally efficient procedure. Compared to past efforts to adapt Root-MUSIC and ESPRIT to beamspace, this approach circumvents the need for restrictive requirements on the form of the beamforming transformation. An asymptotic theoretical performance analysis is also included to provide an alternative to computationally intensive Monte Carlo simulations. Simulation studies show the validity of the performance predictive expressions and verify that the procedure, when incorporated into the Root-MUSIC/ESPRIT formulations, produces a direction finding technique that nearly attains the Cramer-Rao bound (CRB).