Removing the Outliers from DOA Estimation using Root Classification

Abstract

The situations that are unfavorable for direction of arrival (DOA) estimation methods based on structured covariance matrix of the data can be favorable for DOA estimation methods with unstructured CM and vice versa. Combined application of two types of data CM can gives the improvement of DOA estimation performance. The root classification approach which process the two sets of DOA estimates (roots of polynomials) corresponding to different CM is used in the paper. The proposed approach allows obtaining the good threshold DOA estimation performance and avoiding the saturation of root-mean square error (RMSE) at high signal-to-noise ratio. The simulation results confirming the performance of proposed technique are presented.