Improving the threshold performance of the estimator bank direction finding techniques using outlier identification and cure

The estimator bank approach to direction finding employs a number of parallel randomly weighted MUSIC direction-of-arrival (DOA) estimators to improve the threshold performance of the final DOA estimate. In this paper, a new technique is proposed to select non-outlying estimates among such available parallel DOA estimates using outlier identification and cure (I & C). Simulation results validate an improved threshold performance of the proposed approach relative to several earlier estimator bank techniques.