Enhanced Bias-Compensated NLMS for Adaptive DOA Estimation

Adaptive techniques for estimating the direction of arrival (DOA) of signals using an array of sensors function by passing the array's captured signals through an adaptive filter. However, the input signals received by the array often contain added noise, which introduces a bias in the estimated filter weights, ultimately leading to suboptimal DOA estimation performance. In this letter, we present a computationally efficient approach for compensating this bias in adaptive DOA estimation. The bias compensation is performed by estimating the variance of the noise and subsequently eliminating the noise component from the update rule of the adaptive filter. We introduce a novel technique for approximately estimating the noise variance that relies solely on the error signal of the adaptive process. Further, we propose a method to stabilize our adaptive algorithm during tracking scenarios when the sources change their directions. Through simulations of DOA estimation using uniform linear arrays, we demonstrate that the proposed bias-compensated adaptive algorithm significantly outperforms biased adaptive algorithms and outperforms existing state-of-the-art bias-compensated algorithms, all while having a lower computational requirement.