A Generalized Noise Reconstruction Framework for DOA Estimation in Received Data Domain

Abstract

<p>The performance of direction-of-arrival (DOA) estimation algorithms is affected by noise and clutter, but conventional noise reduction methods are difficult to adapt a variety of DOA estimation algorithms in different noise/clutter environments. To address this issue, a generalized robust framework for noise reconstruction in the received data domain is proposed to enhance DOA estimation performance. The problem of noise reconstruction is formulated as an optimization problem, whose objective function kernel may consist of many different DOA estimation algorithms. First, based on the cumulative sum of an approximate noise first order finite difference, a coarse estimation of the received domain noise is then calculated. In particular, the optimal initial value is obtained using the generalized pattern search algorithm, where the objective function, constructed with the distance between the normalized spatial spectrum peaks and the desired spatial spectrum peaks or the total noise power estimated by least squares (LS), is minimized. Then, exact reconstruction of the received data-domain noise is realized by a new cumulative sum formed using the solution from the previous step. Finally, using the denoised data, the DOA estimation is obtained via different algorithms. The effectiveness of the proposed framework is verified by simulations, with the objective function kernel of the propagator method (PM), Capon, sparse Bayesian learning (SBL), and parallel factorization (PARAFAC), both in terms of accuracy of DOA estimation and effectiveness in suppressing sidelobe and pseudo-peaks, for different cases, including white Gaussian noise, non-uniform noise, colored noise, impulsive noise, and K-distribution sea clutter. In addition, the effectiveness of the proposed framework is verified on the FMCW MIMO radar real-world data. The relevant codes are available online: https://github.com/CJYLostviews/Robust-Noise-Reconstruction/</p>