Iterative Covariance-Aided Weighted Least Squares Localization With Optimized NLMS Interference Mitigation for Automotive MIMO Radar

Abstract

Target localization in automotive MIMO radar systems faces significant challenges in dense urban environments due to mutual interference between vehicles. While 2D-FFT based estimators are widely used, their performance degrades sharply in the presence of high-power chirp interference. This paper proposes a comprehensive localization framework for FDM-based MIMO radar to mitigate such interference and enhance estimation accuracy. First, an optimized Normalized Least Mean Squares (NLMS) adaptive filter is integrated into the fast-time preprocessing stage. Unlike conventional adaptive approaches that treat interference mitigation as an isolated task, we analytically derive the output signals to optimize the filter length, thereby maximizing the output Signal-to-Interference-plus-Noise Ratio (SINR). Second, to address the spatial correlation and masking effects in multi-target scenarios, an Iterative Covariance-Aided Weighted Least Squares (ICA-WLS) algorithm is developed. This approach iteratively reconstructs target signal models based on initial detections and subtracts them from the received signal to refine the estimation of the residual noise and interference covariance matrix. Simulation results demonstrate that the proposed framework significantly outperforms standard LS and WLS methods, effectively unmasking closely spaced targets and providing robust localization even under severe interference conditions. The proposed method bridges the gap between adaptive interference suppression and high-precision spatial localization.