Robust Distributed MIMO Localization based on Binary Hard Weighting

In practical scenarios, the non-line-of-sight errors or impulsive noise in dense urban areas pose great challenges to efficient and robust distributed multiple-input multiple-output radar localization. To achieve robustness against outliers, this paper proposes integrating data selection and localization into a unified framework based on the
$\ell_{0}$
-norm optimization and the idea of hard weighting. To be specific, the proposed model introduces a binary auxiliary variable to indicate the outlier-contaminated and unpolluted bistatic range measurements with 0 and 1, respectively. Considering the sparse nature of outliers, we impose sparsity constraint on the binary variable. Then, the robust localization task is cast in the form of mixed integer programming with an
$\ell_{0}$
-norm constraint. To address the resultant problem, the alternating minimization algorithm is adopted as the solver, where the target location and the binary variable are updated in an alternating manner. In particular, both subtasks have closed-form solutions with low computational complexity. Through numerical experiments, the efficiency and accuracy of the proposed algorithm is verified in comparison to several competing methods.