An ℓ₀-Norm Optimization-Based Algorithm for Robust and Efficient MIMO Localization

Most of the existing localization frameworks are established under the Gaussian noise assumption and thus provide unsatisfactory accuracy in the presence of outliers. This work considers the robust and efficient target localization with multiple-input multiple-output radar by adopting the idea of outlier separation and the ℓ 0-norm. Specifically, we model the outliers with an auxiliary variable and impose sparsity constraint on it. The localization task is then formulated in the form of ℓ 0-norm constrained optimization. In doing so, we integrate outlier detection and target localization into a single problem. An alternating optimization (AO)-based solver is developed for the resultant optimization problem. In detail, the AO-based algorithm consists of two steps, which updates the target location and the auxiliary variable alternately. In particular, both subtasks have closed-form solutions with low-computational complexity. Numerical results on both synthetic and real data verify the efficiency and accuracy of the proposed algorithm in comparison with four competing methods.