Robust Near-Field Localization for RIS-Assisted SISO Systems Under Impulsive Noise

This letter investigates reconfigurable intelligent surface (RIS)-assisted near-field localization in single-input single-output (SISO) systems under impulsive noise. We propose an RIS-assisted blockwise robust squared-sine localization algorithm, termed RIS-BRSS. The squared-sine loss with adaptive majorization-minimization (MM) and truncated singular value decomposition (TSVD) are employed to robustly recover the RIS-domain signals. A blockwise median aggregation strategy is then used to construct a robust covariance matrix. By exploiting its anti-diagonal phase structure, the joint angle-range estimation problem is decoupled into one-dimensional structured subspace estimations with automatic parameter pairing, avoiding an exhaustive two-dimensional search. Simulation results demonstrate that RIS-BRSS provides improved localization accuracy and robustness under different impulsive-noise conditions and RIS training configurations.