Joint Blind Localization and Robust Resource Optimization for RIS-Aided Near-Field SWIPT

Abstract

This paper proposes an integrated framework for intelligent reconfigurable surfaces (RIS)-aided near-field simultaneous wireless information and power transfer systems that unifies blind user localization with robust resource allocation. Blind discovery and coarse positioning are achieved via multi-beam scanning with orthogonal Hadamard sequences, resulting in a 50% reduction in positioning error over the Zadoff-Chu benchmark. Under imperfect channel state information (CSI) arising from residual localization errors, a worst-case robust SWIPT scheme is developed to alternately optimize the base station beamforming, power-splitting ratios, and RIS phase shifts, employing semidefinite relaxation and successive convex approximation to address the non-convex coupling. Simulation results show that the proposed scheme converges within 3-5 iterations, and at low channel errors, attains a worst-case signal-to-interference-plus-noise ratio that is only 14.3% below the perfect-CSI upper bound, while consistently outperforming baseline schemes. This work provides a practical methodology for RIS deployment in the presence of unknown initial user locations.