Sum-Rate Maximization of RIS-Aided Multi-User MIMO Systems With Statistical CSI

TL;DR

This paper develops a low-complexity method for maximizing the sum-rate in RIS-assisted multi-user MIMO systems using only statistical CSI, providing near-optimal performance with practical advantages.

Contribution

It introduces a deterministic approximation for the sum-rate under large RIS, enabling closed-form solutions for beamforming and phase shifts based on statistical CSI.

Findings

Proposed algorithm achieves comparable sum-rate to instantaneous CSI methods.

Method significantly reduces complexity and signaling overhead.

Spatial correlation effects are thoroughly analyzed.

Abstract

This paper investigates a reconfigurable intelligent surface (RIS)-aided multi-user multiple-input multiple-output (MIMO) system by considering only the statistical channel state information (CSI) at the base station (BS). We aim to maximize its sum-rate via the joint optimization of beamforming at the BS and phase shifts at the RIS. However, the multi-user MIMO transmissions and the spatial correlations make the optimization cumbersome. For tractability, a deterministic approximation is derived for the sum-rate under a large number of the reflecting elements. By adopting the approximate sum-rate for maximization, the optimal designs of the transmit beamforming and the phase shifts can be decoupled and solved in closed-forms individually. More specifically, the global optimality of the transmit beamforming can be guaranteed by using the water-filling algorithm and a sub-optimal solution of phase shifts can be obtained by using the projected gradient ascent (PGA) algorithm. By comparing to the case of the instantaneous CSI assumed at the BS, the proposed algorithm based on statistical CSI can achieve comparable performance but with much lower complexity and signaling overhead, which is more affordable and appealing for practical applications. Moreover, the impact of spatial correlation is thoroughly examined by using majorization theory.