Localization-assisted Communication for RIS-aided Distributed MIMO Systems Using Distributionally Robust Optimization

Reconfigurable intelligent surface (RIS)-assisted distributed multiple-input multiple-output (D-MIMO) systems rely on instantaneous channel state information (CSI) to optimize the RIS phase shifts. However, the existence of a large number of RISs and access points (APs) requires the transmission of a significant amount of pilots for CSI estimation, thus resulting in high signaling overhead. In this paper, we propose a localization-assisted communication method that utilizes location information for RIS phase shift optimization. To address the challenge introduced by localization uncertainty, we formulate a distributionally robust optimization (DRO)-based sum-rate maximization problem and then present a successive convex approximation (SCA)-based iterative algorithm to solve it. Numerical evaluations demonstrate that the proposed DRO-based beamforming method can improve the sum-rate by $31.2 \%$ compared to the robust optimization and non-robust schemes.

I. Introduction

Distributed multiple input multiple output (D-MIMO) has received widespread attention owing to its ability to achieve collaboration among distributed antennas and provide higher spatial diversity and spectral efficiency. However, in practical deployment, obstacles present in the environment can cause obstruction between antennas and users, thereby affecting the performance of D-MIMO systems. Recently, the reconfigurable intelligent surface (RIS), a kind of metasurface is proposed to solve this problem. The RIS is comprised of numerous RIS elements that can be adjusted using PIN diodes to alter the phase of reflected signals [1]. This capability allows RISs to redirect signals toward previously obstructed areas, improving the communication performance of the communication systems. RIS-aided D-MIMO systems typically utilize instantaneous channel state information (CSI) to design RIS phase shifts. However, due to the presence of multiple access points (APs) and RISs in the system, estimating CSI requires the transmission of a significant number of pilot signals, leading to high signaling overhead [2].