Model-free Optimization and Experimental Validation of RIS-assisted Wireless Communications under Rich Multipath Fading

TL;DR

This work tackles RIS-aided wireless optimization under rich multipath by proposing a CSI-free, model-free cross-entropy (CE) algorithm to maximize the received SNR with binary RIS phase shifts. The method samples configurations, evaluates performance using only $\xi(\bm{x})$, and iteratively updates the sampling distribution to converge toward high-SNR RIS settings, achieving ${\mathcal O}(T K N)$ time complexity. Through both simulations and reverberation-chamber experiments, the CE approach demonstrates substantial SNR gains (approximately $16\ \mathrm{dB}$ in simulation and $12\ \mathrm{dB}$ in experiment) and stronger channel hardening than benchmark methods, with results closely following the $N^2$ scaling. The study provides practical validation of channel hardening in RIS-enabled systems and shows CE's potential for CSI-free, scalable RIS optimization in real indoor environments.

Abstract

Reconfigurable intelligent surface (RIS) devices have emerged as an effective way to control the propagation channels for enhancing the end-users' performance. However, RIS optimization involves configuring the radio frequency response of a large number of radiating elements, which is challenging in real-world applications due to high computational complexity. In this paper, a model-free cross-entropy (CE) algorithm is proposed to optimize the binary RIS configuration for improving the signal-to-noise ratio (SNR) at the receiver. One key advantage of the proposed method is that it only requires system performance indicators, e.g., the received SNR, without the need for channel models or channel state information. Both simulations and experiments are conducted to evaluate the performance of the proposed CE algorithm. This study provides an experimental demonstration of the channel hardening effect in a multi-antenna RIS-assisted wireless system under rich multipath fading.

Figures (4)

• Figure 1: An illustration of the RIS-assisted communication system with $N = 152$ elements. The RIS elements are dipoles which cause omnidirectional scattering, and the metallic cavity whose symmetry is broken by the stirrer supports rich multipath fading. With a strong line of sight (LOS), the fading channel gain is expected to be Ricean distribution.
• Figure 2: Performance on different numbers of iterations.
• Figure 3: Performance on channel hardening.
• Figure 4: Performance on channel hardening (variance$/$mean$^2$).