Experimental Evaluation of Multiple Active RISs for 5G MIMO Commercial Networks

TL;DR

The paper addresses the challenge of enhancing high-frequency 5G MIMO performance by deploying multiple active RISs. It introduces a low-complexity, codebook-based, sequential beamforming algorithm tailored to multi-RIS configurations, and validates it through both numerical simulations and commercial field tests. The experimental results show up to a 14% improvement in channel rank and throughput over single-RIS deployments, while maintaining low computational overhead and compatibility with commercial UEs. This work demonstrates the practical viability of active multi-RIS systems for current 5G networks and future 6G scenarios, with potential extensions to mobility and multi-user support. The study also formalizes the channel model, performance metrics, and energy efficiency considerations that underpin multi-RIS optimization, including the capacity expression $C = \log_2 \det(\mathbf{I} + \widetilde{\mathbf{H}}\widetilde{\mathbf{H}}^{H})$ and the rank-related metric $\varepsilon = \sigma_{\min}/\sigma_{\max}$.

Abstract

While numerous experimental studies have demonstrated the feasibility of reconfigurable intelligent surface (RIS) technology, most have primarily focused on extending coverage. In contrast, this paper presents an experimental evaluation of multiple active RISs deployed in a 5G multiple-input multiple-output (MIMO) commercial network, emphasizing enhancements in channel rank and throughput. We propose a low-complexity, codebook-based beamforming algorithm specifically tailored for multi-RIS configurations, which diversifies directional channels and reduces reliance on explicit channel state information. Field tests using a commercial base station and user equipment reveal that the multi-RIS system can improve channel rank and throughput by up to 14% compared to single-RIS deployments, while maintaining low computational complexity. These findings underscore the practical benefits of active multi-RIS systems for next-generation networks.

Figures (5)

• Figure 1: Simulation environment.
• Figure 2: Channel capacity heatmaps for a single RIS at RIS1 with (a) 1-bit, (b) 2-bit, and (c) 3-bit PSs.
• Figure 3: Comparison of EE and channel rank for single-RIS and multi-RIS systems.
• Figure 4: Iteration count versus capacity: Comparison among RMS, SCSM MRISblind-2023, BG Chian-2024, and the proposed codebook-based algorithm.
• Figure 5: Commercial testing field: (a) Aerial photograph of an outdoor BS and an indoor office; (b) Indoor testing room plan; (c) Field testing scenario; (d1--d3) Three different measurement scenarios.