Adaptive RIS Control for Mobile mmWave NLoS Communication Using Single-Bit Feedback

TL;DR

This work proposes and experimentally evaluates an adaptive beamforming algorithm that enables RIS reconfiguration via a low-rate feedback link from the mobile user equipment to the RIS controller, operating without requiring UE position knowledge.

Abstract

Reconfigurable intelligent surfaces (RISs) are emerging as key enablers of reliable industrial automation in the millimeter-wave (mmWave) band, particularly in environments with frequent line-of-sight (LoS) blockage. While prior works have largely focused on theoretical aspects, real-time validation under user mobility remains underexplored. In this work, we propose and experimentally evaluate an adaptive beamforming algorithm that enables RIS reconfiguration via a low-rate feedback link from the mobile user equipment (UE) to the RIS controller, operating without requiring UE position knowledge. The algorithm maintains the received signal power above a predefined threshold using only a single-bit comparison of received power levels. To analyze the algorithms performance, we establish a simulation-based Monte Carlo (MC) optimization benchmark that assumes full UE position knowledge, accounts for practical hardware constraints, and serves as an upper bound for performance evaluation. Using a hexagonal RIS with 127 elements and 1-bit phase quantization at 23.8 GHz, we validate the proposed approach in a semi-anechoic environment over a 60 cm by 92 cm area. The results demonstrate that the single-bit feedback-driven algorithm closes much of the performance gap to the MC upper bound while achieving up to 24 dB gain in received power compared to an inactive RIS baseline. These findings highlight the practical potential of feedback-based adaptive RIS control for robust mmWave non-line-of-sight (NLoS) communication with mobile users.

Figures (8)

• Figure 1: RIS coordinate system for a hexagonal RIS element placement in the yz-plane. The BS horn antenna radiates from position ${\hbox{\boldmath$a$}}$ towards the center of the RIS at ${\hbox{\boldmath$0$}}=(0,0,0)$ over a distance of $|{\hbox{\boldmath$a$}}|$, similarly the UE omni-directional antenna at position ${\hbox{\boldmath$b$}}$ is within a distance of $|{\hbox{\boldmath$b$}}|$. The LOS is blocked between BS and UE. The picture is not to scale to improve clarity.
• Figure 2: RIS element grouping structure for the proposed iterative algorithm. Blue and green boundary lines correspond to the subgroups of $\mathcal{G}^1$ and $\mathcal{G}^2$, respectively. For element indices, please refer to Tables \ref{['tab:group1']} and \ref{['tab:group2']}.
• Figure 3: Block diagram of the measurement setup used to evaluate the RIS-aided iterative beamforming algorithm. The feedback link between the UE and the RIS controller forms a closed-loop control system involving the UE, the RIS controller, and the RIS. The UE moves inside the area of interest (AoI), which is in NLoS of BS.
• Figure 4: Photograph of the RIS measurement testbed (LoS absorber removed for clarity). The horn antenna serves as the transmitter, and the UE moves on an $xy$-positioning table equipped with an omnidirectional monopole antenna.
• Figure 5: Simulated received power distribution over the area of interest for two RIS control strategies: (i) Monte Carlo (MC) optimization with full UE position knowledge (upper-bound), and (ii) proposed iterative single-bit feedback algorithm without position knowledge. This simulation provides a benchmark for assessing the performance gap between the ideal MC optimization and the proposed practical feedback-driven approach.