RIS algorithms: Statistical- and Codebook-based algorithms for RIS adjustment

To fully exploit potential, RIS requires an adjustment algorithm, which determines the optimal reflection phase for each RIS Unit Cell (UC) and provides rapid and reliable data transmission. Existing algorithms that adjust RIS based on RSRP (Received Signal Reference Power) feedback typically require extensive RSRP sampling before converging to an optimal configuration, which leads to long adjustment times. This area of research is widely represented by the work of our laboratory.

We develop two major types of RIS configuration algorithms: Predefined codebook-based, Blind Reflection/Beamforming. Algorithms from the first group are widely used in the case of LOS (Line of Sight) channel conditions. Blind Reflection/Beamforming algorithms do not depend on channel conditions and rely on statistical approaches to obtain the optimal RIS configuration using the collected RSRPs.

Predefined codebook-based algorithms
We developed a fast hierarchical RIS configuration algorithm with two novel features. First, on each level of the hierarchy, the reflecting coefficients are adjusted to compensate for destructive interference of reflected and direct channels at the Mobile Station. Second, it relies on a novel approach to synthesizing 2D reflected beams with a given width and direction based on narrow beam phase stitching. The proposed algorithm demonstrates superior performance, surpassing existing solutions by up to 5 dB while requiring only 100 signal power measurements for the RIS with 4096 elements.

Blind Reflection/Beamforming algorithms
In Blind Beamforming, RIS usually switches between uniformly random configurations and does not provide a throughput gain during adjustment procedure. This problem is critical for delay-sensitive applications and high mobility scenarios, driving the need for new algorithms that can simultaneously reduce adjustment time and harvest performance gains. We developed a statistical RIS Blind Beamforming algorithm based on MCMC (Markov Chain Monte-Carlo) and SA (Simulated Annealing) approaches, which provides simultaneous data transmission during RIS adjustment in RIS-aided wireless networks. By selecting the number of readjustable UCs and tuning the effective temperature of SA, the MCMC-SA algorithm becomes reliable in noisy conditions. Under limited adjustment time and with a sufficiently large number of UCs, the proposed MCMC-SA algorithm outperforms all known statistical RIS adjustment methods in terms of SNR during the adjustment.

List of relevant papers: