Physics-Informed Metaheuristics for Fast RIS Codebook Compilation

Reconfigurable intelligent surfaces (RIS) can actively manipulate the wave propagation within a space, even in unconventional ways, enabling software-defined wireless propagation. Enabling real-time RIS operation requires a codebook, that is, a set of pre-calculated states of the RIS constituent elements that yield any of the supported macroscopic RIS functionalities, such as anomalous steering, splitting, and wave absorption. The codebook compilation process occurs offline, and requires the time consuming optimization of either an EM simulator, or an automated RIS measurement system. The process is especially resource-demanding in the case of RIS designs with no known analytical performance model, such as metasurfaces in the general sense. This article studies the synergy between metaheuristic optimizers and the RIS codebook compilation process. Specifically, well-known and widely efficient metaheuristic optimizers are imbued with attributes of the RIS physics, yielding considerable gains in the codebook compilation time. This process leverages correlations between microscopic configurations and macroscopic RIS responses, geometric optics, and EM simulators. The evaluation outcomes indicate a significant potential in the design of high-performing and resource-effective metaheuristics for RIS.