Digital Twins for Generic Radio Environments Parametrized by Reconfigurable Intelligent Surfaces: Physics-Based vs. Physics-Agnostic Surrogate Models
Résumé
Optimizing the configuration of reconfigurable intelligent surfaces (RISs) in generic (potentially complex-scattering) radio environments for a desired communications or sensing functionality is challenging because of the non-linear manner in which the RIS impacts the wireless channel. The availability of a learned surrogate forward model of the mapping from RIS configuration to wireless channel can substantially facilitate the optimization problem. Here, we explore different approaches (physics-based vs. physics-agnostic) to learning such digital twins. © 2023, META Conference. All rights reserved.