Nonlocal Reconfigurable Intelligent Surfaces for Wireless Communication: Modeling and Physical Layer Aspects
Amine Mezghani, Faouzi Bellili, Ekram Hossain
TL;DR
This work develops a physically-consistent, linear scattering framework for RIS and advocates nonlocal redirective RIS (RedRIS) as a scalable, angle-selective alternative to conventional local RIS. By leveraging a lens-based analog DFT transformation, RedRIS decouples ports, enabling efficient multi-beam routing, attenuation of control/estimation overhead, and potential integration of fronthaul and access with directionality-enhanced security. The paper analyzes size/bandwidth constraints, introduces active/wave-amplification concepts within RedRIS, and discusses practical architectures (fractionated/lensed/tiled designs) to address scalability. It also extends RedRIS to retrodirective CSI, beam-splitting, semi-static configurations, and security-enhanced deployments, offering a comprehensive roadmap for mmWave/THz networks. While promising, the approach faces implementation challenges in RF interconnections, switching losses, and channel estimation that necessitate continued design, optimization, and experimental validation.
Abstract
Conventional Reconfigurable intelligent surfaces (RIS) for wireless communications have a local position-dependent (phase-gradient) scattering response on the surface. We consider more general RIS structures, called nonlocal (or redirective) RIS, that are capable of selectively manipulate the impinging waves depending on the incident angle. Redirective RIS have nonlocal wavefront-selective scattering behavior and can be implemented using multilayer arrays such as metalenses. We demonstrate that this more sophisticated type of surfaces has several advantages such as: lower overhead through coodebook-based reconfigurability, decoupled wave manipulations, and higher efficiency in multiuser scenarios via multifunctional operation. Additionally, redirective RIS architectures greatly benefit form the directional nature of wave propagation at high frequencies and can support integrated fronthaul and access (IFA) networks most efficiently. We also discuss the scalability and compactness issues and propose efficient nonlocal RIS architectures such as fractionated lens-based RIS and mirror-backed phase-masks structures that do not require additional control complexity and overhead while still offering better performance than conventional local RIS.
