A scheduling framework for performing resource slicing with guarantees in 6G RIS-enabled smart radio environments

Smart Radio Environments (SRE) transform the wireless propagation phenomenon in a programmable process. Leveraging multiple Reconfigurable Intelligent Surfaces (RIS), the wireless waves emitted by a device can be almost freely routed and manipulated, reaching their end destination via improbable paths, with minimized fading and path losses. This work begins with the observation that each such wireless communication customization occupies a certain number of RIS units, e.g., to form a wireless path with consecutive customized reflections. Therefore, SREs can be modeled as a resource of constrained capacity, which needs to be sliced among interested clients. This work provides a foundational model of SRE-as-a-resource, defining Service Level Agreements (SLAs) and Service Level Objectives (SLOs) for the SRE client requests. Employing this model, we study a class of negative drift dynamic weighted round robin policies, that is able to guarantee specific SRE resource shares to competing user requests. We provide a general mathematical framework where the class of policies map ping user requests to resources does not require statistical knowledge regarding the arrival distribution or the duration of each user communication. We study the meaning of work conserving and non-work conserving modes of SRE operation, and also study the convergence properties of our scheduling framework for both cases. Finally, we perform the feasibility space analysis for our framework and we validate our analysis through extensive simulations.