CSI-Free Optimization of Reconfigurable Intelligent Surfaces with Interference by Using Multiport Network Theory

Abstract

Reconfigurable Intelligent Surfaces (RIS) will play a pivotal role in next-generation wireless systems. Despite efforts to minimize pilot overhead associated with channel estimation, the necessity of configuring the RIS multiple times before obtaining reliable Channel State Information (CSI) may significantly diminish their benefits. Therefore, we propose a CSI-free approach that explores the feasibility of optimizing the RIS for the uplink of a communication system in the presence of interfering users without relying on CSI estimation but leveraging solely some a priori statistical knowledge of the channel. In this context, we consider a multiport network model that accounts for several aspects overlooked by traditional RIS models used in Communication Theory, such as mutual coupling among scattering elements and the presence of structural scattering. The proposed approach targets the maximization of the average achievable rate and is shown to achieve performance that, in some cases, can be very close to the case where the RIS is optimized leveraging perfect CSI.