Optimizing Reconfigurable Intelligent Surfaces in Multi-User Environments: A Multiport Network Theory Approach Leveraging Statistical CSI

Reconfigurable Intelligent Surfaces (RIS) are one of the emerging technologies aimed at meeting the expectations of next-generations wireless systems. In this field, the use of multiport network models for the characterization and optimization of RIS has emerged in recent years. These models take into account aspects traditionally not considered in communication theory, such as mutual coupling of RIS elements and the presence of structural scattering. In this work, we refer to this model and focus on the problem of maximizing the average achievable rate in a multi-user uplink scenario by leveraging statistical Channel State Information (CSI). This approach significantly
reduces the computational burden and communication overhead in CSI estimation compared to schemes requiring instantaneous CSI estimation. These benefits are achieved with performance that, in many cases, is reasonably close to that of the perfect CSI scenario. This is one of the outcomes achievable with the proposed optimization scheme. Moreover, it is shown how in multi-user scenarios, namely in the presence of interference, the use of inadequate models to characterize RIS can lead to very poor performances. For example, models that do not consider structural scattering may fail to account for interference caused by RIS.