Capacity-Oriented RIS Control Through a Genetic Algorithm with Learning Capability

This work addresses the issue of controlling a Reconfigurable Intelligent Surface (RIS) in real-time to maximize the network capacity. A RIS control strategy is proposed exploiting a Genetic Algorithm (GA) augmented with learning capabilities to achieve high computation efficiency and guarantee consistent network performance in highly-dynamic scenarios. The proposed strategy does not require knowledge of the channel between the RIS and the users, enabling the design of RIS without channel-sensing hardware. The proposed strategy is demonstrated in a small-scale numerical example.