Reinforcement Learning-based MAC for Reconfigurable Intelligent Surface-Assisted Wireless Sensor Networks

In this short paper, a reinforcement learning based back-off mechanism is proposed for a Reconfigurable Intelligent Surface (RIS)-assisted wireless sensor network. The proposed scheme has the capability to enable the sensors to access the RIS in an interference-free manner based on the intelligently selected back-off values. One of the main features of the proposed scheme is that sensors can avoid access interference without any need of additional signaling. Simulation results demonstrate that the proposed scheme significantly achieves higher network throughput and energy efficiency compared to benchmark Binary Exponential Back-off (BEB).