Joint Reliability Optimization and Beamforming Design for STAR-RIS-Aided Multi-user MISO URLLC systems

Simultaneous transmitting and reflecting reconfigurable intelligent surfaces (STAR-RISs) are capable of serving users on both sides of it at the same time through active and intelligent control of space electromagnetic waves, and are therefore considered to be a powerful means to facilitate the design of ultra-reliable low-latency communication (URLLC) systems. In this paper, we investigate the joint reliability optimization and beamforming design problem for a STAR-RIS-assisted multi-user multiple-input single-output (MISO) URLLC system in an industrial IoT scenario. A system sum-rate maximization problem is formulated, subject to the STAR-RIS amplitude and phase shift constraints, power and reliability constraints. To solve this problem, we design a joint optimization algorithm based on deep reinforcement learning. The algorithm determines the optimal access point transmit precoding matrix, STAR-RIS reflection- and transmission-coefficient matrices, and the packet error probabilities for actuators based on the channel state information (CSI). On this account, the proposed algorithm dynamically tunes the STAR-RIS to make the optimal beam response for real-time channel changes. Comprehensive simulation results demonstrate that the proposed algorithm can provide substantial performance benefits over several baseline schemes. Moreover, the actual channel model with channel estimation error is also considered for reliability to evaluate the impact of imperfect CSI on system performance.