Title:Risk-Based Optimization of Virtual Reality over Terahertz Reconfigurable Intelligent Surfaces

Abstract:In this paper, the problem of associating reconfigurable intelligent surfaces (RISs) to virtual reality (VR) users is studied for a wireless VR network. In particular, this problem is considered within a cellular network that employs terahertz (THz) operated RISs acting as base stations. To provide a seamless VR experience, high data rates and reliable low latency need to be continuously guaranteed. To address these challenges, a novel risk-based framework based on the entropic value-at-risk is proposed for rate optimization and reliability performance. Furthermore, a Lyapunov optimization technique is used to reformulate the problem as a linear weighted function, while ensuring that higher order statistics of the queue length are maintained under a threshold. To address this problem, given the stochastic nature of the channel, a policy-based reinforcement learning (RL) algorithm is proposed. Since the state space is extremely large, the policy is learned through a deep-RL algorithm. In particular, a recurrent neural network (RNN) RL framework is proposed to capture the dynamic channel behavior and improve the speed of conventional RL policy-search algorithms. Simulation results demonstrate that the maximal queue length resulting from the proposed approach is only within 1% of the optimal solution. The results show a high accuracy and fast convergence for the RNN with a validation accuracy of 91.92%.