Learning-based Strategy for RIS-Assisted Terahertz Virtual Reality Networks

Abstract

The quality of experience (QoE) requirement of wireless virtual reality (VR) can only be satisfied with high data rate, high reliability, and low VR interaction latency. This high data rate over short transmission distances may be achieved via the abundant spectrum in the terahertz (THz) band. However, THz waves suffer from severe signal attenuation, which may be compensated by the reconfigurable intelligent surface (RIS) technology with adjustable phase-shift of each reflecting element. Motivated by these considerations, in this paper, we propose an RIS-assisted THz VR network in an indoor scenario, taking into account the viewpoint prediction and downlink transmission. We first propose a genie-aided online gated recurrent unit (GRU) and integration of online long-short term memory (LSTM) and convolutional neural network (CNN) algorithm to predict the viewpoint, location, and the line-of-sight (LoS) and non-line-of-sight (NLoS) statuses of the VR users over time, with the aim to optimize the long-term QoE of the VR users. We then develop a constrained deep reinforcement learning algorithm to select the optimal phase shifts of the RIS for the downlink transmission under latency constraints. Simulation results show that the proposed ensemble learning architecture achieves near-optimal QoE as that of an exhaustive algorithm, and about two times improvement in QoE compared to the random phase shift selection scheme.