Rate-Splitting for Intelligent Reflecting Surface-Aided Multiuser VR Streaming

The growing demand for virtual reality (VR) applications requires wireless systems to provide a high transmission rate support 360-degree video streaming multiple users simultaneously. In this paper, we propose an intelligent reflecting surface (IRS)-aided rate-splitting (RS) VR system. the proposed system, RS facilitates exploitation of shared interests in streaming, and IRS creates additional propagation channels high-resolution videos. also enhances capability mitigate performance bottleneck caused by requirement that all have be able decode common message. We formulate optimization problem maximization achievable bitrate subject quality-of-service (QoS) constraints users. deep deterministic policy gradient with imitation learning (Deep-GRAIL) algorithm, which leverage reinforcement (DRL) hidden convexity formulated optimize phase shifts, parameters, beamforming vectors, selection tiles. RavNet, is neural network customized our Deep-GRAIL algorithm. Performance evaluation based on real-world dataset shows IRS-aided system outperforms several baseline schemes terms sum-rate, videos, online execution runtime. Our results reveal respective gains obtained from improving QoS multiuser systems.