Adaptive Coordinated Multicast for Holographic Video Streaming Over Wireless Networks: A Deep Reinforcement Learning Approach

Holographic video creates an immersive experience for users with lifelike scene reconstruction, yet it comes with the trade-off of managing massive data volumes. Therefore, to maintain a consistent quality of experience (QoE) in wireless networks with fluctuating channel conditions, it is necessary to develop efficient and adaptive holographic video streaming methods. This paper proposes a coordinated multicast streaming framework for holographic video that integrates coordinated multipoint transmission with transcoding-enabled multicasting techniques. Our framework supports the simultaneous transmission of holographic video tiles at various bitrates from multiple multi-antenna base stations to different multicast groups, significantly enhancing the overall viewing experience. We formulate an optimization problem with the goals of improving the average video quality experienced by all users while reducing the energy consumption for transcoding, which is NP-hard. By employing the proximal policy optimization, a leading-edge deep reinforcement learning algorithm, and convex optimization techniques, we develop a dynamic algorithm for joint bitrate selection and resource allocation. Simulations confirm the effectiveness of our algorithm, showing marked improvements in QoE over existing baselines.