Enhancing Holographic Video Streaming: An Integrated Approach to Adaptive Transmission in Wireless Networks

Holographic video offers an immersive six-degree-of-freedom viewing experience but poses significant challenges in managing massive data volumes, especially in wireless networks with varying channel conditions. To ensure consistent quality of experience, we propose an efficient and adaptive streaming framework for holographic video, which seamlessly integrates coordinated multipoint (CoMP), multicasting, transcoding, and adaptive transmission techniques. CoMP improves transmission quality by coordinating multiple base stations, multicasting enables spectrum-efficient delivery to multiple users but enforces a single bitrate per multicast group, potentially conflicting with the goal of personalized adaptive streaming. To reconcile it, transcoding is introduced at the user side: a unified high-bitrate stream is transmitted using CoMP and multicast, and each user then locally transcodes the received video into lower-bitrate versions. In addition, adaptive transmission techniques divide holographic video into tiles and dynamically allocate resources based on network conditions. We formulate a joint optimization problem for video bitrate selection and communication resource allocation to maximize overall video quality for all users. This problem is a non-convex mixed-integer nonlinear programming problem and is NP-hard. To solve it, we develop two algorithms: one based on block coordinate descent (BCD) and another leveraging deep reinforcement learning (DRL). The BCD-based algorithm relaxes integer variables, solves a biconvex problem, and applies a rounding process. The DRL-based algorithm employs Lagrangian analysis to reduce the action space dimensionality, enhancing learning efficiency. Extensive simulations show that our proposed algorithms improve video quality by up to 60% compared to existing baselines, with the DRL-based method offering a better trade-off between solution quality and computational efficiency for real-world applications.