Distortion-Resistant Dynamic Bitrate Adaptive Transmission for 360-Degree Panoramic Video
Siqi Zhang, Yuang Chen, Hancheng Lu, Cheng Guo, Zhengze Li · 2024
For 360-degree panoramic videos, the sharp increase in the amount of video data has made its transmission suffer from tremendous bandwidth requirements. On the other hand, due to the existence of the user's field of view (i.e. viewport), only a limited part of video can be watched, which leads to the bandwidth wastage of video streaming outside the viewport. Selecting different bitrates for tiles based on the relationship between the tile positions and the viewport can counteract this problem. According to this fact, we propose a distortion-resistant dynamic bitrate adaptive transmission algorithm (DRDT) for 360-degree panoramic video. By cutting 360-degree panoramic video into temporal chunks and spatial tiles, DRDT comprehensively considers multiple influencing factors including viewport and spatial distortion to utilize limited bandwidth resources. It selects reference bitrates for tiles in and outside the viewport respectively by reinforcement learning, then optimizes the bitrate allocation for the tiles within the viewport by game theory. We use the real-world traces to test the performance of DRDT. The simulation results demonstrate that the proposed solution can improve user's QoE effectively compared with UVP and 360Pensieve.