In this paper, we present a novel content caching and delivery approach for mobile virtual reality (VR) video streaming. The proposed approach aims to maximize VR video streaming performance, i.e., minimizing video frame missing rate, by proactively caching popular VR video chunks and adaptively scheduling computing resources at an edge server based on user and network dynamics. First, we design a scalable content placement scheme for deciding which video chunks to cache at the edge server based on tradeoffs between computing and caching resource consumption. Second, we propose a machine learning-assisted VR video delivery scheme, which allocates computing resources at the edge server to satisfy video delivery requests from multiple VR headsets. A Whittle index-based method is adopted to reduce the video frame missing rate by identifying network and user dynamics with low signaling overhead. Simulation results demonstrate that the proposed approach can significantly improve VR video streaming performance over conventional caching and computing resource scheduling strategies.

翻译：本文提出了一种面向移动虚拟现实（VR）视频流的新型内容缓存与分发方法。该方法旨在通过基于用户与网络动态性，在边缘服务器上主动缓存热门VR视频块并自适应调度计算资源，以最大化VR视频流传输性能（即最小化视频帧丢失率）。首先，我们设计了一种可扩展的内容放置方案，依据计算与缓存资源消耗之间的权衡，决定在边缘服务器上缓存哪些视频块。其次，我们提出了一种机器学习辅助的VR视频分发方案，该方案通过分配边缘服务器的计算资源，以满足来自多个VR头戴显示设备的视频分发请求。采用基于惠特尔指数的方法，以低信令开销识别网络与用户动态性，从而降低视频帧丢失率。仿真结果表明，与传统缓存及计算资源调度策略相比，所提方法能显著提升VR视频流传输性能。