Matching Prediction to Communication and Computing for Proactive VR Video Streaming

Proactive tile-based video streaming can avoid motion-to-photon latency of wireless virtual reality (VR) by computing and delivering the predicted tiles in a segment to be requested before playback. All existing works either focus on tile prediction or on tile computing and delivering, overlooking the facts that these three tasks have to share the same duration and the quality of experience (QoE) depends on the worst performance of them. In this paper, we jointly optimize the duration of the observation window for prediction and the durations used for computing and communication to maximize the QoE of watching a VR video. We find the global optimal solution by decomposing the original problem equivalently into subproblems, with which we find prediction-limited or resource-limited region. Simulation results demonstrate the gain of the optimized durations by using two existing prediction methods with a real dataset.