A Combined Field-of-View Prediction-Assisted Viewport Adaptive Delivery Scheme for 360° Videos

Recently, 360° or omnidirectional videos have become increasingly popular for both personal and enterprize use-cases. However, video streaming has very high bandwidth processing requirements. State-of-the-art viewport-based solutions lower these requirements by performing selective based on long-term Field-of-View (FoV) prediction mechanisms. sometimes user movement is extremely unpredictable during some parts of the video, applying adversely affects overall quality experience (QoE). This paper proposes a novel Combined tile-based adaptive solution (CFOV) that improves end-user QoE streaming. CFOV performs interactive tile selection more accurate dynamical viewing area identification combining results two FoV It also employs an innovative priority-based bitrate adaptation approach ensures improved budget distribution between different tiles. We evaluate proposed with comprehensive set experiments involving four immersive videos, diverse tiling patterns (i.e., 4×3, 6×4, 8×6), segment lengths 1s, 2s, 3s), 48 empirical head traces under settings. The evaluation newly defined metric specifically introduced to assess performance objectively. experimental findings show that, compared alternative approaches, our can achieve higher viewport match significantly improve watching behaviors content characteristics.