End-to-End Multi-User 360-Degree Video Delivery Using Users' Fixation Points

Viewport-based 360-degree video delivery is a typical method to reduce video traffic for virtual reality (VR) applications. However, viewport-based solutions cause key issues in multi-user VR applications, including high video traffic due to redundant video transmission across multiple headset users and quality degradation due to viewport transitions. These problems occur in both camera-to-server and server-to-user video transmissions. In this study, we propose a 360-degree video delivery scheme for multi-user VR applications. To overcome the above issues, the proposed approach includes appropriate quality and transmission control for camera-to-server and server-to-user video transmissions. The camera extracts the estimated potential region from the dual fisheye video. The server controls recompression at the server to follow the viewport transition and hybrid unicast and multicast tile delivery to avoid redundant transmissions. Evaluations using 360-degree video and corresponding fixation points from multiple users show that the proposed scheme prevents redundant transmissions across multiple headset users and provides better viewport quality for each user under the same video traffic. For example, the proposed scheme reduces video traffic by up to 36.4% compared to the existing viewport-based 360-degree video delivery scheme for ten headset users.