Edge-assisted Viewport Adaptive Scheme for real-time Omnidirectional Video transmission

The omnidirectional application is immersive and highly interactive. In industrial scenarios, it can be used to improve the efficiency of remote collaborative work between workers. The transmission of omnidirectional video (OV) is a key step. Compared with ordinary video transmission, OV transmission requires more bandwidth, which is still a huge burden even in 5G networks. The tile-based scheme can reduce bandwidth consumption, but it cannot accurately obtain the user's field of view (FOV) area. In this paper, we propose an edge-assisted viewport adaptive scheme (EVAS-OV) to reduce bandwidth consumption. First, EVAS-OV uses a Gated Recurrent Unit (GRU) model to predict the user's viewport. Then, the users are divided into multicast clusters according to their viewports. EVAS-OV reprojects the OV frame to accurately obtain the user's FOV area from the pixel level. At the same time, a redundancy strategy is adopted to reduce the influence of viewport prediction errors and improve the robustness of the system. All computing tasks are transferred to the edge server to reduce transmission delay and improve bandwidth utilization. Experimental results show that compared to the viewport adaptive two-layer scheme, EVAS-OV can save 30% of bandwidth.