Edge-assisted Viewport Adaptive Scheme for real-time Omnidirectional Video transmission
Omnidirectional applications are immersive and highly interactive, which can improve the efficiency of remote collaborative work among factory workers. The transmission of omnidirectional video (OV) is the most important step in implementing virtual remote collaboration. Compared with the ordinary video transmission, OV transmission requires more bandwidth, which is still a huge burden even under 5G networks. The tile-based scheme can reduce bandwidth consumption. However, it neither accurately obtain the field of view(FOV) area, nor difficult to support real-time OV streaming. In this paper, we propose an edge-assisted viewport adaptive scheme (EVAS-OV) to reduce bandwidth consumption during real-time OV transmission. First, EVAS-OV uses a Gated Recurrent Unit(GRU) model to predict users' viewport. Then, users were divided into multicast clusters thereby further reducing the consumption of computing resources. EVAS-OV reprojects OV frames to accurately obtain users' FOV area from pixel level and adopt a redundant strategy to reduce the impact of viewport prediction errors. All computing tasks were offloaded to edge servers to reduce the transmission delay and improve bandwidth utilization. Experimental results show that EVAS-OV can save more than 60\% of bandwidth compared with the non-viewport adaptive scheme. Compared to a two-layer scheme with viewport adaptive, EVAS-OV still saves 30\% of bandwidth.