QoE optimized adaptive streaming method for 360°virtual reality videos over MEC-assisted network

As well-known, virtual reality (VR) video covers the entire 360° scene, while the best visual angle of the human eyes is about 120°, and the Field of View (FoV) supported by VR terminal is only 90 ° ~ 110°. Therefore, the viewpoint-aware adaptive streaming methods have attracted more and more attention in the research of for VR video transmission over the Internet. In this context, to further improve the overall QoE of all users, this paper first design a reasonable, effective and complete system framework for tile-based 360° VR video to achieve the centralized optimization approach on viewpoint-aware adaptive streaming over future Mobile Edge Computing (MEC) assisted 5G heterogeneous networks. And then, based on the video quality, quality fluctuation as well as the error characteristics of the predicted viewpoints, a novel priority-based weighted QoE assessment method is presented to measure the contribution of each tile on the user’s QoE. Besides, the end-to-end delay of each tile is also estimated under the guidance of the optimized MEC-server assignment. On these bases, we finally formulate a novel adaptive optimization problem with the aim of maximizing the QoE of the tiles that delivers to all users within the delay constraint and solve it by developing a low-complexity Lagrange Relaxation (LR) based heuristic algorithm. The simulation results show that our algorithm outperforms the others in the overall QoE of all users, i.e., increasing video quality by at least 3.888dB and decreasing quality fluctuation by at least 1.7477 dB under the premise of guaranteeing smooth playback.