MEC-based QoE Optimization for Adaptive Video Streaming via Satellite Backhaul

Recently, satellite communication has been introduced into terrestrial cellular networks as the backhaul link, which makes it feasible to offer cellular services in some rural and remote areas without mobile cellular network coverage. However, due to the limited bandwidth and expensive price of satellite communication, it is not economical to provide high-resolution video streaming service via satellite backhaul. In this paper, we present a Super-Resolution (SR)-based adaptive video streaming scheme in wireless network with satellite backhaul, which utilizes Mobile Edge Computing (MEC) server’s computational capacity and SR algorithm to transmit high-quality video while saving satellite backhaul’s bandwidth. Firstly, in order to eliminate SR algorithm’s critical reliance on client devices’ computational power, we introduce MEC server to tackle computationally expensive SR tasks. Specifically, we employ a multi-scale SR algorithm that can enlarge low-resolution video chunks with 3 scale factors (x2, x3, or x4) (i.e., zoom in the height and width of a video chunk to 2, 3, or 4 times). Next, in order to reduce the extra delay brought by the SR reconstruction process, a pipelined processing mechanism is proposed. Finally, considering the variation of network condition in Radio Access Network (RAN) and the heterogeneous computational capacity of MEC servers, we design a Reinforcement Learning (RL)-based adaption algorithm to dynamically choose a proper SR reconstruction scale factor. Performance evaluation over real-world network traces shows that the proposed scheme outperforms baseline schemes by 13%-63% in terms of Quality of Experience (QoE) under different computational capacities of MEC server.