| Video streaming service has become one of the essential applications in the mobile end and Io T field,having been highly concerned by the industry and academia.To attain more active customers,all the video service providers attend to provide users with more excellent quality of experience(Qo E),like video resolution,smoothness,and stableness.However,according to our latest investigation,some more emerging Qo E metrics have arisen with the rapid development of 5G and AI technology.On the one hand,based on a large-scale online survey conducted in 2019,we find that the ever-growing power consumption on mobile end service will significantly impact the users’ video Qo E,owning to the low-battery anxiety,even further stressing them to abandon the video interested.On the other hand,users(especially the youth)start to be enthusiastic about the video with higher quality(e.g.,ultra-high frame rate and resolution)or with specific presence(e.g.,special rendering effects or style).Fortunately,these demands can be well met with some existing video processing technologies,such as video low-power transcoding technology and video enhancement technology based on artificial intelligence.However,deploying these computing-intensive video processing in the traditional video transmission network may lead to great problems,such as the resource constraints and heterogeneity within end devices and the high latency that may be brought by remote cloud as well as the overhead on computing and storage.Therefore,this paper proposes the edge-assisted users’ emerging Qo E-aware schemes:(1)for mobile end devices’ power consumption,we offer to deploy the low-power transcoding function on the edge server.By periodically selecting the serving users to optimize the overall device playback power,thereby maximining users’ video viewing time and retention rate;(2)Aiming with the users’ personalized video requirements,we propose a general edgeassisted computing framework,which effectively balances the video enhancement effect and multiple Qo E metrics including video transmission delay and video fluctuation in the way of selecting among a variety of video enhancement models with different complexities on an online manner.Finally,many experiments have been carried out and have demonstrated the effectiveness and efficiency of the edge-assisted user emerging Qo E perception schemes. |
