Research On QoE-Oriented Wireless Transmission Scheme For DASH-Based Video

With the development of mobile communication,the volume of video traffic is exploding,which brings unprecedented challenges to wireless transmission of video services.As one of the hottest video technologies,Dynamic Adaptive Streaming over HTTP(DASH)has been widely concerned by researchers all over the world.Its adaptive feature of various coding rates significantly improves the smoothness of video playback,alleviates the pressure of video transmission and brings great improvement of users' perceptive experience.However,wireless resources are extremely limited and radio environment is always unstable,so it is still hard to deliver DASH-based videos over wireless networks.In addition,the evaluation index of DASH-based video services,Quality of Experience(QoE),is actually complicated to quantify and measure.Therefore,it is extremely difficult for the researchers to enhance the QoE performance of wireless DASH-based video services.At present,wireless transmission technology of DASH-based video is divided into two main aspects:short-term scheduling and long-term management of wireless resources.This paper will investigate wireless transmission technologies of DASH-based video from these two aspects,aiming at improving QoE performance of wireless DASH services.Regarding short-term scheduling of wireless resources,there is no real-time resource blocks scheduling schemes for DASH-based video transmission in LTE multi-cell networks.Thus,we establish a Network Utility Maximization(NUM)problem,namely maximizing the sum of weighted rates of all users in LTE multi-cell systems,jointly concerning the size of requested segments,buffer level and average throughput.Furthermore,a scheme of joint allocation of resource blocks and transmit power for DASH-based video in LTE multi-cell networks is proposed.In this scheme,we divide the original problem into two steps:resource blocks scheduling with fixed transmit power and transmit power allocation with fixed resource blocks.The simulation results show that,compared with the state-of-the-art schemes,the frequency and the duration of re-buffering events are reduced by more than 45%,and the QoE performance is significantly improved.For long-term management of wireless resources,most of researches currently consider novel wireless network architectures,such as cache-enabled heterogeneous networks.In this framework,user association plays a key role in load balancing.At the same time,version selection severely affects users' QoE performance,and there is no joint optimization scheme for user association and version selection at this stage.Hence,we propose a joint optimization problem of user association and version selection,aiming to maximize the sum of users' QoE utility in the systems.The utility function mainly considers the required download rate of the requested version,hit rates of cached video files and dissatisfaction of users denied by the networks.In this paper,the original problem is decoupled into two phases:a multiple-choice knapsack sub-problem and a many-to-one matching sub-problem,which are solved in a distributed way.The results imply that users are associated with small cells and select different versions according to the required rates of video versions,channel conditions,cache status and the traffic loads.Simulation results show the effectiveness of our proposed method.