Video Content Delivery In Mobile Edge Networks

Mobile edge networks,which provide computing and caching capabilities at the edge of cellular networks,have been recently proposed as new solutions to accommodate explosively growing traffic demands from mobile users.However,most studies on content delivery over mobile edge networks do not take the unique properties of videos into consideration.Specifically,unlike other data,videos have unique characteristics in request patterns,multibitrate coding,and adaptive streaming.Note that the network resources required by the delivery of videos are greatly impacted by these characteristics.To improve the efficiency of resource utilization,this thesis thoroughly studies theories of video delivery in mobile edge networks.The contents of this thesis are listed as follows.Intelligent video caching in mobile edge networks is studied.To deal with the redundant traffic caused by asynchronous requests for the same video,we proposed an intelligent edge caching scheme.The proposed scheme is able to intelligently perceive the environment and then automatically learns caching policy according to history and current raw observations of the environment,without any explicit assumptions about the operating environment.Specifically,the problem of maximizing the traffic offloaded by the edge node is formulated as a Markov decision process problem.Then,an algorithm based on the actor-critic reinforcement learning method is proposed to the solve the formulated problem.Finally,real-data-driven simulation results show that the proposed scheme is able to achieve better performance on cache hit ratio and traffic offloading,when compared with traditional caching scheme.Non-orthogonal multiple access(NOMA)-enhanced scalable video multicast in mobile edge networks is studied.To deal with the transmission scheduling problem of scalable video multicast where multiple users request the same video content,we propose a NOMAenhanced scalable video coding(SVC)multicast scheme for mobile edge networks.The proposed scheme combines the successive video-layer decoding in SVC with the successive interference cancellation in NOMA,which enables a further reduction of the bottleneck effect imposed by cell-edge user equipments.Specifically,the joint problem of the resource allocation for multiple groups and the scalable multicast scheduling within each group is formulated as a mixed-integer nonlinear programming problem,which aims at maximizing the overall video quality experienced by all users.The problem is solved by an optimal algorithm based on the recursive approach and the knapsack approach.Extensive numerical results demonstrate the improved performance of the proposed NOMA-enhanced SVC multicast scheme over several baseline schemes.Adaptive device-to-device(D2D)video streaming in mobile edge networks is studied.To deal with the joint resource allocation and rate adaptation problem of device-to-device video sharing,a transmission scheduling scheme is proposed for the case in which multiple D2 D pairs request adaptive video streaming in a cell.Specifically,a dynamic network scheduling problem is formulated,with the objective of maximizing the video quality while maintaining the long-term stable performance of fluency during video playback.Then,we obtain the optimal solution to the formulated problem according to the theory of Lyapunov drift and Lyapunov optimization.Extensive numerical results demonstrate that the proposed scheme outperforms the traditional scheme.In summary,this thesis studies video content delivery in mobile edge networks,including intelligent video caching,NOMA-enhanced SVC multicast,and adaptive D2 D video streaming.This thesis aims at improving the quality-of-experience of video users by utilizing the limited network resources.