Research On Edge Cache And Service Migration In Mobile Edge Computing Based On Software Defined Networks

With the rapid growth of demand for online streaming media services,video streaming media platforms have become more and more popular.Besides,users have increasingly demanded low-latency and high-quality services.Due to the highly diverse content,it is impossible to store all the required content on the edge sever with limited cache space.Therefore,how to allocate cache resources to serve as many requests as possible,while further reducing delays and improving the quality of user experience has become an urgent issue.The limited coverage of edge servers and the high uncertainty of user mobility also bring new challenges for service deployment.Because the edge server is connected to many different access points or base stations,when the user moves outside the service area of the edge server,the question of whether to migrate the ongoing service and where to move the service needs to be considered.Therefore,it is of great theoretical and practical significance to study edge caching and service migration in SDN-based mobile edge computing environments.According to the above problems,this thesis mainly includes the following three aspects.(1)Aiming at the problems of limited edge server cache space and large number of mobile devices,the edge cache method based on latency and energy balance is proposed.Firstly,the multi-layer perceptron-based cached content prediction method is proposed to predict the video content requested by the end user according to the static,dynamic or social characteristics of the video.Then,the objective function that minimizes latency and energy consumption is established.Finally,since the objective function of the edge cache based on delay and energy balance is a mixed binary integer programming problem,so the branch and bound algorithm is used to obtain the optimal edge cache strategy for minimizing the average delay and energy consumption and improving user satisfaction.The proposed algorithms are experimentally verified and compared with other algorithms by building the SDN-based mobile edge computing environment and combining a face recognition benchmark application.In the experiment of edge cache method based on delay and energy balance,the proposed edge cache algorithm is compared with the LRU?PBC?JOC and Q-LCCA algorithms.The experimental results show that the proposed edge cache algorithm can effectively improve the cache hit rate,and control the average access delay and energy cost while reducing the load of backhaul traffic.(2)Aiming at the contradiction between the limited coverage of the edge server and the mobility of user terminals leading to the increase of end-to-end delay and the reduction of user experience quality,the dynamic service migration method based on the deep Q learning is proposed to achieve seamless service migration.Firstly,the service migration problem can be described as a Markov decision process.Then,the service migration reward function is established with considering the factors such as server load capacity,transmission cost and migration cost.Finally,the deep Q-learning is used to further obtain the optimal service migration strategy which solves the problem of service interruption due to the high uncertainty of user mobility and request patterns.In the experiment of the dynamic service migration method based on the deep Q learning,the proposed service migration algorithm is compared with the Always?Myopic and Mig-RL algorithms.The experimental results show that the proposed service migration algorithm can effectively reduce the number of service migrations and improve the success rate of service migrations,while controlling the migration cost and reducing the average traffic consumed by migrations.