Research On Cache Algorithm And Implementat Verification Platform In Mobile Edge Computing

With the development of mobile communication technology and the popularization of the mobile Internet,emerging industries represented by live video and short videos are booming.The data flow and content of the mobile Internet are growing rapidly,while the repeated distribution of the same content in the mobile Internet This is the main reason why the backhaul link is overloaded.In the mass content distribution scenario,caching the content,which users prefer,at the edge of the network is an effective means to improve user satisfaction.With the advent of the 5G era and the gradual development of Mobile Edge Computing(MEC),the conditions for deploying caches at the edge of the network have gradually matured.Deploying caches at the edge of the network and responding to user requests at the edge of the network can offload most of the traffic during peak traffic periods,save valuable backhaul link bandwidth,and significantly reduce user content acquisition delays and improve user experience.This thesis is based on National Natural Science Foundation of China," Research on Multi-level Cooperative Caching Method in UAV-Assisted Cellular Network ".This thesis studies the caching algorithm and implement the verification platform in mobile edge computing.Research on caching algorithm in mobile edge computing based on live network user data,predicts user preferences from two dimensions in space and time,designs and implements an MEC verification platform,and discusses the performance of the edge cache algorithm under actual network data.The main work of this paper is as follows:1)This thesis introduces the concept of MEC and the related technologies used in building the verification platform,reviews the current research results and ideas which were proposed in user preference prediction and edge caching,and points out that the current research does not consider the impact of spatial information on user preferences.2)The current MEC system edge cache lacks consideration of the impact of user space-time preference differences on content popularity.This is the reason why we proposed a caching algorithm based on space-time two-dimensional user preference prediction.Firstly,the characteristics of user requests in the time and space dimensions are analyzed,and the user's current preference is pointed out by the combination of long-term preference and short-term preference.Then,user preference characteristics are modeled from the two dimensions of time and space and the user preference spatio-temporal two-dimensional feature extraction algorithm is popersed.Subsequently,the graph convolutional network(GCN)is used to predict user preferences.At the same time,when constructing the user association relationship graph,the user association coefficient calculation algorithm is proposed,and the influence of several key parameters on the prediction results is discussed.Also,the impact of several key parameters on the prediction results are discussed,and the performance of several mainstream algorithms are compared in the current data set.When solving the cache placement problem,a multi-base station collaborative cache method is adopted,with the goal of minimizing the average user content acquisition delay,and then it is modeled as a nonlinear mixed integer programming problem,finally the problem is solved by genetic algorithm.Simulation verification shows that user preferences have a greater impact on the placement of the cache,and the algorithm proposed in this article can effectively reduce the user's content acquisition delay and improve user experience.3)In order to verify the effect of MEC system edge caching technology on improving network performance,this paper designs and implements an MEC verification platform.First,it analyzes the key functions that the MEC verification platform needs to implement.Then,based on the MEC reference architecture given by the European Telecommunications Standards Institute,the platform's architecture and function definitions are proposed.In the specific implementation process,software-defined radio technology was used to build its network layer,relying on virtualization technology to complete the construction of the infrastructure layer,three types of nodes were defined in the instance layer,and detailed design points were explained in-depth for each node And implementation plan.Finally,the platform was deployed and performance tested.It was found that MEC technology can significantly reduce the user's content acquisition delay,and can ensure the smoothness of user data transmission,and the caching algorithm and comparison algorithm proposed in this article are on the live network.Compared with the data environment,it has significant advantages.