Study On Multi-user Cooperative Service Caching Strategy In Multi-access Edge Computing

The rapid growth of innovative mobile services such as augmented reality,online gaming,and autonomous driving has prompted the need to access computing resources with low latency to ensure user experience,while the existing centralized cloud systems cannot fully meet the low latency requirements of these applications due to their inherent transmission latency.Multi-access edge computing(MEC)enables applications to be executed closer to the user by sinking storage,computing,and service functions to the edge of the network adjacent to the end-user,thus effectively reducing transmission delays.As an emerging and promising computing paradigm,it is expected to be a solution to meet low latency service requirements and alleviate the physical resource bottleneck of smart mobile devices(SMDs).At the same time,the MEC system can provide a service platform with low latency,high bandwidth and direct access to network information for SMDs.Although a number of recent studies have addressed the issues of computation offloading and content caching under MEC systems,the focus has primarily been on efficient utilization of computing resources,while ignoring the fact that some applications need to store important data in advance to enable services.Therefore,the concept of service caching is proposed to further optimize the system's Quality of Service(Qo S).In MEC system,service caching can effectively improve the link traffic load and system Qo S.However,it is very challenging to flexibly configure the edge service cache within the limited edge storage capacity to enhance the system performance caused by the high coupling relationship between the service caching and the offloading decision.At the same time,the rapid development of 5G technology has constituted a complex multi-cellular network environment,which enables multiple MEC servers to cooperate and share computing tasks through backhaul link,so as to realize load balancing and resource sharing within the system.In this thesis,the asymmetric bandwidth requirements,combined service caching and computation offloading are first considered in a multi-cellular network with MEC support,and describes the problem of joint service caching and computation offloading to maximize the number of users that the base station can serve.By analyzing the approximate submodularity of the objective optimization function,a configuration update algorithm based on random rounding is proposed to increase the utilization of MEC servers to the greatest extent.Further,we minimize the average task execution time in the MEC system by considering the heterogeneity of task requests,the pre-storage of the application data,and the cooperation of the base stations.On this basis,we formulate the problem of joint computation offloading,service caching,and resource allocation as a mixed integer nonlinear programming(MINLP)problem.In order to overcome the coupling relationship between optimization variables,we solve the MINLP problem by the generalized Benders decomposition based on decomposition theory.Moreover,we develop an efficient algorithm of collaborative service caching and computation offloading,called Gen COSCO,to improve Qo S while reducing computational complexity.In particular,for special case when the service cache configuration is fixed,the Fix SC algorithm is proposed to obtain a system computation offloading policy through dynamic cache updates.Finally,the simulation results show that compared with other benchmark algorithms,the proposed configuration update algorithm based on randomized rounding can effectively increase the number of user tasks offloading to the base stations and maximize the benefits of MEC system operators.At the same time,our proposed framework of joint computation offloading,service caching and resource allocation can effectively reduce the average execution time of computing tasks in the system,improve the quality of experience of users and reduce the burden of network links.