Research And System Construction Of Collaborative Caching And Offloading Mechanisms For Mobile Edge Computing

The development of the mobile Internet and the Internet of Things has led to a surge in data traffic and the emergence of more computing-intensive and delay-sensitive applications.The deployment of large-scale cloud computing platforms enables users to transfer high-intensity computing tasks to remote cloud servers with rich computing resources,but the transmission delay is relatively long.To reduce long backhaul transmission latency to the cloud,Mobile Edge Computing(MEC)has emerged to support latency-sensitive applications.In the research on mobile edge computing,the research on cooperative caching and offloading mechanism on edge servers is abundant.On the one hand,the caching mechanism can pre-store the resources such as programs and data required for performing computing tasks on the edge server.On the other hand,the computing offloading mechanism offloads tasks to the cloud for execution when the processing capability of the edge is insufficient,thereby reducing the processing delay.However,due to the limited resources of edge servers,how to formulate effective caching and offloading strategies to decide which services are cached at the edge and which tasks are executed in the cloud is the key to improving system efficiency.Aiming at the above two research goals,this thesis studies the caching and offloading mechanism in the mobile edge computing environment.The main work includes the following aspects:(1)In the research of cooperative caching mechanism,in the scenario where the user sends a video request and the cache size and processing capacity of the edge server are given,how to better serve the user request,that is,the problem of maximizing the hit rate,we propose an adaptive video caching mechanism under mobile edge computing.Specifically,we first formulate the video cooperative caching problem as a hit-rate maximizing integer linear programming,and then decompose the caching problem into three steps:cache initialization,cache scheduling,and cache replacement.For cache initialization,we transform the optimization problem into a grouped knapsack problem,and use a dynamic programming algorithm to obtain the maximum initial hit rate,and then reversely determine the initial cached video.For cache scheduling,we specify how each video request is scheduled on different edge servers.For cache replacement,we propose a method for cache content replacement based on the value in the current edge server cluster.We conducted experiments on VECSims,an edge video simulation platform based on CloudSim,and the results show that compared with the existing latest caching algorithm,the algorithm we proposed is significantly better than the latest algorithm in all indicators.(2)In the study of joint service caching and computational offloading mechanism,for the service caching and computational offloading problem of computationally intensive tasks in mobile edge computing,we firstly models the problem as an integer linear programming with delay minimization as the optimization objective problem,and then propose a Lyapunov-optimized joint caching and computation offloading algorithm.The algorithm first defines the energy consumption queue,and constructs a quadratic Lyapunov function to represent the congestion degree of the energy consumption queue,and uses the delay and the change value of the energy consumption queue in a time slot to construct a new optimization problem to further minimize the delay.It is transformed into a balance between the computational delay and energy cost of the system;the random sampling algorithm is used in each time slot to find the optimal caching decision and unloading decision to obtain the optimal solution for the current time slot;finally,the energy consumption is updated queue.In this thesis,a realistic simulation environment is constructed for experimental verification.The results show that,compared with the existing caching algorithms,our proposed algorithm significantly outperforms the state-of-the-art algorithms in all indicators.(3)Based on the above algorithms,a caching and offloading management system in edge computing scenarios is constructed.The system mainly includes edge video scene modules and edge task offloading scene modules.The edge video scene module uses the adaptive video cooperative caching algorithm proposed in this thesis.The user first creates an edge video scene by setting basic information,edge server information,and video information,then enters the cache algorithm information,and finally checks the running result of the algorithm;the edge task offloading scene module uses the joint service caching and computing offloading algorithm proposed in this thesis.First,we set the relevant parameters of the edge task offloading scenario,then enter the cache and unloading algorithm,and finally check the running result of the algorithm.On the home page,the operation process of the algorithm in the edge computing scenario is vividly displayed.