Research On Optimization Of Resource Allocation In Mobile Edge Networks

Many intelligent mobile devices have surged with the rapid development of Io T technology and the 5G communication network.As a result,many new interactive applications(such as AR,autonomous driving,image recognition,etc.)have emerged,usually computationally intensive and delay-sensitive.Mobile Edge Computing(MEC)extends traditional cloud computing in the edge network.MEC enables mobile devices to access the edge network’s services through a one-hop wireless network connection by deploying service resources near users.Therefore,the MEC architecture can well meet the Qo S requirements of mobile users for new applications.The resource allocation problem of the edge network has an important impact on the quality of the service provided by the edge network and the overall operating cost of the system.On the premise that the mobile edge computing architecture provides users with request services,this paper studies the problem of resource allocation and task offloading in edge networks and proposes practical solutions.In summary,the main work of this paper is highlighted as follows:(1).Since different Cloudlet deployment strategies directly impact the response delay of the edge network and the cost of Cloudlet deployment,this paper aims to minimize the Cloudlet deployment cost of the edge system for the optimized deployment of Cloudlets in the edge network.On the premise of satisfying user access delay requirements,a Fitness-based Cloudlet Placement Algorithm(FCPA)is proposed.FCPA determines Cloudlet deployment sites and deployment numbers by assessing the fitness of access points by affecting various performance indicators that provide services to mobile users.The simulation results show that FCPA has a good performance in reducing the cost of Cloudlet deployment.(2).When the needs of mobile users in the edge network are more diverse,a practical solution is to configure a corresponding Virtual Machine(VM)for each mobile service in the edge server of the edge network to provide users with specific applications services.Aiming to deploy virtual machines that support multiple application services in mobile edge networks,this paper builds a cloud-edge collaboration computing architecture to minimize edge network data traffic.According to the differentiated demands of users for different application services,this paper proposes a Divideand-Conquer Based Heuristic Placement Algorithm(DCBHPA).The experimental results show that,compared with the benchmark algorithm,DCBHPA can minimize the data traffic,effectively relieve the traffic pressure of the remote cloud,and reduce the potential core network congestion problem.(3).When a user’s complex requirements consist of a series of interdependent subtasks,it is challenging to ensure that the dependent subtasks are executed in sequence and,at the same time,determine the server that the subtasks are scheduled to execute.This paper adopts the form of Directed Acyclic Graph(DAG)to describe the dependencies between subtasks and studies the task offloading problem in ultra-dense networks to minimize the average offloading cost of the system.A ranking method is designed to ensure the dependencies of subtasks,and a Clustering-based Structured Task Offloading Algorithm(CSTOA)is proposed.The simulation results show that CSTOA can effectively reduce the overall task completion time and reduce the energy consumption of mobile devices,so it is better than other offloading strategies in reducing the average offloading cost of the system.In conclusion,this paper analyzes the challenges faced by the optimal configuration of edge service resources and proposes practical solutions.