| With the rapid development of mobile Internet technology and Internet of things(IOT)terminals,people's lives have been greatly affected.Intelligent wearable devices such as smart phones and smart bracelets are playing an increasingly important role in people's work and life.With them,our lives become more convenient.However,most IOT terminals have the same characteristics,that is,small size and weak computing power.With the change of the times,many existing IoT terminals are unable to meet the needs of persistent intensive computing.Experts and scholars propose mobile edge computing offloading(MECO)to get rid of this dilemma.The key point of mobile edge computing is to sink the function of cloud computing center to the edge of network,and use edge server to provide computing services for terminal nodes near the edge of network.In the practical application scenario of edge computing,terminal nodes deliver tasks that cannot be processed by themselves or consume a lot of energy to the edge server to enhance the overall service ability of the system.And computing migration is the top priority of the delivery process.Computing migration means that the edge terminal delivers its own computing task migration to the edge server or other idle nodes,completes the task in time,improves the utilization of system resources and reduces its own energy consumption.The core of edge computing is the choice of migration strategy.In this paper,the task migration of edge computing is studied as follows:(1)This paper summarizes the related research of computing migration,analyzes the selection of computing migration strategies and the division of computing resources in different application scenarios,and summarizes the advantages and disadvantages of different migration schemes.(2)In the edge collaboration scenario,the task migration between nodes is constrained as a task allocation problem to improve the throughput between nodes.Build the system model of multi terminal edge collaboration;analyze the task status of node tasks,build task buffer queue,build the system task migration model,and get the calculation cost of system tasks;analyze the deployment environment in the edge collaboration scenario,and build the system link cost model.The transmission cost of system migration is obtained;the evaluation mechanism of node trust is designed,and the credibility of each node is obtained;the edge collaborative task migration algorithm(ECTMA)based on greedy algorithm is proposed considering the calculation cost,transmission cost and energy consumption.The experimental results show that compared with traditional local processing,ECTMA algorithm can double the task throughput and reduce the time cost of task completion by 40%.Compared with exhaustive algorithm,it can reduce the time complexity of the algorithm from n(2n)to o(N3),and it is more universal.(3)In the MEC scenario with edge server,the addition of edge server reduces the computing pressure of the terminal and provides more powerful computing and storage support for the system.In order to prevent unnecessary transmission overhead and complete task processing within the delay constraint,the migration time constraint model is established.The characteristics of time constraint are analyzed and transformed into a 0-1 knapsack problem.Hybrid genetic algorithm(HGTMA)is designed to solve the migration problem.The experimental results show that HGTMA algorithm can reduce the energy consumption by 55% compared with the traditional full local processing scheme,and significantly reduce the task completion delay.Compared with the random migration scheme,HGTMA algorithm can also reduce the energy consumption by 20% and the delay by 60%.On this basis,we get the optimal number of users that the system can accommodate under different time slot conditions,so as to set the time slot according to the needs of the actual application scenarios and meet the needs of different user groups. |
