A Research Of Task-offloading Algorithm Based On Docker

With the rapid development of mobile communication technology,the current network is facing the challenges of high data rate,high traffic,high reliability and low delay brought by the development of mobile Internet and Internet of things.Facing these challenges,mobile edge computing emerges.Mobile edge computing extends cloud computing capabilities to the edge of the network,close to the geographical location of users.At the edge of the network,it provides users with services of computing processing,data storage and business perception,which meets the requirements of users for information processing speed,quantity and reliability,and greatly improves the user experience.Mobile edge computing extends the traditional network architecture.Task offloading is the key technology in mobile edge computing.Task offloading migrates the computing tasks of the client to the edge of the network for execution,and reduces the energy consumption of the client.Compared with cloud computing,the task offloading of mobile edge computing also reduces the occupancy of network resources,providing lower latency and more stable services.The key technologies in task offloading involve task-offloading execution framework,task-offloading decision-making and computing resource allocation.The task-offloading execution framework determines the process of task offloading;the task-offloading decision-making determines whether and how to migrate the computing task;the computing resource allocation determines how to allocate the computing task to the appropriate location for execution.This paper analyzes the finegrained task-offloading execution framework and the coarse-grained task-offloading execution framework.A task-offloading execution framework which uses fine-grained task partition and processes computing tasks through MEC server cluster based on Docker is proposed,its framework and task-offloading process are designed.On the basis of it,do the following two aspects of algorithm research:First,the problem of task-offloading decision-making is studied.The scheme of finegrained computing tasks brings many new features to the application,including the topological model of tasks,the calculation amount of sub-tasks after task partition and the amount of data transfer between sub-tasks,etc.The above factors should be considered in task-offloading decision-making,and the delay and energy consumption in the process of task execution should be optimized.In this paper,a specific task model is constructed and analyzed.The task-offloading decision-making problem is constructed as the optimization problem of minimizing user energy consumption under the restriction of task-execution time.The optimization problem is solved by the task-offloading decision-making algorithm based on the artificial bee colony algorithm,the domain search method in the binary artificial bee colony algorithm is improved,generate a new solution by using multi-dimensional randomly update and the new solution is determined by the Bayesian formula,thus reduces the repetition probability of new solution,speeds up the optimization speed,and the optimal task-offloading decision-making is obtained.The simulation results show that the task-offloading decision-making algorithm based on the artificial bee colony algorithm proposed in this paper can significantly save the user energy consumption while ensuring the task execution time,and has higher performance compared with similar algorithms.Secondly,the problem of computing resource allocation in task-offloading is studied.Computing resource allocation requires collaboration between servers in cluster and consideration of load in cluster.In this paper,the analytical model of computing resource allocation is constructed,a system efficiency equation is established,and the joint optimization problem of computing resource and load is constructed into the optimization problem of maximizing efficiency in the system.The optimization problem is solved by the joint optimization algorithm of computing resource and load based on the branch and bound method,and the estimates of upper and lower bounds are improved,the lower bound is estimated by the greedy algorithm,the upper is estimated based on the Hungarian algorithm,and the most efficient computing resource allocation scheme is obtained.The simulation results show that the joint optimization algorithm of computing resources and load based on the branch and bound method proposed in this paper can achieve higher efficiency,save the total time of task execution,effectively balance the load in the cluster,and have better algorithm performance.