Research On Energy-efficient Task Offloading Methods In Cloud-edge Collaborative Environment

With the development of 5G and the rapid growth of traffic,mobile devices,such as tablet PC,are more popular because of their convenience,and more applications have emerged as the times require.The completion of the tasks of some types of applications requires efficient transmission,as well as a large amount of computing resources and energy.Mobile devices cannot meet the processing requirements of some tasks,such as the requirements of computing-intensive tasks,etc.,due to the limitation of their own computing capabilities.Therefore,some studies choose to offload tasks to powerful cloud data centers for processing.However,since cloud data centers are far away from mobile devices,they cannot meet the low-latency requirements of time-sensitive tasks,and some researchers have proposed task offloading methods based on edge computing environments.The proposal of edge computing can meet the task requirements such as low latency to a large extent,and can also effectively relieve the burden of cloud data centers.However,compared with cloud data centers,the computing and communication capabilities of edge nodes are relatively limited.Therefore,this paper mainly conducts research based on the cloud-edge collaborative environment,solves the task offloading problem under the “End-Edge-Cloud” architecture,and obtains an efficient set of task offloading strategies.The main research achievements are as below:1)Aiming at the task offloading problem with the optimization objectives of reducing energy consumption,delay and multi-node load balancing,this problem is defined as a multi-objective optimization problem.Then,a task offloading algorithm based on “End-Edge-Cloud” collaborative architecture(TO-EEC)is proposed by improving ARMOEA initialization,crossover,mutation probability,etc.,and strives to obtain a set of hybrid task offloading strategies under the premise of satisfying constraints.Finally,through experimental verification,the proposed algorithm not only has a faster convergenc speed,but also has a significant optimization effect on energy consumption,delay and multi-node load variance under the condition of satisfying multiple constraints compared with other similar algorithms.2)Since the task offloading problem pays attention to the timeliness,the multi-objective evolutionary algorithm runs relatively slowly.To improve this problem,a distributed evolutionary task offloading agorithm(DE-TOA)is proposed to solve the task offloading problem.Divide the population into multiple sub-populations,and multiple sub-populations evolve in parallel to speed up the evolution.In order to ensure the high quality of the individuals in the population,different crossover and mutation probabilities are adopted for the superior and inferior individuals.It is verified by experiments with other comparison algorithms that DE-TOA can minimize the energy consumption and delay,and the algorithm running time is significantly shortened.In addition,as the number of tasks gradually increases,the running time advantage of the proposed algorithm becomes more obvious,which verifies the efficiency of DE-TOA for solving the task offloading strategy set when the number of tasks is large.