Study On Task Scheduling Strategy Based On Multi-Objective Optimization For Cloud Computing

With the development of network technology,cloud computing as the most widely used business service model will have a profound impact on the development of society and economics.Task scheduling is one of the key technologies in cloud computing.The quality of scheduling strategy directly affects the scheduling efficiency and operating performance of the entire cloud system.A reasonable scheduling strategy can reduce the finish time and cost of the task,improve the utilization,reliability and user satisfaction,bringing good user experience.However,due to the complexity of the cloud system and the diversity of users,the research on task scheduling in the cloud environment has become extremely difficult.At present,most researches on task scheduling based on single target.However,in the process of cloud computing task scheduling,optimizing single target can not satisfy the actual needs.At this stage,there is still much room for research on multi-objective scheduling.In addition,combining with the characteristics of cloud computing,the traditional scheduling algorithm can not realize multi-objective optimization and solve other issues.This thesis chooses an intelligent heuristic algorithm for task scheduling.By studying the task scheduling models,targets and algorithms in the cloud environment,aiming at solving the single task scheduling target,this thesis selects the finish time,cost of tasks,bandwidth and load balancing as scheduling indicators,and defines multi-objective task scheduling model.Combined with the improved DE_ACO algorithm,a multi-objective task scheduling algorithm based on DE_ACO algorithm is proposed.This algorithm fuses the differential evolution algorithm and the ant colony algorithm to benefit each other.Then the differential evolution algorithm is used to quickly find the global optimization capability,which avoids the blind search solution of the ant colony algorithm in the early stage for the lack of pheromone,and accelerates the convergence speed of the ant colony algorithm,meanwhile,improves the ability of the differential evolution algorithm to optimize.The algorithm also improves the pheromone update rules,adding a load factor to enable dynamic adjustment of the load.This algorithm was performed on the cloud simulation platform called CloudSim.Experiments show that the algorithm has certain feasibility and effectiveness in shortening task completion time,reducing task completion cost and reducing loadimbalance.