Research On Tsak Scheduling In Cloud Computing Based On Improved Particle Swarm Optimization And Ant Colony Optimization

With the explosive growth of the Internet users and various applications,cloud computing has greatly developed through this process.Users get access to resources from the cloud service has become a new business model.Meanwhile,the diversity and uncertainty of the user's tasks have brought new challenges to cloud computing.In this case,efficient task scheduling has become one of the key directions of cloud computing.Task scheduling in cloud computing aims at how to assign appropriate resource to user.A good scheduling algorithm can not only make sure that the environment of cloud computing is robust and stable,but also reduce the reponse time to users,in case to satisfy the cloud computing users.The target of this thesis is to shorten the completion time of task scheduling.Ant Colony Optimization and Particle Swarm Optimization are two kinds of bio-intelligence scheduling algorithms.Ant colony optimization learns from the information transmission mechanism between ants.The core idea of Particle Swarm Optimization is the simulation of the behavior when the birds seek for food.Both algorithms perform well in task scheduling of cloud computing.Three aspects of the work done by this thesis is shown as follows:(1)By researching on task scheduling problems in cloud computing,analyzing the classification and the characteristics and requirements of task scheduling in cloud environment,a cloud platform task schedluling mathematical model is established.And the model is about how to map the tasks in cloud to resources node properly in discrete cases.In this case,Particle Swarm Optimization is discreted to apply to the task scheduling problem in cloud environment.(2)This thesis focuses on Particle Swarm Optimization,by analyzing the core process of Particle Swarm Optimization,some interesting characteristics of Particle Swarm Optimization has been found.Particle Swarm Optimization converges quickly and calculates less compared to other intelligent algorithms,but it is easy to fail to a local optimal.An inertia weight optimization based on sinusoidal strategy is propersed after introducing some basic optimization strategies based on inertia.And Particle Swarm Optimization is discretized for task scheduling after analyzing the characteristics of cloud computing task scheduling problems.The improved discrete Particle Swarm Optimization(IDPSO)is superior to the standard discrete Particle Swarm Optimization in shortening the completion time of task scheduling.(3)By combining Particle Swarm Optimization and Ant Colony Optimization,so called PSO-ACO.PSO-ACO converges quickly in the early stage,calculates less and has better performance in the later period because of the positive feedback mechanism,both advantages of Particle Swarm Optimization and Ant Colony Optimization are absorbed by PSO-ACO.By simulating this three algorithms,it's found that PSO-ACO have better performance in shorten the scheduling time than both Particle Swarm Optimization and Ant Colony Optimization.