Research On Particle Swarm Optimization Based Task Scheduling For Cloud Workflow System

Cloud computing is a large model of resource sharing model.The Cloud computing platforms offer every user ubiquitous,convenient and on-demand network computing resources services by making full use of massive distributed network resources.The key characteristic of the cloud computing is on-demand self-service,broad network access,resource pooling,rapid elasticity,measured service.The workflow is a kind of partly or entirely automatic execution business processes by computer.The workflow management system receives tasks from users and allocates the suitable resource to the task according to the user's demand.The object of the cloud computing is to provide user more efficiency and cheaper resources.The QoS demand of automatic task allocates and execute promote constantly.Therefore,how to complementary advantages cloud computing and workflow and search the best task scheduling plan is an important research area.The cloud workflow systems combine resources of cloud computing with resource allocation methods of workflow.According to the dependency and priority of workflow tasks,the cloud workflow management system can allocate the suitable resources to the corresponding tasks.In the cloud environment,the resource is used with compensation.Therefore,if the resource allocation plan isn't suitable,it will increase the cost of cloud service provider and insufficient resource utilization.Above all,the way how cloud workflow system allocates the suitable resource to the corresponding tasks is a very important issue.The task scheduling algorithm in cloud workflow system can allocate suitable resources to corresponding tasks.In the early research stages,the cloud service providers focus on the execution cost of tasks.Therefore,most optimization object of task scheduling algorithm is to minimize the execution cost of tasks.With the development of cloud computing,the users have more requirement of task makespan.The cloud service providers also need higher resource utilization.By this time,the optimization object of task scheduling algorithm changes to minimizing the makespan of tasks.In recent years,the study of QoS in cloud services has become more and more important.Hence,the task scheduling algorithm needs to optimize the execution cost and makespan of the tasks simultaneously.Therefore,the combination way of those two factors is an important research area.With the development of cloud computing,the number of large cloud data center increases gradually.The proportion of cloud service energy cost in total cost becomes much heavier.The optimization of the cloud data center energy cost is a big challenge.The cloud workflow management system can reduce the energy cost of tasks by optimizing task scheduling in cloud environment.However,there is less research on the energy cost of cloud workflow system.So the task scheduling algorithm can't increase the resource utilization of the cloud server and reduce the energy cost of tasks.The current research of energy optimal tasks scheduling algorithm only optimizes the single object(energy cost or QoS).And then the performance of cloud workflow service can't satisfy the users demand.Therefore,the task scheduling algorithm reduces the energy cost of task by satisfying the users QoS demand,which is a very important issue.At present,the task scheduling algorithms of cloud workflow system often use particle swarm optimization.However,traditional inertia weight strategy of particle swarm optimization(PSO)algorithm has the disadvantages,which are easy to fall into local optimum and slow convergence.Therefore,the PSO algorithm can't effectively reduce the cost and energy consumption of the scheduling plan.To address such a problem,a fine adaptive inertia weight strategy is proposed in this paper.Our method improves the adaptability of inertia weight and accuracy of particles velocity by comprehensively representing the information of particle position.It avoids premature convergence by a proper balance between local and global search ability of particles.Then,an adaptive inertia weight-based PSO algorithm is presented to solve the cost and energy consumption of task scheduling problem in cloud workflow systems.Finally;a cost-optimal adaptive inertia weight based particle swarm optimization(COPSO)and energy-aware adaptive inertia weight based particle swarm optimization(EAPSO)task scheduling algorithm are proposed.The main works and innovations of this paper are as follows:1.According to the disadvantages of traditional inertia weight strategy of PSO algorithm,this paper improves the success value calculation method of adaptive inertia weight(AIW)strategy.A fine adaptive inertia weight(FAIW)strategy is proposed.Through using the algorithm in practice,the cost and energy consumption of task scheduling problem are solved.2.The first optimization object is the task cost.The COPSO task scheduling algorithm combines cost model of task with FAIWPSO algorithm.It can optimize the cost of task in cloud workflow.Experiments compare our fine adaptive inertia weight with other five traditional inertia weights(viz.constant,index decreasing,linear decreasing,random,and adaptive inertia weight).The results show that our fine adaptive inertia weight based scheduling algorithm can always achieve better performance than others in terms of convergence speed,fitness and cost of the scheduling plan.3.The second optimization object is the task energy consumption.The EAPSO task scheduling algorithm combines energy model of task.It can optimize the energy consumption of task in cloud workflow.Experiments compare our EAPSO algorithm with other five traditional inertia weights(viz.constant,index decreasing,linear decreasing,random,and adaptive inertia weight).The results show that our EAPSO task scheduling algorithm can always achieve better performance than others in terms of convergence speed,fitness and energy consumption of the scheduling plan.This paper researches the cost and energy consumption of task scheduling problem in cloud workflow systems.A FAIWPSO algorithm is proposed to solve the cost and energy consumption of task scheduling problems.The experimental results show that the COPSO and EAPSO task scheduling algorithm can always achieve better performance in cost and energy consumption problem in cloud workflow systems.