Research On Workflow Scheduling Scheme Based On Cloud Computing

In recent years,the rapid development of cloud computing has made it become a hot research topic.Among them,workflow task scheduling and resource allocation have also been widely concerned in cloud computing.In cloud computing,workflow task scheduling is the process of mapping tasks,which depend on each other,to resources and execute them.Workflow task scheduling algorithm is to find the optimal scheduling scheme to complete the mapping of tasks and resources,and successfully execute tasks,in the case of meeting the user's needs.Particle Swarm Optimization(PSO)can be applied to cloud computing workflow scheduling to get the mapping of tasks in workflow and resources.First of all,this thesis proposes a scheduling strategy based on resource preprocessing.In the scheduling process,the number of resource is random,not practical,and it will make the particles become very blind when updating the position during the iteration.In order to solve the problem,this thesis re-encodes resources with the cost of the resource and its memory as a standard,so that the number of resources has practical significance in the scheduling process,and the particles will move according to the constraint target when updating the position.Finally,the thesis realizes the improved scheduling strategy,which makes the particle determine which position is more in line with the constraint target during the movement.Secondly,this thesis proposes an adaptive PSO scheduling strategy.During scheduling,inertia weight plays an important role in balancing local search with global search.Setting the inertia weight as a constant can not balance the local search with the global search,which will affect the final scheduling cost.In response to this problem,this thesis presents a dynamic inertia weighting mechanism,and in order to ensure the stability of the scheduling,the acceleration coefficient will change with the inertia weight changes.Finally,the thesis realizes the improved scheduling strategy.The inertia weight and acceleration coefficient in the scheduling strategy have the characteristics of dynamic change.Finally,this thesis proposes a meta-based PSO scheduling strategy.In the process of scheduling,when the local optimal location information of the particle is dominant,the last obtained global optimal location information is not the real optimal information in the search space.In order to solve this problem,this thesis proposes to combine Iterated Local Search(ILS)and its perturbation property so that the particle swarm can jump out of the current search space when it falls into the local optimum so as to obtain the optimal search space information.