Research On Grid Resource Scheduling Policy Based On Ant Colony Algorithm

Grid is a group of technologies built on Internet, which integrates the various computing resources locating in different places, offering an environment of high performance to its users and realizing the resource sharing in a wide range. Resource scheduling is very important in Grid system, one of the large-scale remote operating systems, as it matters the resource utilization of the Grid system directly and whether the whole functions could be realized effectively. Grid is featured with heterogeneousness, dynamic adaptability, resources' sharing and wide distribution, autonomy and multi-layer management, all of which determine that the importance and complexity of selecting resource scheduling policy for Grid environment.Resource scheduling in Grid system is one of NP problems, to which Ant Colony Algorithm (ACA) has been proved to be one of the effective solutions. That's why the research focuses on designing and simulating a Grid resource scheduling strategy based on ACA.In the very start, resource scheduling in Grid system has been studied, for which the work below has been done: the definition of Grid together with the main features and popular architectures are introduced, followed by the introduction and compare of several Grid simulating tools among which GridSim, whose resource scheduling process is studied in detail, is considered as the best choice for the later work of simulation after analyzing. Then the study on ACA has been done, the biological principles of ant colonies and its mathematical model illustrated, and practical experiments on ACA applied to TSP executed, whose results are showed out and analyzed to conclude the significant factors in the algorithm performance.With the work above, the resource scheduling policy in Grid environment based on ACA is designed on the software engineering basis, which means firstly the feasibility of ACA applied into Grid resource scheduling is analyzed and the main factors taking important parts in scheduling performance under Grid environment are determined after simulating resource scheduling process with GridSim, and then the preliminary designing and detailed designing are completed, in which the design of resource allocating policy and resource processing policy is the key point and the Grid resource scheduling policy based on ACA is simulated, which includes the creation of resources and tasks, the design of Pseudo Code and illustration and evaluation of the experimental results.The experimental results have showed that the resource scheduling policy based on ACA is feasible and improves the usage efficiency of resource and the general performance of Grid system by shortening the completing time of tasks and bettering the load balancing among Grid resources.