Resource-cost-based multi-agent systems scheduling for Grid resource management

A resource-cost-based multi-agent systems Grid resource management method is proposed in this paper. The multi-agent systems dispatch a set of individual agents to coordinate a user job over Grid in a decentralized manner. There agents in this method are divided into three types according to different functions. This method utilizes a resource-cost-based Grid resource scheduling to select the most appropriate resource for each Grid task. In this paper, resource-cost concept and corresponding algorithm are proposed. The individual agent´s implementation based on Java is illuminated. Finally the experiment result is presented via comparing the general Grid with our multi-agent systems Grid. The result indicates that our proposed approach can reduce about sixteen percent executing time in the same numbers of subtasks condition and obtains almost fifteen percent higher bandwidth in the point-to-point communication..