Multiagent Model for Grid Computing

For supporting the grid computing in dynamic network environment, a multiagent model is proposed in this paper. A series of formal definitions, such as the architecture of the model, the dynamic network environment (DNE), the manage agent system, the independent computing agents (ICA) which support the traditional computing base on migration, the cooperation computing team (CCT) which supports the data parallel computing , and the relations among them are given. The dynamic learning method and the fuzzy partition technique for logical computer cluster, on which the CCT runs, are studied. The computing process is described. The experiment results show that this model resolves effectively the problems of optimization use of resources in DNE. It can be fit for grid computing.