Research on Task Assignment Based on Agents in Grid Computing

In hybrid heterogeneous and dynamically changing distributed grid environment, how to assign tasks to appropriate resources is one of the most challenging techniques. In this paper, we present a novel grid task assignment model and corresponding algorithm based on agents, small-world network model theory, and power-law. With the help of agents to assign tasks, the proposed model and algorithm can exactly assign grid tasks to suitable resources in real time. The proposed model and algorithm are also designed applying with the two basic rule of complex networks, which makes the nodes of the grid system more searchable, and thus helps the task assignment agents to find needed resources for specific tasks more quickly and efficiently. Simulation experiments show that the model and algorithm can make sound load balance and performance improvement compared with other traditional method.