An Efficient Dynamic Load Balancing Scheme for Multi-agent System Reflecting Agent Workload

In ubiquitous computing the multi-agent system can provide intelligent services to the users by effectively utilizing the resources distributed in the environment. The traditional load balancing approach based on the load condition of the nodes is not suitable to the multi-agent system (MAS). In this paper we propose an efficient load balancing scheme for MAS employing the models formulated to identify the condition of the agents and the relevant policies required for dynamic load balancing. The proposed scheme significantly reduces the response time by migrating some agents to balance the loads of the system. An experiment verifies this property by comparing it with the existing scheme. We also propose a MAS architecture effectively supporting the registration, modification, and deletion of agent-based services, and easy retrieval and provisioning of registered services for the users like Web service.