An open framework for load balanced multi-agent systems

The open, heterogeneous and dynamic environments of pervasive computing require scalable and resource adaptive applications. Replication and dynamic load balancing are well-known mechanisms for scalability and efficient utilization of resources in parallel and distributed systems. We present an open framework for the creation of load balanced multi-agent systems. Its openness allows agent developers to develop tailored load balancers, which take customized application- and system-related information into account and are able to integrate customized balancing algorithms. Pre-defined classes and methods ensure its easy programmability. The realization of our framework is based on a Java based, FIPA compliant agent platform.