Distributed execution of aggregated multi domain workflows using an agent framework

In e-Science, meaningful experiment processes and workflow engines emerge as important scientific resources. A complex experiment often involves services and processes developed in different scientific domains. Aggregating different workflows into one meta workflow avoids unnecessary rewriting of experiment processes and thus improves the reuse efficiency. Remote workflow engines explore the computing power of distributed environment. However, the diversity of workflow description and execution models makes the integration between engines difficult. An agent framework uses ontology based communication language and makes the integration to semantic information of resources seamless, it is thus suitable for coupling distributed engines. In this paper, we present our work in the context of Dutch Virtual Laboratory for e-Science (VL-e) project. A semantic registry for describing workflow engines is implemented, and mobile agents are used to manage distributed workflow coordination.