Computational Environment of Software Agents.docVIP

Acta Polytechnica Hungarica Vol. 5, No. 2, 2008 Computational Environment of Software Agents1 Martin Tomá?ek Department of Computers and Informatics, Faculty of Electrical Engineering and Informatics, Technical University of Ko?ice Letná 9, 042 00 Ko?ice, Slovakia E-mail: martin.tomasek@tuke.sk Abstract: Presented process calculus for software agent communication and mobility can be used to express distributed computational environment and mobile code applications in general. Agents are abstraction of the functional part of the system architecture and they are modeled as process terms. Agent actions model interactions within the distributed environment: local/remote communication and mobility. Places are abstraction of the single computational environment where the agents are evaluated and where interactions take place. Distributed environment is modeled as a parallel composition of places where each place is evolving asynchronously. Operational semantics defines rules to describe behavior within the distributed environment and provides a guideline for implementations. Via a series of examples we show that mobile code applications can be naturally modeled. Keywords: software agent, multi-agent system, mobile code, communication, process calculus, mobile agent 1 Introduction Mobile agent [1] is an autonomous program that decides which places of the distributed application visits and what operations uses there. Distributed systems based on mobile agents are more flexible than static ones: they support mobile users and can reduce network bandwidth [2]. It means the user just sends an agent then disconnects from network and finally receives the agent with result upon new connection. Formal description and specification of such systems is very important for modeling and successful implementation of the application. If we think of most important system characterist