Multi-agent Systems on the Internet: Extending the Scope of Coordination towards Security and Topology

Abstract

The Internet is rapidly becoming the privileged environment for today’s Multi-Agent Systems. This introduces new issues in MAS’ design and development, from both a conceptual and a technological viewpoint. In particular, the dichotomy between the openness of the execution environment and the need for secure execution models makes governing agents’ interaction a really complex matter, especially when mobile agents are involved. If coordination is managing the interaction, then the issue of agent coordination is strictly related with the issues of topology (how the space where agents live and possibly move is modelled and represented), authentication (how agents are identified), and authorisation (what agents are allowed to do). To this end, we first discuss the TuCSoN model for the coordination of Internet agents, then show how it can be extended to model the space where agents live and interact as a hierarchical collection of locality domains, where programmable coordination media are exploited to rule agent interaction and to support intelligent agent exploration. This makes TuCSoN result in a single coherent framework for the design and development of Internet-based MAS, which takes coordination as the basis for dealing with network topology, authentication and authorisation in a uniform way.