Considerations on Secure Fipa Compliant Agent Architecture

In the paper proposed secure architecture is FIPA compliant in the sense that every common FIPA compliant agent should be able to interact with agents having this extended architecture even not knowing the security mechanisms. On the other hand, agents demanding for security can, of course, refuse plain, non secured communication with non authenticated agents, etc. The security mechanisms can utilize existing agent platform mediators such as the Agent Management System (AMS), Directory Facilitator (DF), or Agent Communication Channel (ACC), e.g. for authentication purposes. But these new tasks enforce to make requirements for these mediators much more thoroughness, as FIPA specifies them too loosely for these purposes (FIPA 1998, FIPA 2000). Other approaches could be found in (Wulf 1995, Wong 1998). This paper is aimed mainly to “static”, i.e. non-mobile, agents, but it should be possible to extend the architecture by mobility support easily.