Adam: an Agent-based Middleware Architecture for Distributed Access Control

This paper outlines a conceptual architecture for an autonomic middleware component designed to provide application-independent access control for use in largescale highly-dynamic computing environments. In such environments, most notably ambient/pervasive computing environments, centralised access control policy determination is impossible or inadvisable because of the complexity of trust relationships. In the absence of centralisation, network resources are forced to make trusting decisions locally, in the light of information that they themselves can gather. Thus the architecture that is described in this paper is founded around an automatic knowledge acquisition and processing mechanism, acting as the foundations of a semi-autonomous multi-agent system (MAS). The agents dynamically organise themselves into cooperating distributed communities that mediate between users and devices (collectively known as trustees) and network resources (principals). Once activated by their owners, agents maintain user credentials, negotiate amongst themselves to establish the credibility of prospective trustees identities and cooperate to gather evidence about the likelihood of trustees adhering to the policies of principals.