No abstract

We present a framework for decentralized authorization for physical access control, using smart cards, where access to individual rooms is guarded by context-dependent policies that are dynamically evaluated. Policies are specified using a logical language parameterized by events. A policy analyzer converts policy specifications into equivalent executable automata and also generates initialization information about the contexts used in these policies. While the automata are stored in users' smart cards, context initialization information is disseminated in the system. We also provide a context modeling mechanism that supports construction and propagation of contexts in the system. Upon an access request, user automata are executed at the point of access in the presence of current context information. This results in an allow/deny decision. The benefit of this approach lies in resolving authorizations in a decentralized manner in situations where the solution needs to scale with increasing number of users.