A Decision Procedure for Solving Constraint Systems in Presence of Multiple Independent Intruders

We consider a model of multiple independent intruders that have no ability to share knowledge between each other. We use this model to analyze security in wireless ad-hoc networks, where each intruder has a local control in the network, i.e., he can read and send messages only to his direct neighbors. Another application is the mobile ambient calculus where several intruder processes are not able to exchange their knowledge. Both these security problems can be reduced to satisfiability of lazy intruder constraint systems, for a bounded number of steps of the honest agents. However, the constraint-based verification method usually relies on a well-formedness property of constraints. This well-formedness entails that the constraints can be ordered so that the intruder knowledge is monotonically growing. This does not hold for several intruders that learn independent of each other. For the resulting generalized class of weak-well-formed constraints, we give a novel constraint reduction procedure and prove that it is sound, complete and terminating. We also prove that it is NP-complete. KeywordsMultiple independent intruders, constraint-based deduction, routing protocols, mobile code.