A branch and bound approach for the design of decentralized supervisors in Petri net models

The paper addresses the design of compact and maximally permissive decentralized supervisors for Petri nets, based on generalized mutual exclusion constraints. Decentralization constraints are formulated with respect to the net transitions, instructing each local supervisor to detect and disable transitions of its own control site only. A solution is characterized in terms of the states it allows and its feasibility is assessed by means of two separate tests, one checking the required behavioral properties (e.g., liveness, reversibility and controllability) of the induced reachability subgraph and the other ensuring the existence of a decentralized supervisor enforcing exactly the considered set of allowed states. The second test employs an integer linear programming formulation. Maximal permissivity is ensured by efficiently exploring the solution space using a branch and bound method that operates on the reachable states. Particular emphasis is posed on the obtainment of the controllability property, both in the structural and the behavioral interpretation.