Improvements in BDD-based reachability analysis of timed automata

Abstract

To develop efficient algorithms for the reachability analysis of timed automata, a promising approach is to use binary decision diagrams (BDDs) as data structure for the representation of the explored state space. The size of a BDD is very sensitive to the ordering of the variables. We use the communication structure to deduce an estimation for the BDD size. In our experiments, this guides the choice of good variable orderings, which leads to an efficient reachability analysis. We develop a discrete semantics for closed timed automata to get a finite state space required by the BDD-based representation and we prove the equivalence to the continuous semantics regarding the set of reachable locations. An upper bound for the size of the BDD representing the transition relation and an estimation for the set of reachable configurations based on the communication structure is given. We implemented these concepts in the verification tool Rabbit [BR00]. Difierent case studies justify our conjecture: Polynomial reachability analysis seems to be possible for some classes of real-time models, which have a good-natured communication structure. © Springer-Verlag Berlin Heidelberg 2001.