An automata-theoretic approach to fair realizability and synthesis

Abstract

Over the last few years, the automata-theoretic approach to realizability checking and synthesis of reactive modules, developed by Pnueli and Rosner, by Abadi, Lamport, and Wolper, and by Dill and Wong-Toi, has been quite successful, handling both the synchronous and the asynchronous cases. In this approach one transforms the specification into a tree automaton. The specification is realizable if this tree automaton is nonempty, i.e., accepts some infinite tree, and a finite representation of an infinite tree accepted by this automaton can be viewed as a finite-state program realizing the specification. Anuchitanukul and Manna argued that the automata-theoretic approach cannot handle realizability checking and synthesis under fairness assumptions. In this paper we show to the contrary that the automata-theoretic approach can handle realizability checking and synthesis under a variety of fairness assumption.