Designing reactive systems: integration of abstraction techniques into a synthesis procedure

Abstract

This paper presents a new paradigm for designing reactive systems. It combines the use of formal methods widely recognized in software engineering and synthesis procedures developed within the framework of the Supervisory Control Theory for discrete event systems. It promotes design exploration by means of a synthesis approach with the sole aim of producing reliable reactive systems. The adoption of these particular synthesis procedures is, however, not sufficient to achieve this objective, because of scalability and computational complexity issues. To circumvent these difficulties, this paper suggests two extensions with respect to conventional synthesis procedures. The first one concerns the representation of reactive programs by attributed controllers. This requires that the process to be controlled must be described not only in terms of controllable active components but also in terms of uncontrollable passive components by using timed transition graphs and algebraic specifications, respectively. The second one involves abstraction and equational reasoning to take into account the use of strongly typed objects. This requires various kinds of transformation applied to the original problem specification as well as to intermediate solutions.