Incremental development of deadlock-free communicating systems

Abstract

Abstract.A basic property which distributed communicating systems have to fulfill is deadlock-freedom. For systems consisting of the parallel composition of subsystems it is complex to check deadlock-freedom because the global state space of the composition has to be investigated. This paper presents an approach by which the absence of deadlocks is preserved during the development. Small initial deadlock-free systems are stepwise extended with new functionalities to large complex systems by transformation rules which preserve deadlock-freedom. Systems are represented by finite automata extended with arbitrary local variables. A verification rule is presented which ensures that the parallel composition of such extended automata is deadlock-free. The advantage of this rule is that only information over pairs of connected subsystems is needed and not over the complete state space.