Partial order reductions for failures refinement

Abstract

Partial order reduction techniques have been introduced to avoid the problem of state space explosion arising in the verification of concurrent systems. The overall idea is to exploit only some of a large number of interleavings of concurrent transitions while retaining the possibility of checking certain properties. In this paper we investigate how much structure of a system has to be preserved for checking failures refinement. In contrast to previously proposed techniques we focus here on exploiting independencies between visible actions.