Partial Order Reduction with Compositional Verification

Abstract

This thesis expands the usage of partial order reduction methods in reducing the state space of large models in model checking. The work done can be divided into two parts. In the first part we introduce two new ample conditions that utilise strongly connected components in place of two existing ample conditions that use cycles. We use these new conditions to optimise existing partial order reduction verifiers and extend them to verify nonblocking properties. We also introduce two selection strategies for choosing ample event sets and an improved ample algorithm in order to improve the efficiency of ample set computation, and investigate how the various combinations of these suggested algorithmic improvements effect several models of varying size. The second part of the thesis introduces the concept of using partial order reduction techniques in combination with compositional verification techniques. We introduce a modified version of the silent continuation rule that makes use of the independence relationship from partial order reduction methods and include algorithms by which they may be implemented in a model verifier. All of the original concepts developed in this thesis are also proven correct.