Model-based higher-order mutation analysis

Abstract

Mutation analysis is widely used as an implementation-oriented method for software testing and test adequacy assessment. It is based on creating different versions of the software by seeding faults into its source code and constructing test cases to reveal these changes. However, in case that source code of software is not available, mutation analysis is not applicable. In such cases, the approach introduced in this paper suggests the alternative use of a model of the software under test. The objectives of this approach are (i) introduction of a new technique for first-order and higher-order mutation analysis using two basic mutation operators on graph-based models, (ii) comparison of the fault detection ability of first-order and higher-order mutants, and (iii) validity assessment of the coupling effect. © 2010 Springer-Verlag Berlin Heidelberg.