Faulty interaction identification via constraint solving and optimization

Abstract

Combinatorial testing (CT) is an important black-box testing method. In CT, the behavior of the system under test (SUT) is affected by several parameters/components. Then CT generates a combinatorial test suite. After the user executes a test suite and starts debugging, some test cases fail and some pass. From the perspective of a black box, the failures are caused by interaction of several parameters. It will be helpful if we can identify a small set of interacting parameters that caused the failures. This paper proposes a new automatic approach to identifying faulty interactions. It uses (pseudo-Boolean) constraint solving and optimization techniques to analyze the execution results of the combinatorial test suite. Experimental results show that the method is quite efficient and it can find faulty combinatorial interactions quickly. They also shed some light on the relation between the size of test suite and the ability of fault localization. © 2012 Springer-Verlag.