Evolutionary Testing of Unstructured Programs Using a Testability Transformation Approach

Abstract

Evolutionary testing is an effective technique for automatically generating good quality test data. However, under the Node-Orient criterion, the technique is hindered by the presence of unstructured control flow within loops, this is because the control dependence is effectively ignored by the fitness function. In this paper a method of testability transformation is proposed in order to circumvent the problem, the approach is a source-to-source transformation that aims to improve the performance of evolutionary testing for unstructured programs. An experimental study is then presented, which shows the power of the approach, comparing evolutionary search with transformed and untransformed versions of two programs, the results show that our new fitness calculation rule could effectively guide evolutionary search to successsfully find the required test data at low cost.