Dynamic search-based test data generation focused on data flow paths

Abstract

Test data generation approaches produce sequences of input values until they determine a set of test cases that can test adequately the program under testing. This paper focuses on a search-based test data generation algorithm. It proposes a dynamic software testing framework which employs a specially designed genetic algorithm and utilises both control flow and data flow graphs, the former as a code coverage tool, whereas the latter for extracting data flow paths, to determine near to optimum set of test cases according to data flow criteria. Experimental results carried out on a pool of standard benchmark programs demonstrate the high performance and efficiency of the proposed approach, which are significantly better compared to related search-based test data generation methods.