Traxtor: An automatic software test suit generation method inspired by imperialist competitive optimization algorithms

Abstract

Software testing refers to a process which improves the quality of software systems and also is one of time and cost consuming stages in software development. Hence, software test automation is regarded as a solution which can facilitate heavy and laborious tasks of testing. Automatic generation of test data with maximum coverage of program branches is regarded as an NP-complete optimization problem. Several heuristic and evolutionary algorithms have been proposed for generating test suits with maximum coverage. Failure to maximally branch coverage, poor success rate in test data generation with maximum coverage and lack of stable results are considered as the major drawbacks of previous methods. Enhancing the coverage rate of the generated test data, enhancing the success rate in generating the tests data with maximum coverage and enhancing the stability and speed criteria are the major purposes of the present study. In this study, an effective method (Traxtor) is proposed to automatically generate tests data by using imperialist competitive algorithms (ICA) optimization algorithms. The proposed method is aimed at generating test data with maximum branch coverage in a limited amount of time. The results obtained from executing a wide range of experiments indicated that the proposed algorithm, with 99.99% average coverage, 99.94% success rate, 2.77 average generation and 0.12 s average time outperformed the other algorithms.