This algorithm involves some modifications of genetic operators and the employment of memory mechanism to enhance its effectiveness. The proposed approach is implemented to generate test data for Simulink models. The obtained results indicated that IGA outperformed the conventional genetic algorithm in terms of the quality of test sets, and the execution time. | Journal of Computer Science and Cybernetics, , (2017), 50–69 DOI AN IMPROVED GENETIC ALGORITHM FOR TEST DATA GENERATION FOR SIMULINK MODELS LE THI MY HANH, NGUYEN THANH BINH, KHUAT THANH TUNG The University of Danang - University of Science and Technology, Vietnam ltmhanh@; ntbinh@; thanhtung09t2@ Abstract. Mutation testing is a powerful and effective software testing technique to assess the quality of test suites. Although many research works have been done in the field of search-based testing, automatic test data generation based on the mutation analysis method is not straightforward. In this paper, an Improved Genetic Algorithm (IGA) is proposed to increase the quality of test data based on mutation coverage criterion. This algorithm involves some modifications of genetic operators and the employment of memory mechanism to enhance its effectiveness. The proposed approach is implemented to generate test data for Simulink models. The obtained results indicated that IGA outperformed the conventional genetic algorithm in terms of the quality of test sets, and the execution time. Keywords. Genetic algorithm, mutation testing, simulink, test data generation. 1. INTRODUCTION Software testing is an expensive, tedious, and time-consuming activity but it is a crucial step to improve the quality and the reliability of software. The process of generating test data decides the effectiveness and efficiency of software testing. The quality of test data is normally measured by the adequacy criteria. Adequacy criteria, also referred to as coverage criteria, pose certain requirements that should be fulfilled by test cases. Mutation testing proposed by DeMillo et al. [6] is a powerful and effective testing technique to assess the quality of test suites. The driving principle of mutation testing is the use of faults which mimic mistakes that a competent programmer would make. These faults are introduced into
