| dc.description.abstract | Researchers and practitioners have extensively studied various testing techniques and their importance on affecting the cost and the quality of software. One of these techniques is Model-Based Testing (MBT). MBT concentrates on test models that are software artifacts exploited for test automation. The goal of this project is to evaluate whether we can reduce the number test cases and time of test case execution in order to detect faults in MBT, by implementing structured test case generation methods, such as HSI method. To test our hypothesis, we conducted an experiment where we compare the efficiency and effectiveness of fault detection between the HSI method implemented by us and the random test case generation method implemented in a model-based testing tool called ModelJUnit. The experiment is done after investigating the existing random walk algorithm in ModelJUnit and implementing the HSI method in the presented tool. Our results indicate that the traditional technique which employs mutation, for some mutants, has better fault detection efficiency than the random walk, regarding the length of the test cases generated. But, concerning the effectiveness, measured by the number of mutants killed, the HSI method only showed better results than the random method for some cases. In these cases, the FSM model of the implementation consists of an increasing number of states, where the random walk cannot reach the deeply injected faults. | sv |
