Empirical analysis of a genetic algorithm-based stress test technique

Evolutionary testing denotes the use of evolutionary algorithms, e.g., Genetic Algorithms (GAs), to support various test automation tasks. Since evolutionary algorithms are heuristics, their performance and output efficiency can vary across multiple runs. Therefore, there is a strong need to empirically investigate the capacity of evolutionary test techniques to achieve the desired objectives (e.g., generate stress test cases) and their scalability in terms of the complexity of the System Under Test (SUT), the inputs, and the control parameters of the search algorithms. In a previous work, we presented a GA-based UML-driven, stress test technique aimed at increasing chances of discovering faults related to network traffic in distributed real-time software. This paper reports a carefully-designed empirical study which was conducted to analyze and improve the applicability, efficiency and effectiveness of the above GA-based stress test technique. Detailed stages and objectives of the empirical analysis are reported. The findings of the study are furthermore used to better calibrate the parameters of the GA-based stress test technique.