A Novel Fitness function of metaheuristic algorithms for test data generation for simulink models based on mutation analysis

Abstract

Testing is one of the crucial activities to assure the software quality. The main objective of testing is to generate test data uncovering faults in software modules. There are a variety of testing techniques in which mutation testing is a popular approach to generate test sets and evaluate their fault detection ability. Simulink is an environment widely used in industry to design and simulate critical systems. Testing such a system at the design phase could help to detect faults earlier. This study aims to propose a novel fitness function of metaheuristic algorithms to generate test data based on the mutation technique for the Simulink models. The fitness function is designed by analyzing each mutation operator and the features of blocks in the Simulink environment in order to guide the search process to reach the test data killing mutants more easily. Then, this fitness function is used in the multi-parent crossover genetic algorithm to generate test sets. The obtained results indicated that the mutation score has been significantly improved for all models when using the novel fitness function. In addition, each stubborn mutant was killed with a lower number of test data evaluations in comparison with the work of other authors.