An approach to generate software test data for a specific path automatically with genetic algorithm

We focus on software reliability with testing coverage, which will grow with increment of the coverage. We expect to improve quality of software testing with it automated. An approach of generating test data for a specific single path is presented in this paper, different from the predicate distance applied by most test data generators based on genetic algorithms. A similarity between the target path and execution path with sub path overlapped is designed as fitness value to evaluate the individuals of a population and drive GA to search the appropriate solutions. Several experiments are taken to examine the effectiveness of the designed fitness function, which evaluate performance of the function with the convergence ability and consumed time. Results show that the function performs well compared with other two typical fitness functions for specific paths.