Path-oriented Test Data Generation Based On Modified Genertic Algorithm

Test data generation is a key point in software testing process. How to generate effective test data under a limited time and resource is of great value in both theoretical and applied aspects. Manually generating test data consumes too much resource and the test data is always insufficient and redundant. Automatic test data generation will make software test more efficient.In structural testing, path coverage is a popular coverage criterion. Genetic algorithm (GA) is a random-based search algorithm which simulates the natural evolution process. By conducting the selection, crossover and mutation operations, GA is trying to generate the solution during a mount of iterations. Due to its adaptability and global searching ability, genetic GA has been widely used in path-oriented test data generation. However, GA does not consider the structural information of the program under test when generating path-oriented test data thus it suffers from high iteration times and low efficiencyIn order to improve the test data generation efficiency, this paper proposes a modified genetic algorithm (MGA) which uses structural information of the program under test to help choosing the crossover and mutation point. With the help of precise crossover and mutation operations, the iteration times needed when generating test data can be reduced. Besides, a path-oriented test data generation prototype system which uses the modified genetic algorithm has been developed. Using this prototype system, test data of C programs can be generated automatically. A lot of experiments show that MGA has faster convergence speed and higher test data generation efficiency when applied in path-oriented test data generation.