Automatic Test Data Generation Technology Based On Data Flow Rules

In software testing field, the choice of test data is one of challenging issues in structural testing. Its appropriateness is directly related to that errors can be detected. With the growing size of software systems, the limitations of traditional testing methods have become increasingly evident. In this background, the paper combines Data-Flow and Genetic Algorithm (GA) to research the automatic generation of test data.Above all, the paper introduces the basic concepts and methods about software testing, especially data-flow testing techniques. GA's basic principles, steps, characteristics and its application in software testing field are introduced as well. According to characteristics of their complementary, an automatic test data generation algorithm based on Data-Flow is proposed in the paper. This algorithm extends data-flow technology and is given as follows. Firstly designing parameters coding and create the initial population based on GA. Secondly using nodes'information of testing requirements path to construct a new fitness function to evaluate individual. Thirdly by means of genetic operators to complete the evolutionary process for generating optimal testing data.To validate the proposed algorithm, this paper carries out experiments on three programs compared with control flow testing techniques based on Random Algorithm (RA). The results show that the algorithm proposed in this paper is better than RA in the coverage rate, searching time, number of iterations and size of the generated test data. It's proved that the designed fitness function is able to take full advantage of nodes'information in data-flow graph and find the corresponding optimal solution. It can be concluded that, the algorithm proposed can generate high-quality test data and serve software testing well.