| Software testing is an important measure used to discover the errors and defects of software in order to assure the quality of software. Software testing is a key part of software development cycle. It is difficult to generate appropriate test cases for testing. Test data generated artificially is often accompanied by heavy workload and blindness, resulting in low efficiency and long test cycle and low sufficiency. Automatic test data generation could compensate for the above mentioned shortcomings effectively, reducing the cost and improving the efficiency of software test.Two basic problems of software testing are Automatic test data generation and test adequacy. Path coverage is a more stringent coverage criterion. It is easier to find errors and defects in the early stage of software development by using path testing. Therefore, path testing has a high practical value. However, it is not feasible to achieve the complete path testing in actual software testing. The general practice is to select a specific path or a finite set of all paths according to certain criteria to test. Based on path coverage, this paper introduces how to use the ACO, Particle Swarm- Genetic hybrid Algorithm, and Genetic-ACO hybrid Algorithm to solve the issue of automatic generation of software testing respectively. The method of constructing the searching model and the principles of pheromone release of the ACO are described in detail. How to set the extreme values of the individual and populations used in the Genetic-Particle Swarm hybrid algorithm is fully introduced. The theory of utilizing the principles of crossover in GA to generate new paths for ants to search in the Genetic-ACO hybrid algorithm is explained concretely. Finally, the above three methods are compared with other techniques separately in order to verify their performance. The experimental results show that the three methods proposed in the paper are flexible, effective and having a certain theoretical significance and practical value. |
PDF Full Text Request