| Software testing is one of the most important techniques to validate software and reduce potential risk. However, software testing is a labor-consuming process, typically consuming at least 50% of the total costs of developing software. The most labor-intensive component in software testing is the generation of test data, which consumes at least 40% of the total testing costs. Thus, the automation of test data generation is a promising way to drastically reduce the cost. Most of existing approaches of automatic test data generation are concentrated on the programs written in simple languages using simple test adequacy criterion. This thesis addresses the empirical studies of application of four optimization techniques on the test data generation. Condition-decision coverage is used in the experiments. Empirical results on five C/C++ programs are provided in this thesis, as well as the detailed analysis and discussion of the performance of these optimization methods. |
