Search-based approach of automated test generation for state-based components

Software testing is an important and expensive process in software development. Test generation is one of the most difficult and time-consuming activities in software testing process. To improve software testing process and reduce the cost of software development, test generation must be automated.; Many software applications are developed using component technology. And the quality of software applications essentially depends on the quality of software components. To assure the quality of software applications, the components must be evaluated that they have desired behavior or properties required by the applications. The existing methods of automated test generation can be used to evaluate state-less components. However, these methods are not applicable to identify properties of software components for multi-operation components that exhibit state-like behavior. Notice that the search not only needs to identify values of input variables but also a sequence of operations. Using the existing automated test generation methods for state-based components can not achieve adequate component testing or component evaluation.; The goal of this Ph.D. research was to develop and investigate automated execution-oriented methods to generate test data for multi-operation state-based components to identify component properties. To accomplish this goal, novel search based test generation approach was developed as part of this PhD research. The approach by probing identifies operations that influence the fitness function related to a property under evaluation. These operations are then used in the search process. The experimental study was conducted to evaluate the effectiveness and the cost of the approach. The experimental study has shown that the approach may be very effective in identifying component properties with a reasonable cost. In addition, we have investigated application of static dependency analysis to reduce the cost of the search by identifying all influencing operations before the search. The experimental study has shown that static analysis can significantly reduce the cost of the search.