A novel approach for test case generation from UML activity diagram

Software testing approaches are mainly divided into three types i.e. code based testing, specification based testing and model based testing. In model based testing, the testing can be started from the design process at the beginning phase. So, early detection of faults can be achieved by using this approach. An approach for the generation of test cases from UML (Unified Modelling Language) activity diagram using genetic algorithm has been presented in this paper. Early detection of faults can be achieved by this approach and will certainly reduce the time, cost and effort of the developer to a large extent. We propose a model to generate an Activity Flow Table (AFT) from an Activity Diagram and then convert it to Activity Flow Graph (AFG). Coverage criteria are very important in test case generation. By using activity coverage criterion we traverse the AFG and the test paths are generated. Finally, we generate the test cases from these paths. Genetic Algorithm has been applied to generate test cases and also to optimise them. The model is implemented on a case study of ATM Withdrawal system.