Software reliability testing data generation approach based on a mixture model

To solve the problem about software reliability testing cases and testing data generation of real-time control systems, this study applies the reliability testing cases generation approach based on the mixture of operation profile and Markov chain which describes software operation profile by the use cases of UML, establishes the use model based on UML model for automatically deriving the testing model from the use model, generates a reliability testing case set based on the testing model and generates testing input data of reliability testing semi-automatically by eliciting input and output variables and abstracting testing input and output classes. The results of reliability testing on the actual airborne software show that the framework of the test case set generated by the proposed model is fairly stable. Thus, the proposed approach is suitable for generating reliability testing data of the real-time control system software semi-automatically, greatly simplifies the reliability testing process, improves testing efficiency and ensures testing validity. © 2010 Asian Network for Scientific Information.