Research On Software Fault Detection And Conformance Test Generation Based On EFSM Model

Usually traditional software testing designs appropriate test cases to check whether it can output the desired result, which is the detection of error phenomenon; While the software fault detection which based on fault model can detect the nature of the error, and the advantage of fault model which based on software fault detection is to detect system failure of one or more specific. The sequential circuit and network protocol, as well as other state transfer system can be well described by Finite State Machine model. However, the model can't describe the system's data part clearly. Extended Finite State Machine extends the Finite State Machine with variables and operations on the variables, it can accurately model the data and control part of the state transfer system. For software testing in the Extended Finite State Machine- related studies, consistency of test generation and fault detection based on fault model have been the focus of the studies.The thesis analyzed the concepts and principles of software fault detection, as well as conformance testing for Extended Finite State Machine, established software fault detection model to detect the logic fault and proposed a efficient conformance testing test generation method in the Extend Finite State Machie model. The main contributions of this thesis are as follows:Firstly, summarize the techniques about software testing and fault detection based on fault model, according to the flow of formal software development, put forward a formal model of fault detection. Analyze and compare the four classical test sequence generation methods for Finite State Machine. Through the analysis of Extended Finite State Machine generated by the concept and step of conformance testing, proposes a conformance test generation fault model based on Extended Finite State Model.Secondly, according to the proposed fault detection model, set up a kind of internal logic fault detection model, and propose a heuristic detection algorithm which based on slicing the state and transfer system. The detection algorithm will curtail the state variance of the fault model. It can only detect part of the fault detection model's state and transitions changes of detection system, which can improve the ability to deal with fault detection. The experiment proved that the algorithm was effective for the model.Finally, present a method for test generation based on user-defined faults for Extended Finite State Machine specifications. This method is based on a formally defined fault model (conformance relation and types of faults) and guided by some conditions established for the test generation, proposed to use appropriate slices of the specification Extended Finite State Machine. It avoids transforming from Extended Finite State Machine to Finite State Machine. Finally the experiment proved that the algorithm was effective for getting the shorter test sequence.