Test Generation Based On Composite Dependence Relation Coverage Criteria For EFSM

Software testing is crucial for detecting software faults in both software development and software maintenance.The effectiveness of software testing largely depends on the quality of tests.A high-quality test case set can effectively detect the failure of the software under test(SUT).On the other hand,testing coverage criteria are often used in validating the adequacy and effectiveness of tests.In software testing,the model is an abstract representation of the software at a particular level,such as the expected behavior of the software under test,the testing strategy,and the testing environment.The EFSM model is a commonly used model that can describe the software under test more accurately.The test coverage criteria commonly used in the EFSM model include state or transition coverage criteria.Dependence analysis is an important analytical technique that can be used to testing coverage criteria.Studies have shown that using data dependence as a coverage criterion in the EFSM model can detect data flow failures well.However,EFSM models are non-termination and non-deterministic,the traditional program based control dependence cannot be used in the EFSM models.The control dependence reflects the most important structural information in the EFSM model.Compared with state or transition coverage criteria,control dependence coverage criteria can better detect the control flow fault.At the same time,compared with the data flow coverage criteria,it can can avoid the incompleteness of the test in the EFSM model fault detection.The pre-order study of the research group proposed new types of EFSM control dependence.Based on the study of the control dependence based testing coverage criterion,this paper further combines the data dependence in the EFSM models to propose a composite dependence based testing coverage criteria for EFSM,which can avoid the incompleteness of the test in the fault detection of the data flow and the control flow.This paper also explores the inclusive relation between control dependence and data dependence and gives the proof.One case study is introduced to show that the paths generated based on the control dependence can discover the failure that the paths generated based on the data dependence cannot detect.What's more,this paper shows a test generation framework based on composite dependence relation coverage criteria for EFSM.We further presents the algorithm for generating composite dependence and the rules for test path generation.In order to further verify the effectiveness of the coverage criteria proposed in this paper,ten commonly used analysis benchmark models in the EFSM model are selected in this paper.The experiment verifies the non-mutually inclusive relationship between control dependence and data dependence through the experiment.What's more,the test case set generated based on the composite dependence coverage criterion has a higher error detection rate than the test case set generated based on the data dependence coverage criteria and the transition coverage criteria.Finally,we compare the error detection rate of the test case sets generated by the three coverage criteria based on the new control dependence.The results show NTSCD and UNTICD are more suitable to be applied in the composite dependence coverage criterion.