Research Of Test Case Automatic Generation Method Based On EFSM

Along with the development of information technology, which making the software products applied to various of society, as software quality and security problems are being paid more and more attention to. Therefore, how to ensure software quality is more and more researched by the majority of experts, scholars and engineering researchers. The software testing is an important means to ensure software quality and reliability. However, with the increasing complexity and scale of software, software testing has become a complex and time-consuming work, in order to reduce the human and material cost of software testing throughout the process of software development and improve the overall quality and reliability of software, software testing automation technology has become an urgent issues need to solve nowadays.Software testing automation technology one of the key points lies in the automatic generation of test cases. Model-based testing is an effective software testing methods, which can effectively describe the state and behavior of complex software systems. The Extended Finite State Machine(EFSM) expands the context variable and the predicate conditions and actions above context variable on the base of Finite State Machine(FSM). EFSM can more accurately describe the state behavior of software systems, so it is widely used among the actual project. However, due to the contents EFSM added on basis of FSM, making the automatic generation of test cases based on EFSM model more complex, and thus remains a daunting task. In this paper, on the basis of existing research methods, the paper studies automatically generate test cases further based on EFSM model, including the following aspects:(1)For the problem of EFSM model is static, but you must make EFSM model can be implemented according to the input variable during testing, and then get the test cases. In this paper, we use SMC tool based on UML modeling method to make EFSM model into Java code, and then use path traversal algorithm to make static EFSM model with dynamic execution capability combining graph theory, and thus create the executable model.(2)For the problem of automatic generation of test sequences based on EFSM model, this paper analyzed and studied the previous method for automatic generation of test sequences on the basis of the EFSM model, we convert state diagram into transition diagram, and then using the modified breadth-first search algorithm for automatic generation of test sequences which from the initial transition to reach all other feasible target transition. The paper use conflict detection methods based on data-flow and control-flow to identify the conflict problem of test sequences in process of generating the test sequence automatically. And then considering the length of test sequence and the type of operation variables, this paper designed a feasibility measure method to analysis the feasibility and effectiveness of test sequences quantificational.(3)For the problem of automatic generation of test data based on EFSM model, this paper analyzed and discussed the effectiveness and efficiency of the intelligent optimization algorithms for automatic generation of test data aspects. And then we focus on the genetic algorithm and the scatter search algorithm. For the test sequence already obtained in front, this paper use scatter search algorithm to automatically generate test data.Finally, this paper designed and implemented a series of experiments to verify the effectiveness of the test sequence generation methods described frontal, and then we compared the scatter search algorithm and genetic algorithm for automatically generate test data to verify the effectiveness of scatter search algorithm. Experimental results show that the test sequence generation method proposed in this paper can effectively generate feasible test sequence and the scatter search algorithm for automatic generation of test data has more advantages as compared to the genetic algorithm in this experiment.