| With the development of computer technology, embedded systems are being suede more and more in many safety-critical areas. The usage of these embedded systems has brought great convenience to human beings at the same time the failure of these systems will also bring incalculable loss to life and property. How to improve the reliability of embedded systems and how to improve the reliability of embedded software in particular has become a widely noticed focus for researchers. Software testing plays an important role in enhancing the credibility of software systems, and model-based statistical testing has been widely used because of its high efficiency, low consumption at home and abroad.This paper focuses on the automation of statistical testing mainly including the following: Firstly, an automatic optimization method of Markov usage model and its framework are proposed in this paper. This optimization method is based on the actual use of the target software therefore it can be widely applied to all Markov usage model based statistical testing. Secondly, the state entropy-based model generation method of automatic optimization is researched and the use constraints and the steps of the optimization algorithm are analyzed in detail. Experiments show that this method can not only increase the automation level of statistical testing but also can effectively reduce the model complexity to improve testing efficiency and save cost of tests. At the same time the generation problem of smooth distribution under the condition of the unknown of transition probabilities can be solved with the optimization methodFinally, a statistical testing automation tool named STATool is developed based on the Windows and Microsoft Visual C++ 6.0 and supports the automatic optimization generation of usage model, the parametric analysis and the test case generation. The STATool increases the automation level of statistical testing and gives a reference to the researches on automated testing tools. |
PDF Full Text Request