Software Reliability Verification Testing To Reduce Test Case The Amount Of Method Study

The dependable software is currently widely used in various fields of social life, and compared to hardware reliability, software reliability is laggard. Now it has become a bottleneck for further enhancing the level of dependable software. Software Testing is playing an important role in raising the level of software reliability, and in general it can be divided into reliability growth testing and reliability validation testing. In order to reduce the number of test cases and test time in the reliability validation testing, this paper gives two effective methods, and at last give a method based on the second maximum likelihood estimation theory to choose the better one. The main contents and contributions of this paper are as follows:First of all, this paper focuses on two widely-used methods for producing the test cases. They are based on operational profiles and based on Markov chain model. Then, this paper designs an application example of the latter method, and finds that the convergence indicator from test chain to usage chain is flawed, and it also can not effectively reduce the number of test cases.Secondly, in order to reduce the number of test cases without reducing the reliability indicator, this paper puts forward two methods based on empirical Bayes statistics, they are, calculating priori distributions based on prior order and based on conjugate maximum entropy. And they will be explained in both discrete software implementation and continuous software implementation. At last this paper gives a method based on the second maximum likelihood estimation theory to choose a better one from the two methods.Finally, numerical simulation is given to verify the methods mentioned above. At last by comparing the method of calculating priori distributions based on prior order and based on conjugate maximum entropy, it verifies that the second maximum likelihood estimation can choose a better one from the two methods.