Software testing, especially software reliability testing (SRT) is an effective approach to find the defects in software and improve the software reliability. So it plays an important role in software development process. Nowadays, with the widespread use of the software, people more and more focus on the software reliability. And the model of the software reliability is the core function of the estimation in software testing. this dissertation introduces the background and the present conditions of the software testing, regards the software reliability testing as the point, Main works are as follows:1) The several models of software reliability are discussed in details, including the several typical stochastic process models and non-stochastic process models. Comparisons between them are listed. The applicability and disadvantages of these models are pointed out, and the problems of software reliability model that present stage exists are analyzed.2) On the basis of the software reliability, in order to improve the efficiency of detecting fault during software testing, constructs the optimal allocation model of the test-case in partition testing. Transform the model into a classic unbounded knapsack problem and present a dynamic programming algorithm to solve it. Then draw a scheme of the optimal allocation.3) By imitating on trial about the model of the optimal allocation, when we compare with the proportion and average allocation methods, the experiential results show that this approach can improve the ability of detecting fault, and satisfy the desired reliability. It manifests the better result. |