Theory Of Optimal Search-based Software Testing Resource Allocation Method

With the arrival of information age, the application of computers in the national economy, military, national defense and all aspects of social life becomes more and more extensive and in-depth, the soul of the computer - software plays an important role in this process, and naturally the quality of software products comes to be the focus of people's attention. As the key steps in the process of software development, software-testing is essential tools and stage for developing a quality and reliable software. and it has an irreplaceable position.Software-testing is an expensive, resource-intensive job, its cost may occupy half of the whole project expenditures, so in the process of software development, deciding how to allocate software-testing resources is very important for the entire testing process and the testing results. A reasonable distribution of software-testing resources between the software modules can shorten the testing stage and improve quality and reliability. Until now, there is no software-testing resources allocation model which is sophisticated enough, the optimization of software-testing resource has not attracted enough attention. Therefore, the study of software testing resource allocation model, particularly with the optimal search theory, is very meaningful.Optimal search theory which is first proposed by G. Kimball and B. Koopman in 1940's is developed from the statistical decision theory in operations research. The optimal search problem involves computing how resources can be allocated so as to maximize the probability of detection, or to minimize the consumption of resources used. We may categorize optimal search problems according to whether the space and time are discrete or continuous, and whether the target is stationary or mobile. In this thesis, we study the situation with stationary target in the discrete space.With limited software-testing time, the testing can not expose all defects. In order to spending the testing resources in the process of software development, we use the optimal search theory to optimize the software testing resources distribution process to acquire a better testing result. Taking into account affects that the prior probability distribution of defect impacts on optimal search model, this article discusses a variety of methods to predict the prior probability distribution of defect. Besides, we introduce the software reliability growth model and determined the form of detection function according to the Non-homogeneous Poisson process model. Finally, we propose an optimal allocation of testing resource based on optimal search theory.In the end, we do experiments to simulate real allocation of software-testing resources and the software-testing process, the experimental results show that, with limited resource, our optimal allocated strategy of software-testing resources lower the software failure probability effectively, that means, it do optimize the allocated plan.