Software Reliability And Cost Model

As the software technique development, scale and complexity of software system raise continuously, people's requirements of software quality is also higher and higher. Software reliability as an important measure of software quality, its quantitative estimate and forecast of software quality has become a focus of people's concern and research.First, this paper introduces the significance and the domestic and foreign development in the research of software reliability. Also it introduces some basic concepts, methods and mathematical foundations which are related to software reliability models. Then it enumerates two kinds of classic models of software reliability growth model (SRGM), one of them which calls G-O model is the basic model of this paper. This paper builds a new software reliability model, and using it to find the optimal released time. The main results are as follows:1. Making up a function of fault removal efficiency p (t ). Most NHPP software reliability growth models assume that faults detected during tests will eventually be removed. Consideration of fault removal efficiency in the existing models is simple and limited. Practically some faults detected during tests will not be removed immediately, and the fault removal efficiency will be lower and lower during the test, so the fault removal efficiency should be a function of testing time.2. Taking p (t ) into G-O model, and building a new developed model. According to the assumption and independent incensement of Poisson process, it solves following model indexes: the number of detected faults and removed faults during ( 0,t ]; software failure efficiency; software reliability; the number of the remnant faults after t; mean time to failure (MTTF).3. Base on the above model, a software development cost model is proposed and the optimal software released time is obtained by minimizing the cost function. Finally a numerical example is given to show the model applicability.