Software reliability modeling with removed errors and compounded-decreased-rate

In this paper, with the compounded rate of removed errors, a new software reliability model is proposed. By the least squares and first-order differential methods, the approximate parameters of the model are obtained. The applicability of proposed model is demonstrated on three well-known real software data sets. The results show that our model has a good capability of fitting and predicting. Moreover, when we own enough history failure data of software, predicting results will be fairly accurate.