Parameter estimation of discrete logistic curve models for software reliability assessment

Abstract

We describe software reliability growth models that yield accurate parameter estimates in spite of a small amount of input data in an actual software testing. These models are based on discrete analogs of a logistic curve model. The models are described with two difference equations, one each proposed by Morishita and Hirota. The difference equations have exact solutions. The equations tend to a differential equation on which the logistic curve model is defined when the time interval tends to zero. The exact solutions also tend to the exact solution of the differential equation when the time interval tends to zero. The discrete models conserve the characteristics of the logistic model because the difference equations have exact solutions. Therefore, the proposed models provide accurate parameter estimates, making it possible to predict in the early development phase when software can be released.