Predicting software reliability using an imperfect debugging Jelinski Moranda Non-homogeneous Poisson Process model

Abstract

Earlier software reliability models have been criticised for their unrealistic assumption of perfect debugging and that they give their answers far too late. Although several software reliability models that incorporate imperfect debugging have been presented during the last several years only few attempts have been made to predict further into the future. These attempts were confined to models that assume perfect debugging. In the present study the imperfect debugging variant of the JMNHPP model is used to obtain one and further step ahead predictions of the reliability of a program. The preliminary results of the study are encouraging. When imperfect debugging is incorporated in the JMNHPP model, predictions can be obtained much earlier in the development and are reasonably consistent with their one step ahead predictions.