Distance-based approach for software reliability growth model selection

Abstract

This paper proposed an optimized distance-based approach (DBA) for Software Reliability Model (SRM) selection. Reliability of software is important in making design decision. It is useful in software test planning as well as to give estimation or prediction of how long the software can deliver its function after it is delivered to client. However, the accuracy of reliability prediction is still remain an issue in reliability prediction. SRM selection may affect the prediction accuracy and it may prone to human error since one need to have deep mathematical knowledge. There are many works that proposed the methods to select suitable SRM for specific failure data. However, they did not consider applying optimization technique in the model parameters estimation which is also one of the way to improve prediction accuracy. In this paper, parameter of models are estimated using Genetic Algorithm (GA). Initial finding shows that GA are able to minimize deviation from the expected data.