A NEURAL NETWORK METHODOLOGY FOR SOFTWARE RELIABILITY PREDICTION OF LONG-TERM MTTF

Abstract

Accurate prediction of software MTTF (Mean Time To Failure) is necessary in critical applications. 
This paper presents the prediction of long-term MTTF using neural network. This proposed method is to learn 
a sequence of cumulative observation times and the corresponding observed accumulated failures up to the 
present testing time. The results obtained through proposed method compared with CASRE Computer-Aided 
Software Reliability Estimation) tool. It is observed that information provided by the proposed method is more 
meaningful and relevant as compared to the existing methods.