A practical software-reliability measurement framework based on failure data

Software-reliability measurement based on failure data can provide estimations of software reliability at its present state and predictions of software reliability in the future as testing goes on. This paper analyzes the problems existed in the measurement process and proposes a practical software-reliability measurement framework based on state of the art to give relatively better software reliability predictions. Study on the validity of the modification methods i.e. the recalibration technique and the combination method also has been done. The results show that the modification methods could give better prediction results but not always, so predictive quality analysis still needs to be done.