Improved Estimation of System Reliability with Application in Software Development
Beidi Qiang, Edsel A. Peña · 2018
This chapter is concerned with the problem of estimating the reliability of a software system either during the software development stage or during real-time operation. It looks at the situation where white-box testing is used in software performance assessment, which means component-level failure times (or workloads) are observed. The chapter proposes improved estimators of system reliability by applying shrinkage estimators for each of the component reliabilities. It utilizes the structure function to combine these estimators to obtain the system reliability estimator. Some numerical examples with different system structures and component lifetimes are provided. Simulation results are also presented to examine and compare the performance of the different estimators. The statistical inference problem of point estimation of functionals of a parameter falls into the decision-theoretic framework. The chapter demonstrates the advantage of the improved estimator over the standard maximum likelihood estimator (MLE) and PLE using simulated data.