Reliability estimation for a software system with sequential independent reviews

Abstract

Suppose that several sequential test and correction cycles have been completed for the purpose of improving the reliability of a given software system. One way to quantify the success of these efforts is to estimate the probability that all faults are found by the end of the last cycle, We describe how to evaluate this probability both prior to and after observing the numbers of faults detected in each cycle and we show when these two evaluations would be the same.