Statistical Analysis of a Debugging Model
Suppose that a system is undergoing a sequence of trials and design changes in an effort to find and eliminate design flaws in the system. Specifically, suppose that each trial is classified as an assignable cause failure (if the failure is due to one or more design flaws), an inherent failure (if the failure is not due to a design flaw), or a success. After each assignable cause failure, the design flaws that caused the failure are removed. In the case of statistically independent and identical design flaws, a probabilistic model is developed to describe this debugging process. Explicit expressions are obtained for the distributions of the important random variables and for the important measures of reliability. Statistical methods are developed for estimating the parameters of the model from accumulated test data.