A generalized software reliability process simulation technique and tool

Abstract

The paper describes the structure and rationale of the generalized software reliability process and a set of simulation techniques that may be applied for the purpose of software reliability modeling. These techniques establish a convenient means for studying a realistic, end-to-end software life cycle that includes intricate subprocess interdependencies, multiple defect categories, many factors of influence, and schedule and resource dependencies, subject to only a few fundamental assumptions, such as the independence of causes of failures. The goals of this research are dual: first, to generate data for truly satisfying the simplified assumptions of various existing models for the purpose of studying their comparative merits, and second, to enable these models to extend their merits to a less idealized, more realistic reliability life cycle. This simulation technique has been applied to data from a spacecraft project at the Jet Propulsion Laboratory; results indicate that the simulation technique potentially may lead to more accurate tracking and more timely prediction of software reliability than obtainable from analytic modeling techniques.<>