Realistic assumptions for software reliability models

Abstract

A definition of reliability appropriate for systems containing significant software that includes trustworthiness and is independent of requirements is stated and argued for. The systems addressed encompass the entire product development process as well as both product and its documentation. Cost incurred as a result of faults are shown to be appropriate as a performance measurement for this definition. This and more realistic assumptions are shown to lead to the use of auto-regressive integrated moving average (ARIMA) mathematical models for the modeling of reliability growth.<>