Software Reliability: Estimation and Prediction
Abstract
The principal purpose of this research was to provide new tools for measuring probabilities of failure free operation of software systems and to develop methods for predictions of software reliability. Links are established between stochastic models of fault occurrence suggested by Scholtz (1985) and Miller (1986) and an important class of finite population sampling models called 'successive sampling' in the sample survey literature. Successive sampling consists of sampling a finite population of objects, each with an assigned magnitude, proportional to magnitude and without replacement. Recognition of linkages between Schoz and Miller's 'Exponential Order Statistics Models' and successive sampling allows application of an emerging body of research on methods of estimation for successive sampling models to software reliability estimation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 31, 1992
- Accession Number
- ADA268895
Entities
People
- Gordon Kaufman