Date of Award


Document Type


Degree Name

Master of Science

First Advisor

Daniel Ferens, PhD

Second Advisor

Maj. James Skinner

Third Advisor

Maj. Terry Adler


This research explores early life cycle software reliability prediction models or techniques to predict the reliability of software prior to writing code, and a method for increasing or improving the reliability of software products early in the development life cycle. Five prediction models and two development techniques are examined. Each model is statically analyzed in terms of availability of data early in the life cycle, ease of data collection, and whether data is currently collected. One model and the two techniques satisfied those requirements and are further analyzed for their ability to predict or improve software reliability. While the researchers offer no significant statistical results of the model's ability to predict software reliability, important conclusions are drawn about the cost and time savings of using inspections as a means of improving software reliability. The results indicate that the current software development paradigm needs to be changed to use the Cleanroom Software Development Process for fixture software development. This proactive approach to developing reliable software saves development and testing costs. One obvious benefit of this research is that cost savings realized earlier in the software development cycle have a dramatic effect on making software development practices better and more efficient.

AFIT Designator


DTIC Accession Number



Co-authored thesis presented to the Faculty of the Graduate School of Logistics and Acquisitions Management.