Diagrams and Languages for Model-Based Software Engineering of Embedded Systems: UML and AADL
Abstract
Even after years of research and practice in computer science and, in particular, in software engineering, software projects are still largely risky and unpredictable. There is significant evidence to support this observation. Consider, for instance, a NIST (National Institute of Standards and Technology) study in 2002 that found software errors cost the U.S. economy $59.5 billion annually, about 0.6 percent of the national gross domestic product [NIST 2002]. Based on that total, software users and developers pay more because of error-ridden software than gamblers do at the slot-machines and tables in Las Vegas, Atlantic City, and all other commercial venues that provide gaming. Gamblers accept the risk associated with the roll of a dice; software users should not have to. It is not that developers do not apply resources to discovering and fixing errors. They do. The same NIST study reported that nearly 80% of the money spent in development goes to correcting defects. Yet, software, unlike almost any other product, is provided to customers with a high-level of errors.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 2007
- Accession Number
- ADA633428
Entities
People
- Dionisio de Niz
Organizations
- Carnegie Mellon University