80 likes | 196 Vues
Explore high-impact research solutions for defense needs, emphasizing dependable software, development process, and technology transition in the 21st century.
E N D
Software Dependability: The Next Step Michael Evangelist Fraunhofer Center for Experimental Software Engineering College Park, MD
August 7, 2001 - Fraunhofer Center for Experimental Software Engineering
Barry’s Questions • Look at software engineering • products • processes (including methods, tools, and engineering principles) • management • technology transition • Focus on high-impact research solutions for high-priority DoD needs August 7, 2001 - Fraunhofer Center for Experimental Software Engineering
High Dependability • Ubiquity of computing and the coming nano revolution demand higher dependability • for defense and commercial systems • Not only dependable software but also • dependable development process • dependable acquisition process • dependable research process August 7, 2001 - Fraunhofer Center for Experimental Software Engineering
High-Dependability Research Model Careful Problem Discovery New Principles and Technologies Rigorous Evaluation and Feedback Framework-Based Technology Transition August 7, 2001 - Fraunhofer Center for Experimental Software Engineering
Example Evaluation Results in High-Dependability Computing • [Basili et al.] Controlled evaluation experiments • comparing code reading vs. functional testing vs. structural testing • on effectiveness of Ada for flight-dynamics software • on effectiveness of Independent Verification & Validation August 7, 2001 - Fraunhofer Center for Experimental Software Engineering
Potential Experimentation on High-Dependability Technologies • Verification: model-checking (Clarke et al.), proof-carrying code (Lee and Necula), ... • Synthesis: generate distributed programs from high-level temporal-logic specifications (Attie et al.) • Self-stabilizing systems: system guaranteed to converge to “good” state under stress (Arora et al.) • .... August 7, 2001 - Fraunhofer Center for Experimental Software Engineering
Summary • In 21st Century, DoD must focus on high-dependability computing • new products, principles, theory • plus high-dependability research model to provide sound framework for research and technology transition • Without improved model, DoD gets less ROI • Research community loses impact August 7, 2001 - Fraunhofer Center for Experimental Software Engineering