The SEI is currently conducting research in the area of quality attribute analysis.
Software-reliant systems depend not only on achieving acceptable functionality, they also depend on non-functional properties, which may include security, reliability, interoperability, openness, or agility. When these non-functional properties, or quality attributes, are not addressed sufficiently early in the system-development or evolution life cycle, unacceptable system behavior and delays in fielding can result.
SEI research focuses on the structure and behavior of software-reliant systems and the relationship between structure and quality attributes. Our approach uses a combination of formal notations, quantitative analyses, and qualitative methods to address conflicts among quality attributes that may arise at any phase of system development. The SEI’s work provides foundations for handling multiple critical quality attributes in both dynamic and resource-constrained environments.
Our approach recognizes that systems of increasingly large scale will
- manifest new quality attributes
- require the use of non-traditional theories and analyses for their software design, construction, deployment, adaptation, and evolution; and
- require new technologies and methods to accommodate their scale.
Specific tasks under investigation this year are
- Resource allocation for massively parallel multicore platforms—developing task models, resource abstractions, and scheduling strategies for predicting real-time performance
- Static analysis for multicore—investigating use of scalable static analysis to ensure that concurrency-related invariants are preserved as systems move to multicore platforms
- System reliability framework—developing new metrics and approaches for using architecture knowledge to assure the safe and reliable operation of software-reliant systems