Safety-critical, performance-critical, and mission-critical systems have become increasingly software-reliant. The cost of developing such systems has increased exponentially under the current practice of "build then test" and has become unaffordable, reaching $10B for the next generation aircraft, for example. Eighty percent of faults introduced during development are currently not caught until integration/acceptance testing and actual operation and repaired at a cost factors as high as 110.
Architectural models with well-defined semantics that include the interactions between the physical mission system, the computer system, and the embedded application software, support discovery of system-level problems through predictive model-based analysis of critical non-functional properties early in and throughout development to complement and refocus testing. Industrial proof-of-concept case studies have shown that this model-based engineering approach can dramatically reduce the cost and increase confidence in that expected system behavior will be met.
This white paper describes the Architecture Analysis and Design Language (AADL), an industry standard for modeling and analyzing the architecture of the operational software, the computer system, and the mission system of safety-critical, performance-critical, and mission-critical software-reliant systems.