Technical Report
Verification, Validation and Accreditation
-
Systems Engineering and Systems Management Transformation
Report Number: SERC-2011-TR-018-1
Publication Date: 2011-05-03
Project:
Verification, Validation And Accreditation Shortfalls For Models And Simulations
Principal Investigators:
Sue O’Brien
Dr. Russell Peak
Co-Principal Investigators:
As systems become more software intensive and complex, managing their development
and implementation also becomes more complex. Models in development today are
isolated, domain-specific artifacts that are created throughout the design lifecycle. A
mechanism is needed to integrate the design models with simulation environments as
the models are being developed and refined in order to rapidly see the impacts as the
design matures. The ability to perform verification, validation, and accreditation
(VV&A) early in the modeling process and throughout the lifecycle could greatly
improve the model and its contribution. But in order to perform VV&A on complex
systems, a precise language would be required to model these systems in an integrated
fashion to remove ambiguity and the segmented developmental lifecycle.
Objectives of this research included exploring the unique capabilities of Architectural
Analysis and Design Language (AADL) for developing high confidence (verified and
validated) models as part of a system development lifecycle and to determine the
maturity of the AADL tools for VV&A model refinement. To show how AADL could be
used to embed the Verification and Validation of architectural models into the
development process, an architectural model of the Army’s Systems Integration and
Test Laboratory (STIL) was developed and used as a test bed.
The results of the research showed that a portion of a real world DoD representative
system could be modeled using AADL in a very short time with little previous experience
in AADL. AADL’s well-defined semantics supported Architecture/System Design
verification by allowing a precise specification of the architecture so that the analysis
performed is trustworthy and repeatable.