Verification, Validation & Accreditation (VV&A)
EBR is engaged in the process of Verification, Validation and Accreditation (VV&A) of Human, Social, Cultural and Behavioral (HSCB) simulations and system dynamics models that contribute to policy development and military decision making. Today, these models support a challenging national security environment characterized by political instability, insurgency and asymmetric warfare. Our VV&A process is targeted at mitigating the risk associated with using such complex models and building users’ confidence and trust in the results generated by the model.
- Verification checks that the model and its data accurately represent the conceptual description and specifications, i.e. the model does what it was designed to do.
- Validation checks the model’s accuracy in terms of representing the real world from the perspective of its intended i.e. the results are realistic .
- Accreditation is an official determination that the application and its associated data are acceptable for use for a specific purpose, i.e. the model is a viable tool.
While all of these components are important, the one that is most challenging is validation. One of EBR’s tenets of success regarding VV&A is that validation itself is a complex endeavor. This complexity means it is not realistic to expect to validate or accredit a model generically for all possible uses across the enterprise. It is critical to define the scope of the problems to which the model will be applied, which in turn, bounds the VV&A process. In other words, the validation of models must be closely coupled to the analysis being performed. This principle of intended use reduces (or eliminates) the validation challenge to one that is more manageable and computable.
EBR is currently doing research on validation of large HSCB models. We are investigating methods for segmenting the validation space into pieces that reflect the use of the model over a portion of the possible domain. A part of this segmentation process addresses the computability challenge associated with validation over that segment. The goal of our research is to develop techniques that minimize the computational challenge on decomposed segments of the model’s domain. However, the decomposition is done in a way that will allow us to extend the results back to a much larger and more complex domain space.