The number of very large-scale identity management (IDM) systems requiring multimodal biometric enrollment is rapidly increasing. Within USG alone, systems such as FBI / IAFIS, DHS / IDENT, and DOD / ABIS collect some combination of finger, face, & iris, but these systems are currently limited in their ability to intelligently exploit these multiple biometrics. Very large Federal procurements are in the works (such as FBI / NGI) to expand multimodal capabilities, and as a result there is an immediate and growing need to fill an existing gap in technical knowledge and standards.

The integration of multimodal biometrics provides enhanced capabilities such as: higher accuracy, very large scalability, opportunistic acquisition, compensation for low quality samples, ability to handle missing biometrics, integration of new modalities, enrollment consolidation /de-duplication, and interoperability. While these benefits are anticipated and generally accepted, much is yet to be known about how to architect very large-scale multimodal biometric 1-to-many IDM systems. New standard methods for evaluation are needed to assess the effectiveness and efficiency of these systems, once engineered. With potential for very large scalability, new standard methods for performance assessment must be developed. Testing methods that are applied today at the level of 100,000 probes matched to 1,000,000 enrollment records may/will not work with enrollment databases anticipated to grow to 100s of millions approaching "global" populations of more than a billion people.