Advanced Algorithms and Architecture for Multimoda

Identity management (IdM) problems arise in many applications, but the accuracy and reliability of IdM technologies in the military and homeland security domains are of the utmost importance. Numerous technologies exist to exploit biometrics for IdM, but such technologies are limited in the number of features they exploit and in their robustness to data issues such as missing, overlapping or spoofed data. Toyon proposes to research and develop an architecture and associated algorithms for multimodal biometric and contextual data fusion across multiple data repositories in a network centric environment. Toyon’s proposed approach provides automated repository searching and identity estimation in a feedback control architecture. The automated search capability will be tightly coupled with the identity estimation algorithms to optimize the collection of data available to identify the person of interest (POI). Toyon will develop and employ a Bayesian Network algorithm for identity estimation along with techniques to automatically detect and rectify data inconsistencies such as identity spoofing and unintentional data entry errors. The overall architecture and algorithms will be flexible to the number and heterogeneity of the available data repositories thereby enabling an enterprise-level solution.