Feature Aided Association Module (FAAM)

In this SBIR project we will develop a mathematically rigorous automated Feature Aided Association Module (FAAM) for accurately combining both kinematic and non-kinematic sensor information in order to create a consistent Single Integrated Air Picture (SIAP) of all aircraft and cruise missile objects. The use of features, or non-kinematic data, will significantly improve the ability of FAAM to create hypotheses that are more likely to contain the correct correlation decisions. The underlying technical mechanism by which FAAM will treat non-kinematic data will be a Bayesian Network based on the taxonomy of air targets and the types of measurements available for estimating the various attributes that characterize the targets. FAAM will leverage a sophisticated Wagner Associates multiple hypothesis architecture that can process high volumes of data within the context of a distributed fusion environment. A key innovation of our approach will be the use of kinematic history as an input feature to the identification/classification estimate of the target. While information such as maximum speed, maximum turn rate, adherence to a given flight profile, etc. does not provide definitive evidence, it can improve the quality of the current (ID/Classification) state estimate, which in turn improves the association scoring.