Distributed Decision Architecture for Adaptive, Intelligent BMD Information Fusion

"Defense against Ballistic missiles is a complicated problem domain that entails multi-dimensional considerations, not only in the space-time domain such as threat kinematics, but also constraints such as threat type and classification and weaponeffectiveness. Other dimensions include "physics based" modeling and considerations of sensor management and tasking. Beyond this BM defense should also consider non-physics based "epistemological" information such as enemy dispositions, plans, andintentions. The JDL fusion model defines four levels of fusion: level 1 - Object refinement, level 2 - Situation refinement, level 3 - Threat Refinement, and level 4 - Process refinement. Within the BMD domain, the different problem and solution spacesspanned by the JDL fusion levels require data and information of apparently vastly different types. Heretofore, the prototype techniques to solve these problems have used disparate representational and processing techniques such as state vectors andKalman filters for tracking, Bayesian fusion for discrimination, and knowledge based rules, frames, and "fuzzy" reasoning for modeling group behaviors in order to infer enemy dispositions and courses of action. Partly reflective of the relative immaturityof fusion levels 2-3, extremely different and specialized point-designs have been devised to address the different levels of fusion. A consideration of the high level functional decomposition of BMD f