Distributed Battle Management Techniques

In the future Ballistic Missile Defense system, it will be necessary to coordinate the operations of many geographically dispersed system components to create an architecture with the best chance for defeating the threat. Though challenging, it is desiredthat this architecture be implemented in a distributed manner to avoid the need for a centralized controller to which all sensors transmit all of their information for processing and decision-making. In the research effort proposed here, the DecisiveAnalytics Corporation Team will implement a distributed C2BMC inference architecture using a framework called Multiply-Sectioned Bayesian Networks. Using this approach, which is based on recent advances in the artificial intelligence literature, it ispossible to decompose a unified Bayesian Network into subcomponents that operate at different locations, while exchanging data to maintain global consistency. Our implementation will employ the actual system models under development for Project Hercules,and we will assess computational performance with respect to various implementations of these cases.This background research will result in the development of a distributed multi-source fusion engine for inference. Potential application includes use in military command and control systems, and other decision support systems that involve time-criticaldecision-making under uncertainty, including crisis monitoring, and medical applications.