A Graph-Based Approach to Situation Assessment
Duong Nguyen, David Fisher, Andrew Jeselson, Rebecca J. Bussjager, Ryan Stephens, Carolyn Sheaff · AIAA Infotech@Aerospace 2010 · 2010
We describe a multi-algorithm Situation Awareness and Threat Indication and Warning system with the following components: (a) We use a newly-developed continuous-time Hidden Markov Models (HMMs) with observations. Given a sequence of events, HMMs build a model of underlying ‘hidden’ states. Our approach differs from usual HMM approach in that events can come at varying times with no inherent time-step. It distinguishes between not observing an event because we are not looking for it as opposed to not observing an event because it is not there. (b) Dynamic Bayesian Networks (DBN) allow fusion of information from multiple sources or locations into a multi-tiered model. (c) Fuzzy Cognitive Maps (FCM) integrate and display relations between elements of the developed model of adversarial behaviors, and (d) Case-Based Reasoning (CBR) compares resultant models with previous experiences to issue threat indications and warnings.