Signal Processing for Layered Sensing

As a foundational component of a Layered Sensing architecture, Toyon proposes to develop a Multi-Modal Fusion Module (M2FM) capable ofconfirming detections from Chemical, Biological and Radiological (CBR) sensing networks by fusing together information from other (non-CBR)sensors. The module will provide information to assist operators in confirming an initial CBR detection or rejecting it as a false alarm. M2FM willalso aid in characterizing the threat scenario by estimating key parameters for the source: time, location, size and temporal profile of the release.The estimated source term can be an input to existing plume models capable of predicting the extent of the resulting hazard. These results will beprovided as a Decision Aid for guiding tactical response actions.The M2FM will utilize Bayesian-based sensor fusion algorithms, which have been used on similar "evidence accrual"-type problems. The CBRapplication involves several unique features that will require adapting the theory and existing tools (e.g., Bayesian Belief Networks) to emphasizereal-time accrual based on features related to the CBR threat, and to infer the underlying threat properties based on processed sensor outputs.The scope of Phase I effort involves defining, analyzing and demonstrating all aspects of M2FM in more detail.