A Hybrid Detector Network for Nuclear and Radioactive Threat Detection

Abstract:

Identification and tracking of nuclear materials is a problem of obvious importance, and has attracted a great deal of engineering effort in the recent past. The technical challenges of this problem are compounded by the fact that nuclear materials that are of interest vary widely in their concentrations and compositions, and are often concealed with shielding material, and buried in background radiation that is always present, but varies widely in strength and spectrum over time and space. Finally, there are also many types of detectors and sensors used in practical tracking systems, and these too vary widely in their quality and sensitivity. The overarching objective of our research is to advance the theory and practice of a cyber-­‐physical system that uses a hybrid network of interconnected sensors to monitor a region of space for the presence of radioactive sources, and detect, identify and track individual sources of interest. Our poster focuses on the simple and general estimation problem of finding the location of a nuclear source from radiation measurements. Our objective is to study the effect of the inherent quantum randomness of radioactive emissions on the accuracy to which nuclear sources can be localized. To this end, we consider an ideal mobile detector making perfect, noiseless measurements and formulate a general problem of maximum likelihood estimation of source location using such measurements. For the case of a stationary source and a detector moving with uniform speed in a straight line, we derive solutions to the maximum likelihood location estimate as well as the corresponding Cramer-­‐Rao lower bounds. We present numerical results showing that the maximum likelihood estimates approach the Cramer-­‐Rao bounds, and comment on the implications of these theoretical results with ideal detectors and perfect estimators for the real-­‐world problem of nuclear source localization.