MODELING OF QR SENSORS FOR OPTIMIZED EXPLOSIVESDETECTION

摘要

Quadrupole Resonance (QR) sensors have the unique capabilityof detecting explosives with remarkably high detection rates and lownumber of false alarms. The sensitivity of a QR-based sensor in inductivedetection can be assessed in terms of the signal-to-noise ratio (SNR), whichdetermines the Receiver Operating Characteristics (ROC) curves of thedetector and provides a fundamental limitation to the performance of theQR explosive detection system. The main goal of the QR sensor design is,therefore, to maximize the SNR to achieve the highest possible detectionperformance with the lowest number of nuisances.This paper describes part of the work performed at Quantum MagneticsInc. to model the characteristics of the QR signal detection process, whichincludes the receiver's response, data processing, and signal detectionalgorithm. These theoretical and numerical models allow us to predict ROCcurves for the QR detector and evaluate trade-offs between experimentalparameters, sample characteristics, data processing, and hardware featuresof the detector. Numerical and experimental results are presented to validateour models, and demonstrate their usefulness for QR sensor design andoptimisation.