Device Classification Performance Modeling Using UWB Stimulated RF-DNA Fingerprinting

Authors: Mathew Lukacs, Peter Collins, Michael Temple

Source: FERMAT, Volume 14, Article 4, Mar.-Apr., 2016

Abstract: Quality control is critical for all industrial processes, but often times defect detection is labor intensive. A novel approach to industrial defect detection is proposed using a Digital Noise Radar (DNR), coupled with Radio Frequency Distinct Native Attribute (RF-DNA) fingerprinting processing algorithms to non-destructively interrogate microwave devices. The DNR is uniquely suitable since it uses an Ultra Wideband noise waveform as an active interrogation method that will not cause damage to sensitive microwave components. Additionally, it has been demonstrated that multiple DNRs can operate simultaneously in close proximity, allowing for significant parallelization of defect detection systems resulting in increased process throughput. Using this method, 100% sampling for quality control may be attainable in many cases. The ability to classify defective units from properly functioning units was demonstrated in [1] with potential applications including assembly line inspection of automotive collision avoidance systems, wireless or cellular antenna defect detection during manufacture, and phased array element defect detection prior to RF system assembly. However, prior research into active interrogation has been strictly empirical. This paper will develop an analytical model and simulation for interrogating a passively terminated antenna with an overall objective of improving classification performance through optimization of the interrogation signal bandwidth.

Index Terms: Noise Radar, Device Classification, RF-DNA, Ultra-Wideband

View PDF

Device Classification Performance Modeling Using UWB Stimulated RF-DNA Fingerprinting