Classification of Electronic Devices With Power-Swept Signals Using Harmonic Radar

Abstract

A novel approach for classification of electronic devices using a harmonic radar is proposed in this article. We transmit single-tone time-varying signals with varying power to the electronic circuits under test (ECUTs). The novelty of the proposed approach stems from representing the harmonic response of the ECUT in the harmonic and statistical feature spaces. Besides, as a new approach, we sweep the transmitting power to capture the nonlinear behavior relevant to nonlinear I - V characteristics of the ECUT. Furthermore, we utilize both the time and frequency content of the received harmonic response. Statistical and Fourier domain features of the received power levels at harmonic frequencies are determined and used for classification. k-nearest neighbors is chosen as the classification method. We demonstrate the effectiveness of our method in the presence of noise through Monte Carlo simulations with different signal-to-noise ratio values. Extensive simulation studies show that our new approach of power-swept signals and new features we have developed are very effective at classifying nonlinear devices. © 1965-2011 IEEE.