Performance Analysis of Feature Sets for Harmonic Radar to Classify Electronic Devices

In this study, we exploit new feature sets to classify electronic devices via harmonic radar. Nonlinear characteristics of the devices are recorded as a response to a frequency-swept signal with constant power. The harmonic characteristics of electronic devices are described by statistical and Fourier feature sets. Contrary to previous studies, statistical and Fourier feature sets are composed of both feature parameters and harmonic information. Exploiting these feature sets provides a good baseline for performance comparison rather than analyzing the classification performance of the harmonics individually. The performance assessment is performed by Monte Carlo simulations.