Fractal segmentation of signal from noise for radio transmitter fingerprinting

This paper presents a study of segmentation between channel noise and turn-on transient for radio transmitter fingerprinting. The segmentation is based on a multifractal measure, which is called variance fractal dimension trajectory. Previous studies have shown that this multifractal measure is effective in nonstationary signal characterization. We employ two techniques for the variance fractal dimension implementation and evaluate them in terms of noise segmentation. An optimal window size for the fingerprinting is presented. We also present the Renyi dimension trajectory based on generalized entropy as an alternative of the variance fractal dimension measure for noise segmentation