Fractal feature extraction of radar emitter signals

In this paper, a novel effective approach based on fractal theory is proposed to extract features of radar emitter signals. Box dimension and information dimension are used as classification features to recognize the types of intra-pulse modulation of radar signals. It is proved in both theory and experiments that the features are not sensitive to noise. Experimental results show that the features have good characteristics of clustering the same radar signals and separating different radar signals.