Specific emitter identification based on normalized frequency spectrum

Abstract

Specific emitter identification (SEI) refers to the capability to associate a received pulse waveform with a unique emitter. It can be used to identify illegal emitters and detect camouflaging attackers. As communication security become increasingly important and frequency hopping technique is widely used, a SEI method for frequency hopping signals is meaningful. In this paper, a novel SEI method based on normalized frequency spectrum features is proposed. Box-counting fractal dimension is used to extract the transient signal. Principal component analysis is used to extract radio frequency fingerprint by analyzing the transient signals' normalized frequency spectrum. Signals from four mobile phones are collected and a support vector machine classifier is created for classification. Experimental results demonstrate that the proposed method is effective.