Method of Empirical Mode Decomposition in Specific Emitter Identification

Abstract

RF fingerprint (RFF) extraction is a fundamental issue of specific emitter identification (SEI). In most of current SEI techniques, the extracted RFFs are adversely affected by other components. This paper introduces a novel SEI system for the common communication signals. Empirical mode decomposition (EMD) method is applied in such system to separate components of a signal, which efficiently mitigates the obscurity problem of RFF. The principal component analysis (PCA) is then used to select the critical feature and a support vector machine (SVM) classifier is finally used to realize the emitter classification. In the experiments, the transient signals captured from four mobile phones and the steady signals from four WLAN cards are utilized to evaluate the performance. The experimental results highlight that the proposed method can achieve high accuracy of recognizing to transmitters.