Research On Radio Frequency Fingerprint Extraction And Recognition Algorithm Based On EMD

The rapid development of wireless communication technology makes people pay more attention to electromagnetic environment monitoring.With the development of large-scale networking monitoring activities,identification of individual signal sources has become an important research topic in wireless communication security.According to the application requirements of signal source equipment identification algorithm in low-end signal acquisition equipment,it is designed in this paper that a set of radio frequency fingerprint extraction and identification algorithm,extracting signal features through optimization algorithm of empirical mode decomposition,and designing classification scheme to improve the accuracy of identification.The main work of this paper is as follows:1.The RF fingerprint characteristics of the transmitter are analyzed.The source of the transmitter's features is explored and analyzed,with the focus on the formation of stable RF fingerprint.The RF fingerprint features used in this paper are modeled and analyzed,and the extraction and identification model of RF fingerprint features is determined.2.An optimization algorithm based on empirical mode decomposition is proposed to extract RF fingerprint features of signal.At first,the empirical mode decomposition method is simulated and analyzed,and experiments are carried out on the basis of measured data.The problems existing in the application of empirical mode decomposition algorithm to actual signals are pointed out.Based on this,an optimization algorithm of empirical mode decomposition algorithm is proposed,including signal preprocessing method and optimization algorithm of signal decomposition.To some extent,the influence of mode aliasing problem on subsequent work is reduced,and the stability of signal decomposition results is improved.The complexity feature of the decomposed signal is extracted,and the complexity of subsequent classification work is reduced through data dimensionality reduction.3.A classification scheme based on combination classifier is designed.Based on the existing classification algorithms,neural network classifier and support vector machine(SVM)classifier are selected to classify the fingerprint feature vectors obtained by RF fingerprint extraction algorithm,and the classification results of different classifiers are analyzed and compared based on which an optimization scheme of classification is proposed.Through data splitting and voting,the accuracy of classification recognition is improved.After the completion of the whole scheme,the simulation analysis of the measured data of the various algorithms proposed in this paper is carried out in stages,and a certain degree of comparative experiments are carried out with the algorithms in the literature.Through the comparative results,the effectiveness of the algorithm in this paper is illustrated.