Research On Individual Radiation Source Identification Based On Fingerprint Feature Extraction

With the development of information technology,Specific emitter identification has gradually become an important research topic in the field of communications.The technology is of great significance in many fields such as electronic warfare,military communications,frequency band protection,malicious device identification,and network security.Although it has obtained many application results in both military and civilian fields,it still faces some challenges.The basis for realizing the identification between individuals is the fingerprint of the radiation source signal,that is,the subtle characteristics of the radiation source individual's internal hardware differences reflected in the emitted signal,which is universal and discriminative.The process of obtaining subtle features through signals is called the feature extraction of individual radiation sources.It is the key and difficult point in individual identification.With the continuous improvement of the production process of electronic components,the differences between individual radiation sources will become more and more difficult to distinguish.In the complex electromagnetic environment,signal acquisition is also facing greater challenges,and the research on the feature extraction technology of the radiation source signal is of urgency.In this paper,starting from the actual problem of collecting radiation source signals in a complex electromagnetic environment,it studies the feature extraction technology of small samples and fragmented radiation source signals.The main work is as follows:First,the background and challenges of the research on fingerprint feature extraction and individual identification of the radiation source signal are analyzed,and several feature extraction methods based on the non-stationary characteristics of actual signals are discussed.Taking the feature extraction method based on bispectrum as an example,the performance of the traditional method in the case of incomplete and fragmented data is analyzed,which shows that the traditional feature extraction method has certain limitations.Then the relationship between the characteristics of the nonlinear system and the chaos theory is studied,combined with the individual differences of the radiation source from the nonlinear characteristics of the internal components of the radiation source,it is demonstrated that the feature extraction method based on the chaos theory is an important method for studying the feature extraction of nonlinear time series.Feasibility,and the characteristics of the nonlinear system,that is,the parameters reflecting the characteristics of the nonlinear system in the phase space,are elaborated.Finally,a feature extraction scheme based on phase space reconstruction is designed,and the characterization effect of the nonlinear characteristics of the signal on the inherent characteristics of the individual radiation source is verified in the experiment,and the feature extraction scheme proposed in this paper is verified through the measured radiation source data.At the same time,the feature extraction experiment was carried out in the case of incomplete and fragmented small data volume.The results show that this scheme has obvious advantages over traditional methods.The feature extraction scheme designed in this paper is used to identify the individual of six mobile devices,and an average recognition rate of 98% is achieved.In the case of small samples and fragmented signal data,this solution will perform feature extraction after splicing the data into fragments,and achieve an average recognition rate of more than 90% under different fragmentation degrees.The recognition effect is double under the same data condition.The average recognition rate of spectral features of 50%-60% is increased by 30%.