Specific Emitter Identification Research Based On Radio Frequency Fingerprint

With the rapid development of signal processing technology,more and more signal modulation methods,more and more radio frequency equipments,the electromagnetic environment is far more complex.Traditional modulation recognition technology has been difficult to meet the individual identification needs of emitters.Whereas the specific emitter identification technology based on radio frequency fingerprint could solve the identification problem of the same type,the same model,even the same batch of emitters by utilizing the characteristics caused by the difference of different transmitter hardware.It is one of the research hotspots in this field and has a broad application prospects.This paper starts with the fingerprint formation mechanism of emitters signal,and studies the signal fingerprints brought by the important hardware such as power amplifier,modulator,frequency source and the whole transmitter,then gives the corresponding signal fingerprint definition and extraction method,lastly identifies emitters utilizing machine learning.The main work is as follows:The behavior model of power amplifier with no memory effect and memory effect is established respectively.On this basis,a method for individual identification of emitters based on power amplifier nonlinearity is proposed for narrowband signal conditions.The method does not need to know the drive signal of power amplifier in advance,which reconstructs the drive signal directly according to the received signal,then estimates the power amplifier model coefficients as the signal fingerprint,lastly uses the support vector machine to complete the individual identification.The experimental platform was built to verify the method.The experiment was carried out on three power amplifiers of the same batch of the same model.The results show that the recognition accuracy is higher than 90% when the received signal-to-noise ratio is greater than 11 dB.The proposed method can effectively complete individual identification.As for wideband signal,the characteristics of power amplifier show memory effect.Thus,a method for individual identification of emitters based on power amplifier predistortion error is presented.This method is based on the predistortion and power amplifier structure widely used in mobile communication networks.This paper extracts characteristics of the power amplifier predistortion error as the signal fingerprint,and the individual identification is performed by the neural network classifier.The simulation results show that there are significant differences among different predistortion structures.This method can extract stable signal fingerprints and effectively identify emitters.Aiming at the phenomenon that the distortion of the transmitter causes the distortion of the constellation and carrier frequency offset causes the rotation of the constellation,a WLAN device identification method is proposed.Based on the analysis of the IEEE 802.11a/g protocol,the method extracts the carrier frequency deviation and error vector magnitude of the received signal as the signal fingerprint,and the individual identification is completed by the neural network classifier.The IEEE 802.11a/g device individual recognition experimental platform was built by using the USRP B210 and GNU Radio.The experimental results show that the proposed method has a recognition rate of more than 90% for wireless devices,which has achieved good recognition results.In view of the existing research on the overall characteristics of the transmitter,the research under the theoretical analysis and experimental conditions is lack of practical application.A multi-channel WLAN device identification method based on channel state information processing is studied.Based on the in-depth analysis of the channel state information,the method selects the unwrapped phase as the signal fingerprint and performs individual identification through the neural network classifier.The channel state information tool is used to build an IEEE 802.11 n device individual recognition experimental platform.The experimental results show that the proposed method has a recognition rate of more than 80% for multi-channel wireless devices.