Research On Extraction Classification And Identification Of Drone Radio Frequency Fingerprint

Drones first appeared in the 1920 s.Since its inception,it has quickly become popular in many fields due to its advantages of low price,high flexibility and remote control.The highly flexible nature of the drone allows it to enter private places and shoot at will,which causes serious privacy leaks.In addition,drones may be used for terrorist attacks after modification.Due to the low cost of the drone,its attack effect is terrible.In order to prevent the possible security threats from drones,an effective classification and identification system is needed to control the drones.In the large-scale industrial production of wireless devices,due to the limitations of cost and technology,the circuits and devices of the devices inevitably introduce random defects which do not hinder normal usage.Since these defects are randomly generated,they are unique.And these defeats can remain unchanged for a long time and cannot be cloned.The principle of the Radio Frequency(RF)fingerprint is to extract the characteristics from the RF signals transmitted by the device,and use this as the fingerprint of the device.These fingerprints can be used to uniquely classify and identify wireless devices.In this thesis,we use the RF fingerprint technology to uniquely mark each drone by extracting the Differential Constellation Trace Figure(DCTF)feature of RF signal,and use the extracted DCTF feature to classify and identify different drones.The main research work and innovations of this thesis are listed as follows:1.This thesis studies two kinds of signals commonly used in drone communication: the picture transmission signal used to transmit the video captured by the drone and the control signal used to transmit the control information of the drone.The picture transmission signal is an IEEE 802.11 based Wi-Fi signal,which includes a BPSK modulated Beacon frame signal and an OFDM modulated data signal.The control signal is a single carrier frequency modulated signal modulated by Gaussian Filtered Minimum Shift Keying(GMSK).GMSK is a modulation method formed by adding Gaussian low-pass filtering before Minimum Shift Keying(MSK)modulation.By using DCTF process,different RF fingerprint characteristics are extracted and studied.Wi-Fi Beacon frame signal and GMSK frequency hopping control signals are used for RF fingerprint extraction.Aiming at the characteristics of the extracted DCTF,a method based on Haar-like feature DCTF RF fingerprint extraction method is proposed to realize the feature extraction of drone RF fingerprints in the actual wireless environment.After that,a RF fingerprint classification algorithm based on Support Vector Machine(SVM)algorithm is designed.A method based on RF fingerprint for classification and recognition of drones is proposed.The performance of the method was tested under different Signal-Noise Ratio(SNR)conditions.2.This thesis proposes an effective signal detection method using adaptive triangulation threshold method to dynamically acquire the threshold.With the help of this method,the effective signal and the noise signal can be separated,and the valid signal region can be quickly located.This method can reduce invalid calculations and increase the speed of extracting RF fingerprints.At the same time,the method can quickly split a received complete signal file into a plurality of short signal data.Using these shorter signal data to extract RF features can improve the utilization of signal files and provide more samples for subsequent classifier training and recognition.3.Based on the wireless target recognition system software developed by the laboratory and the Universal Software Radio Peripheral(USRP)platform,this thesis designs a complete processing system for the classification and identification of drones.This system can realize the whole process of collecting and identifying the drones by using the USRP to collect the RF signal of the drones,training the RF fingerprint classifier according to the collected drone image transmission and control signals,and using the newly acquired RF signal.