CN-116782236-B - Eye pattern-based radio frequency fingerprint identification method and system

CN116782236BCN 116782236 BCN116782236 BCN 116782236BCN-116782236-B

Abstract

The application provides an eye pattern-based radio frequency fingerprint identification method and system, and relates to the technical field of wireless communication. The method comprises the steps of demodulating a transmitter signal to be identified to achieve carrier separation and obtain a Gaussian baseband waveform of the transmitter signal, calculating an eye diagram UI, intercepting and segmenting the Gaussian baseband waveform by taking the eye diagram UI as a period, superposing all intercepted eye diagram track frames to obtain an eye diagram, calculating probability density of the eye diagram, reversely deducing the probability density of the eye diagram to obtain a BT value of a GMSK signal, differentiating an actually measured signal eye diagram and a modeling eye diagram under the condition of the same BT value to obtain fine differences between the actually measured signal eye diagram and the modeling eye diagram, further extracting statistical features of the eye diagram differences, combining the BT value features to construct a radio frequency fingerprint feature vector, and inputting the radio frequency fingerprint feature vector into a support vector machine to perform identity authentication of equipment. The authentication method can accurately authenticate the identity of different wireless devices with the modulation mode of GMSK by utilizing the hardware slight difference between the different wireless devices.

Inventors

ZHANG JINGBO
ZHENG XIAOHAN
CHEN BING
LIN RONGWEN

Assignees

大连海事大学

Dates

Publication Date: 20260508
Application Date: 20230526

Claims (9)

1. The radio frequency fingerprint identification method based on the eye pattern is characterized by comprising the following steps of: the method comprises the steps of obtaining a transmitter signal to be identified, demodulating the transmitter signal to be identified, and separating carrier waves to obtain a Gaussian baseband waveform of the transmitter signal; Calculating an eye diagram UI, intercepting and segmenting the Gaussian baseband waveform by taking the eye diagram UI as a period, and superposing all intercepted eye diagram track frames to obtain an eye diagram; calculating the probability density of the eye pattern; The BT value of the Gaussian Minimum Shift Keying (GMSK) signal is obtained by means of inverse deduction of probability density of an eye pattern, wherein the BT value is the product of the bandwidth B of a Gaussian filter and the width T of an input code element; Constructing an eye pattern model of the BT value; The actually measured signal eye diagram and the modeling eye diagram under the condition of the same BT value are subjected to difference, statistical characteristics are further extracted from the eye diagram difference, and a radio frequency fingerprint characteristic vector is constructed by combining the BT value characteristics; Inputting the radio frequency fingerprint feature vector to a machine learning algorithm for identity authentication of equipment, wherein the machine learning algorithm is a support vector machine.
2. The eye-based radio frequency fingerprinting method of claim 1, wherein calculating the eye-diagram UI comprises: find time when amplitude of Gaussian baseband waveform is 0 In the ideal case of signal transmission, the bit period is unchanged, the zero crossing point position is a bit sequence jump edge, and the relationship between the bit number of the bit sequence and the corresponding time is linear; calculating the deadline less than each bit Time of data point closest to the time in time range And (3) with Is the time difference of (2) And calculate to be greater than Time of data point closest to the time in time range And (3) with Is the time difference of (2) ; If it is The deadline of the bit is The next bit data point is preceded by 0, otherwise the deadline of the bit is 。
3. The eye-based radio frequency fingerprinting method of claim 1, wherein calculating the probability density of the eye pattern comprises: deriving and inverting the eye diagram track of each state group of the eye diagram to obtain the probability density of the eye diagram track of each state group; The probability density of the eye trace of each state group is multiplied by the number of the state groups and added, and then divided by the number of all state groups, thereby obtaining the probability density of the eye.
4. The method for radio frequency fingerprint identification based on an eye pattern according to claim 1, wherein the step of obtaining BT values of GMSK signals by back-deriving using probability densities of the eye pattern comprises: obtaining the amplitude corresponding to four maxima in the curve by clustering in the probability density curve, wherein the amplitude comprises a large 0 level Small 0 level Small 1 level And a large 1 level ; Respectively to small 0 level And a large 0 level Large 1 level And a small 1 level Substituting the difference into a level difference function related to the BT value, and calculating to obtain a small BT value And large BT value 。
5. The method for eye-based radio frequency fingerprinting as defined in claim 1, wherein constructing an eye model of the BT value comprises: Constructing an eye path mathematical model of each state group: ; Wherein: Is a set of states for an eye pattern, , Is of width of The gaussian impulse response of the input rectangular pulse of (a) is specifically expressed as: ; At the position of B is the 3dB bandwidth of the filter; And superposing all the state group eye patterns to obtain an eye pattern model.
6. The method for eye-based rf fingerprint identification of claim 1 or 4, wherein constructing the rf fingerprint feature vector in combination with the BT value feature and the eye difference feature comprises: acquiring large BT values Small BT value The average of these two BT values ; Obtaining the average value, the maximum value and the minimum value of the eye pattern difference of each state group; and constructing a radio frequency fingerprint feature vector based on the acquired values.
7. The eye-based radio frequency fingerprint identification method of claim 1, wherein inputting the radio frequency fingerprint feature vector into a machine learning algorithm for device identity authentication comprises: The features are randomly divided into a training set and a testing set according to a preset proportion, the training set and the testing set are input into a Support Vector Machine (SVM) for classification, and a label result output by the trained model is compared with an actual label to obtain classification accuracy.
8. The method of claim 1 or 7, further comprising using principal component analysis to reduce the dimension of the RF fingerprint feature vector for visualization.
9. An eye-based radio frequency fingerprinting system, comprising: the baseband waveform acquisition unit is used for acquiring a transmitter signal to be identified, demodulating the transmitter signal to be identified, and separating carriers to obtain a Gaussian baseband waveform of the transmitter signal; The eye diagram acquisition unit is used for segmenting the Gaussian baseband waveform and superposing all the intercepted eye diagram track frames to obtain an eye diagram; The device comprises a BT value acquisition unit, a Gaussian Minimum Shift Keying (GMSK) signal processing unit and a Gaussian filter, wherein the BT value acquisition unit is used for calculating the probability density of an eye diagram and reversely deducing the BT value of the Gaussian Minimum Shift Keying (GMSK) signal by utilizing the probability density of the eye diagram, and the BT value is the product of the bandwidth B of the Gaussian filter and the width T of an input code element; The radio frequency fingerprint feature vector construction unit is used for constructing an eye pattern model of the BT value, making differences between an actually measured signal eye pattern and a modeling eye pattern under the condition of the same BT value, further extracting statistical features from the eye pattern differences, and constructing a radio frequency fingerprint feature vector by combining with the BT value features; the identity authentication unit is used for inputting the radio frequency fingerprint feature vector into a machine learning algorithm to perform identity authentication of equipment, wherein the machine learning algorithm is a support vector machine.

Description

Eye pattern-based radio frequency fingerprint identification method and system Technical Field The invention relates to the technical field of wireless communication, in particular to an eye-diagram-based radio frequency fingerprint identification method and an eye-diagram-based radio frequency fingerprint identification system. Background In recent years, with the rapid development of wireless communication and internet of things technologies, wireless devices have become an integral part of human daily life. Especially, the gradual application of novel Internet of things equipment such as unmanned vehicles, unmanned aerial vehicles and unmanned ships brings great convenience to the life and work of people, and simultaneously, the communication security problems such as the trusted authentication of wireless equipment and the like are more and more required, so that the research on the wireless network security problems becomes more and more critical. Because of the open nature of wireless channels, the communication information of wireless devices becomes very vulnerable to eavesdropping and attack, and ensuring the security of wireless communications has become an urgent issue. Traditional wireless network security authentication schemes often utilize the complexity of encryption algorithms and the privacy of the initial distribution key to ensure the security of data. However, with development of cryptography and computer technology, encryption algorithms are easier to crack, and counterfeiting of hardware addresses and even tampering of various hardware information of devices have already provided mature schemes, so that encryption/decryption algorithms and secure transmission protocols face higher mathematical complexity and huge challenges. In order to solve the above-mentioned security problem, students at home and abroad begin to study how to utilize the physical layer characteristics of wireless equipment to realize the identity authentication of the equipment. The authentication method is to analyze the radio frequency signal of the wireless equipment so as to extract the radio frequency fingerprint of the equipment. The radio frequency fingerprint is physical characteristic information extracted from the wireless signal and can reflect nuances in the signal transmission process of the wireless transmitter. These differences may result from different manufacturing processes of the hardware device, as well as from manufacturing tolerances. These features are fixed when the hardware device leaves the factory, and have the characteristics of uniqueness and difficulty in cloning. Therefore, the wireless device is difficult to be artificially tampered and counterfeited, the possibility of the wireless device being counterfeited is greatly reduced, and the safety and the reliability of the identification system based on the radio frequency fingerprint are improved. An eye diagram (EYE DIAGRAM) is a digital signal pattern superimposed on an oscilloscope from the data levels and edge transitions of multiple symbols. The eye diagram continuously superimposes a new signal waveform over the previous signal waveform while preserving the previous signal waveform. The superimposed pattern shape looks much like an eye, so the eye pattern is named. The method comprises abundant information, can observe the influence of inter-code crosstalk and noise from an eye diagram, and embody the integral characteristics of digital signals, thereby estimating the quality of the system. At present, most of researches on radio frequency fingerprint feature recognition based on eye patterns are that eye patterns are directly input into a convolutional neural network with a relatively simple structure as pictures for automatic recognition, and although related experiments show that the eye patterns have better classification performance, the obtained features generally have no interpretable physical significance, have dependence on specific data sets and have limited generalization capability. Disclosure of Invention According to the prior research on the radio frequency fingerprint feature recognition based on the eye pattern, the eye pattern is mostly converted into pictures to be directly input into CNN for classification recognition, the obtained features usually have no interpretable physical meaning, have the problems of dependence on specific data sets and limited generalization capability, the invention provides a radio frequency fingerprint identification method and a radio frequency fingerprint identification system based on an eye diagram, which are used for deeply analyzing the eye diagram, extracting effective features and inputting the extracted features into machine learning for classification verification, wherein the extracted features have interpretable physical significance and smaller calculation complexity, and can realize effective identification and authentication of Gaussian minimum frequency shift keyin