An eye pattern based radio frequency fingerprinting method and system

Abstract

The application provides a kind of based on eye diagram radio frequency fingerprint identification method and system, it is related to wireless communication technical field.The method comprises: the demodulation processing of transmitter signal to be identified, the separation of carrier is realized, the Gaussian baseband waveform of transmitter signal is obtained;Calculate eye diagram UI, with eye diagram UI as period to the Gaussian baseband waveform is intercepted section, all eye diagram trajectory frames intercepted are superimposed to obtain eye diagram;Calculate the probability density of eye diagram;The BT value of GMSK signal is obtained by backstepping using the probability density of eye diagram;The difference between the measured signal eye diagram and modeling eye diagram under the condition of same BT value is obtained, the subtle difference between them is obtained, the statistical characteristics of eye diagram difference are further extracted, and the radio frequency fingerprint feature vector is constructed by combining BT value characteristics;Radio frequency fingerprint feature vector is input to support vector machine to do the identity authentication of equipment.The authentication method of the application can utilize the hardware subtle difference between different wireless devices of modulation mode GMSK to accurately identify the identity of the same.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and more particularly to a radio frequency fingerprint recognition method and system based on eye diagrams. Background Technology

[0002] In recent years, with the rapid development of wireless communication and Internet of Things (IoT) technologies, wireless devices have become an indispensable part of daily life. The gradual application of new IoT devices such as unmanned vehicles, drones, and unmanned ships has brought great convenience to people's lives and work, but it has also placed increasingly higher demands on communication security issues such as trusted authentication of wireless devices. Therefore, research on wireless network security has become increasingly crucial. Due to the open nature of wireless channels, the communication information of wireless devices is easily eavesdropped on and attacked, making the security of wireless communication a top priority. Traditional wireless network security authentication schemes often rely on the complexity of encryption algorithms and the secrecy of the initial distribution key to ensure data security. However, with the development of cryptography and computer technology, cracking encryption algorithms has become increasingly easy. Forgery of hardware addresses and even tampering with various hardware information of devices have mature solutions, posing higher mathematical complexity and significant challenges to encryption / decryption algorithms and secure transmission protocols. To address these security issues, scholars both domestically and internationally have begun to study how to utilize the physical layer characteristics of wireless devices to achieve device authentication. This authentication method involves analyzing the radio frequency signals of wireless devices to extract their radio frequency fingerprints. Radio frequency (RF) fingerprints are physical feature information extracted from wireless signals, reflecting subtle differences in the signal transmission process of a wireless transmitter. These differences may stem from different manufacturing processes of the hardware or from manufacturing tolerances. These features are fixed at the time the hardware leaves the factory, possessing uniqueness and being difficult to clone. Therefore, they are very difficult to tamper with or forge, greatly reducing the possibility of wireless devices being counterfeited and improving the security and reliability of RF fingerprint-based identification systems.

[0003] An eye diagram is a digital signal graph created by superimposing the data levels and edge transitions of multiple symbols onto an oscilloscope. While preserving the previous signal waveform, the eye diagram continuously superimposes new signal waveforms onto the earlier ones. The resulting graph resembles an eye, hence the name. It contains a wealth of information; the effects of inter-symbol interference and noise can be observed from the eye diagram, reflecting the overall characteristics of the digital signal and thus allowing for an estimation of the system's performance.

[0004] Current research on RF fingerprint feature recognition based on eye diagrams mostly involves directly inputting the eye diagram as an image into a relatively simple convolutional neural network for automatic recognition. Although related experiments show that it has good classification performance, the obtained features usually do not have interpretable physical meaning and are dependent on specific datasets, resulting in limited generalization ability. Summary of the Invention

[0005] Most existing research on RF fingerprint feature recognition based on eye diagrams converts the eye diagram into an image and directly inputs it into a CNN for classification and recognition. The resulting features usually lack interpretable physical meaning, are dependent on specific datasets, and have limited generalization ability. This invention provides an RF fingerprint recognition method and system based on eye diagrams. It conducts in-depth analysis of the eye diagram itself, extracts effective features, and then inputs them into machine learning for classification and verification. The extracted features have interpretable physical meaning and low computational complexity, enabling effective recognition and authentication of Gaussian minimum frequency shift keying signals in practical applications.

[0006] Therefore, the present invention adopts the following technical solution:

[0007] This invention provides a radio frequency fingerprint recognition method based on eye diagrams, comprising the following steps:

[0008] The transmitter signal to be identified is acquired, and the transmitter signal to be identified is demodulated to separate the carrier wave and obtain the Gaussian band waveform of the transmitter signal.

[0009] Calculate the eye diagram UI, segment the Gaussian band waveform with the eye diagram UI as the period, and superimpose all the segmented eye diagram trajectory frames to obtain the eye diagram;

[0010] Calculate the probability density of the eye diagram;

[0011] The BT value of the GMSK signal is obtained by inversely calculating the probability density of the eye diagram;

[0012] Construct an eye diagram model of the BT value;

[0013] The difference between the measured signal and the modeled eye diagram under the same BT value is calculated, and statistical features are further extracted from the differences in the eye diagram. The BT value features are then combined to construct an RF fingerprint feature vector.

[0014] The radio frequency fingerprint feature vector is input into a machine learning algorithm for device authentication.

[0015] Further, the eye diagram UI is calculated, including:

[0016] Find the time t when the amplitude of the Gaussian band waveform is 0. nIn the ideal case of signal transmission, the bit period remains unchanged, the position of the zero crossing should be the bit sequence transition edge, and the relationship between the number of bits in the bit sequence and the corresponding time is linearly related.

[0017] Calculate the cutoff time t less than each bit end The time index of the nearest data point within the time range. l With t end Time difference Δt l And calculations for values ​​greater than t end The time index of the nearest data point within the time range. r With t end Time difference Δt r ;

[0018] If Δt l <Δt r The cutoff time for this bit code length is t. e (index r If the next data bit is not specified, a 0 is added before it; otherwise, the cutoff time for that bit's code length is t. e (index l ).

[0019] Further, the probability density of the eye diagram is calculated, including:

[0020] The probability density of the eye diagram trajectory for each state group is obtained by taking the derivative and then the inverse of the derivative.

[0021] The probability density of the eye diagram trajectory is obtained by multiplying the probability density of each state group by the number of state groups, summing the results, and then dividing by the total number of state groups.

[0022] Furthermore, the BT value of the GMSK signal is obtained by inversely calculating the probability density of the eye diagram, including:

[0023] Clustering is used to obtain the amplitudes corresponding to the four maxima in the probability density curve, including: Level 0 (highest level 0). max Level 0 (small 0 level) min Level 1 min Level 1 and higher max ;

[0024] Level 0 (small 0 level) min and Level 0 max Level 1 max and Level 1 min Substituting the difference into the difference function for BT values, we can calculate the smaller BT value BT.min And large BT value BT max .

[0025] Further, constructing an eye diagram model for the BT value includes:

[0026] Construct a mathematical model for the eye diagram trajectory for each state group:

[0027] Y(t)=[A1G(t),A2G(t),...,A8G(t)] T ;

[0028] Among them: A i For the state group of the eye diagram, G(t) = [g(t+T)] b ),g(t),g(tT b )] T g(t) is a function with width T b The Gaussian impulse response of the input rectangular pulse is specifically expressed as:

[0029]

[0030] In g(t), B is the 3dB bandwidth of the filter.

[0031] The eye diagram model is obtained by superimposing the eye diagram trajectories of all state groups.

[0032] Furthermore, an RF fingerprint feature vector is constructed by combining BT value features and eye diagram difference features, including:

[0033] Get large BT value BT min Small BT value BT max And the mean of these two BT values, BT mean ;

[0034] Obtain the mean, maximum, and minimum values ​​of the eye diagram differences for each state group;

[0035] Construct an RF fingerprint feature vector based on the acquired values.

[0036] Furthermore, the machine learning algorithm is a support vector machine.

[0037] Furthermore, the radio frequency fingerprint feature vector is input into a machine learning algorithm for device authentication, including:

[0038] Features are randomly divided into training and test sets according to a preset ratio and then input into a support vector machine (SVM) for classification. The label results output by the trained model are compared with the actual labels to obtain the classification accuracy.

[0039] Furthermore, it also includes: using principal component analysis to reduce the dimensionality of the radio frequency fingerprint feature vector and then visualizing it.

[0040] The present invention also provides an eye diagram-based radio frequency fingerprint recognition system, comprising:

[0041] The baseband waveform acquisition unit is used to acquire the transmitter signal to be identified, demodulate the transmitter signal to be identified, separate the carrier, and obtain the Gaussian band waveform of the transmitter signal.

[0042] The eye diagram acquisition unit is used to segment the Gaussian band waveform and superimpose all the captured eye diagram trajectory frames to obtain an eye diagram;

[0043] The BT value acquisition unit is used to calculate the probability density of the eye diagram and use the probability density of the eye diagram to inversely deduce the BT value of the GMSK signal.

[0044] The radio frequency fingerprint feature vector construction unit is used to construct an eye diagram model of the BT value, subtract the measured signal and the modeled eye diagram under the same BT value conditions, further extract statistical features from the eye diagram differences, and construct the radio frequency fingerprint feature vector by combining the BT value features.

[0045] The identity authentication unit is used to input the radio frequency fingerprint feature vector into a machine learning algorithm for device identity authentication.

[0046] Compared with the prior art, the present invention has the following advantages:

[0047] This invention proposes a frame folding synchronization algorithm to calculate the eye diagram UI, thereby reducing the difficulty of feature extraction caused by the superposition of eye diagram trajectory errors. The eye diagram is analyzed in depth, and a probability density function-based algorithm for inversely calculating the BT value is designed. This algorithm is used to calculate the BT value of a given eye diagram. Based on this, the subtle differences between the measured signal and the modeled eye diagram under the same BT value conditions are utilized to further extract statistical features from the eye diagram differences and combine them with the BT value as the feature set of the radio frequency fingerprint. Experiments show that this invention can effectively identify different devices using GMSK modulation with a recognition accuracy of 99.49%. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a flowchart of an eye diagram-based radio frequency fingerprint recognition method according to an embodiment of the present invention;

[0050] Figure 2 This is the radio frequency fingerprint recognition execution process in an embodiment of the present invention;

[0051] Figure 3 This refers to the recognition accuracy achievable in the embodiments of the present invention;

[0052] Figure 4 This is a three-dimensional visualization of the features extracted in the embodiments of the present invention. Detailed Implementation

[0053] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0054] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0055] This invention provides an eye diagram-based radio frequency fingerprinting method. It demodulates the received Gaussian Filtered Minimum Shift Keying (GMSK) signal to generate an eye diagram. The eye diagram is then analyzed in depth, and the Bit-Shift (BT) value is extracted as a set of features for the radio frequency fingerprint. An eye diagram model with the same BT value is constructed. Utilizing the subtle differences between the measured signal eye diagram and the modeled eye diagram under the same BT value conditions, statistical features are further extracted from the eye diagram differences. These BT value features are combined to construct a feature vector, which serves as the radio frequency fingerprint feature of the device. Finally, the device is classified, i.e., identified based on these features.

[0056] like Figure 1 As shown, this embodiment of the invention provides a radio frequency fingerprint recognition method based on eye diagrams, including the following steps:

[0057] S1. Acquire the transmitter signal to be identified, demodulate the transmitter signal to be identified, separate the carrier, and obtain the Gaussian band waveform of the transmitter signal;

[0058] S2. Calculate the eye diagram UI. Segment the Gaussian band waveform with the eye diagram UI as the period. Superimpose all the segmented eye diagram trajectory frames to obtain the eye diagram.

[0059] In an eye diagram, the interval between one eye intersection and the next adjacent eye intersection is a unit interval (UI).

[0060] The calculation of the eye diagram UI in S2 includes the following steps:

[0061] Find the time t when the amplitude of the Gaussian band waveform is 0. n In ideal signal transmission conditions, the bit period remains constant, the zero-crossing point should be the transition edge of the bit sequence, and the relationship between the number of bits in the bit sequence and its corresponding time is linear:

[0062] t n = a + b × n;

[0063] Where n is the number of bits in the bit sequence that is related to t n The corresponding number of bits, a is the time when the amplitude of the first bit in the frame message is 0; b is the bit period of the digital message, which is also a preliminary estimate of the eye diagram UI.

[0064] Calculate the values ​​for t less than and greater than t respectively. end The time of the nearest data point within the time range and t end Time difference Δt l , Δt r :

[0065] Δt l =t end (i)-t e (index l );

[0066] Δt r =t e (index r )-t end (i);

[0067] Among them, t end For the deadline of each bit, index l For less than t end Time range from t end The location of the most recent data point, index r For greater than t endTime range from t end The location of the most recent data point, t e The time corresponding to the Gaussian band waveform;

[0068] If Δt l <Δt r The cutoff time for this bit code length is t. e (index r If the next data bit is not specified, a 0 is added before it; otherwise, the cutoff time for that bit's code length is t. e (index l The data point is appended with a 0 to ensure that the number of data points in each frame of the eye diagram trajectory obtained by segmentation remains consistent.

[0069] S3. Calculate the probability density of the eye diagram;

[0070] The calculation of the probability density of the eye diagram in S3 includes the following steps:

[0071] For each state group A of the eye diagram i Eye trajectory y i (t) Taking the derivative and inverse, we obtain the probability density of the eye diagram trajectory for each state group; where, for each state group A i The row matrix is ​​as follows:

[0072]

[0073] Each state group A i The probability density of the eye diagram trajectory is related to the number of state groups N. i After multiplying and adding the results, divide by the total number of state groups to obtain the probability density formula for the eye diagram, as follows:

[0074]

[0075] in:

[0076] y∈[-1,1];

[0077] S4. Use the probability density of the eye diagram to inversely deduce the BT value of the GMSK signal;

[0078] GMSK modulation is a digital modulation method developed from MSK modulation. Its characteristic is that a pre-modulated Gauss filter is used before the MSK modulator to reduce the switching energy when switching between two carriers of different frequencies, so that the channel spacing can be made closer at the same data transmission rate.

[0079] The BT value is the product of the 3dB bandwidth B of the Gaussian filter and the input symbol width T, and is a key parameter for designing Gaussian filters.

[0080] S5. Subtract the measured signal eye diagram from the modeled eye diagram under the same BT value conditions to obtain the subtle differences between the two. Further extract statistical features from the eye diagram differences and construct an RF fingerprint feature vector by combining the BT value features.

[0081] Steps S4 and S5 are mainly used to construct an eye diagram model based on BT values, and then obtain the measured signal eye diagram and the modeled eye diagram. Figure 2 We extract features from the subtle differences between individuals to construct feature vectors, specifically:

[0082] First, clustering is used to obtain the amplitudes corresponding to the four maxima in the probability density curve. These four amplitudes are ordered from smallest to largest as follows: Level 0 (largest 0). max Level 0 (small 0 level) min Level 1 min Level 1 max ;

[0083] Next, we will analyze the Level 0 level. min and Level 0 max Level 1 max and Level 1 min Substituting the difference into the difference function for BT values, we obtain two BT values, which, in ascending order, are: the smaller BT value BT... min Large BT value BT max The function for calculating the level difference with respect to the BT value is as follows:

[0084]

[0085]

[0086] Then, based on the given BT value, the formula of the eye diagram trajectory mathematical model is substituted into it to construct eye diagram trajectory mathematical models with the same BT value. After superposition, the eye diagram model is obtained. The formula of the eye diagram trajectory mathematical model is as follows:

[0087] Y(t)=[A1G(t),A2G(t),...,A8G(t)] T ;

[0088] Where: G(t) = [g(t+T)] b ),g(t),g(tT b )] T g(t) is a function with width T b The Gaussian impulse response of the input rectangular pulse is specifically expressed as:

[0089]

[0090] In g(t), B is the 3dB bandwidth of the filter.

[0091] Finally, the measured signal eye diagram and the modeled eye diagram under the same BT value conditions are subtracted, and the mean, maximum, and minimum values ​​of the eye diagram differences are extracted and combined with the large BT value. min Small BT value BT max The mean of these two BT values ​​BT mean A total of 27 statistical dimensions are used to construct the radio frequency fingerprint feature vector.

[0092] S6. Input the radio frequency fingerprint feature vector into the support vector machine for device authentication.

[0093] Specifically, this step involves randomly dividing the features into training and test sets according to a preset ratio, and then inputting them into a Support Vector Machine (SVM) for classification. The label results output by the trained model are compared with the actual labels to obtain the classification accuracy. Principal component analysis is used to reduce the dimensionality of the RF fingerprint feature vectors and visualize them.

[0094] This invention provides an in-depth analysis of the eye diagram itself, first extracting effective features and then inputting them into machine learning for classification and verification. The extracted features have interpretable physical meaning and low computational complexity, enabling the identification and authentication of GMSK signals in practical applications. Experiments show that this invention can effectively identify different devices using GMSK modulation with a 99.49% accuracy rate.

[0095] The following specific application examples will further illustrate the solution and effects of the present invention.

[0096] The eye-map-based radio frequency fingerprint recognition method provided in this embodiment includes:

[0097] 1) Obtain the transmitter signal to be identified, demodulate the transmitter signal to be identified, separate the carrier, and obtain the Gaussian band waveform of the transmitter signal.

[0098] In this embodiment, GMSK signals from seven different AIS devices were selected as the target signals, with each device collecting 200 data points. Since the AIS operates in the marine VHF band, the International Telecommunication Union (ITU) designated 161.975MHz (channel 87B) and 162.025MHz (channel 88B) as AIS channels at the 1997 Radiographic Conference. According to the Nyquist theorem, a sampling rate of 324.05 Msps is required to acquire the AIS signal. The actual signal selected in this embodiment was directly captured by a Tektronix real-time spectrum analyzer for a duration of 0.06 seconds. However, since the Tektronix real-time spectrum analyzer contains a local oscillator and a compression filter, its maximum sampling rate is 56MHz. Therefore, according to the bandpass Nyquist sampling theorem, the bandwidth of the local oscillator and compression filter should be determined by the center frequency and bandwidth, thus determining the sampling rate Fs. Since the AIS bandwidth is approximately 25kHz or 12.5kHz, the bandwidth is adjusted to 40MHz. The sampling rate to bandwidth ratio in the Tektronix real-time spectrum analyzer is 1.4, resulting in a sampling rate of 56MHz / s. The received signal is modeled as follows:

[0099]

[0100] For the modulated phase information, cos(ω) c t) and sin(ω c t) represent the in-phase component and quadrature component of the carrier, respectively.

[0101] A 1-bit differential demodulation algorithm is selected to demodulate the GMSK signal. First, r(t) is multiplied by the in-phase and quadrature components of the carrier wave, respectively, to obtain two mixed signals. Then, each signal is passed through a low-pass filter to remove high-frequency components, resulting in two quadrature signals. These two quadrature signals are then processed by a 1T... b After the delay, multiply each signal by the other signal that has not undergone the delay, and then subtract the results to obtain the expression for the Gaussian band waveform:

[0102]

[0103] 2) Calculate the eye diagram UI, segment the Gaussian band waveform with the eye diagram UI as the period, and superimpose all the segmented eye diagram trajectory frames to obtain the eye diagram;

[0104] Specifically, this step first finds the first maximum position close to 1 in the Gaussian band waveform, then finds 10 zero-crossing positions in sequence to its right. The average of these 10 zero-crossing positions can be used to obtain the starting point of the eye diagram. The superposition range does not exceed the last maximum position close to 1 in the baseband waveform.

[0105] Calculate the values ​​for t less than and greater than t respectively. endThe time of the nearest data point within the time range and t end Time difference Δt l Δt r :

[0106] Δt l =t end (i)-t e (index l );

[0107] Δt r =t e (index r )-t end (i);

[0108] Among them, t end For the deadline of each bit, index l For less than t end Time range from t end The location of the most recent data point, index r For greater than t end Time range from t end The location of the most recent data point, t e The time corresponding to the Gaussian band waveform;

[0109] If Δt l <Δt r The cutoff time for this bit is t. e (index r If the next data point is not specified, add a 0 before it; otherwise, the cutoff time for that bit is t. e (index l The data point is appended with a 0 to ensure that the number of data points in each frame of the eye diagram trajectory obtained by segmentation remains consistent.

[0110] 3) Calculate the probability density of the eye diagram;

[0111] Specifically, firstly, for each state group A of the eye diagram... i Eye trajectory y i (t) Take the derivative and then the inverse to obtain the probability density of the eye diagram trajectory for each state group, and then calculate the number N of each state group. i The probability density formula for the eye diagram trajectory is as follows: Multiply the probability density of each state group's eye diagram trajectory by the number of state groups, add them together, and then divide by the total number of state groups.

[0112]

[0113] in:

[0114] y∈[-1,1];

[0115] 4) The BT value of the GMSK signal is obtained by inversely calculating the probability density of the eye diagram. The difference between the measured signal eye diagram and the modeled eye diagram under the same BT value condition is calculated to obtain subtle differences between them. Statistical features are further extracted from the eye diagram differences, and combined with the BT value features to construct an RF fingerprint feature vector, including:

[0116] a. Calculate the voltage level:

[0117] Clustering is used to obtain the amplitudes corresponding to the four maxima in the probability density curve. These four amplitudes, in ascending order, are: Level 0 (largest) and Level 0 (lowest). max Level 0 (small 0 level) min Level 1 min Level 1 max ;

[0118] b. Calculate the BT value:

[0119] Level 0 (small 0 level) min and Level 0 max Level 1 max and Level 1 min Substituting the difference into the level difference function for the BT value, we obtain two BT values, which are ordered from smallest to largest as follows: Small BT value BT min Large BT value BT max The difference function for calculating the BT value is as follows:

[0120]

[0121]

[0122] c. Construct an eye diagram trajectory mathematical model based on the BT value. Superimpose the eye diagram trajectory mathematical models for all state groups to obtain the eye diagram model. The eye diagram trajectory mathematical model for each state group is as follows:

[0123] Y(t)=[A1G(t),A2G(t),...,A8G(t)] T ;

[0124] d. Subtract the measured signal eye diagram from the modeled eye diagram under the same BT value conditions, extract the mean, maximum, and minimum values ​​of the eye diagram difference, and combine them with the largest BT value. min Small BT value BT max The mean of these two BT values ​​BT mean Construct radio frequency fingerprint feature vectors using equal statistical measures.

[0125] 5) Input the feature vector into the support vector machine for device authentication.

[0126] Specifically, 200 samples from each device were divided into a training set and a test set in a 7:3 ratio. The resulting 27-dimensional feature vectors were used as input to a support vector machine (SVM) and a Gaussian kernel function was used to learn and train the model. The classification accuracy of the model was 99.89% on the training set and 99.49% on the test set.

[0127] In another embodiment, the present invention also provides an eye-map-based radio frequency fingerprint recognition system, comprising:

[0128] The baseband waveform acquisition unit is used to acquire the transmitter signal to be identified, demodulate the transmitter signal to be identified, and achieve carrier separation.

[0129] The eye diagram acquisition unit is used to estimate the eye diagram UI, segment the Gaussian band waveform with the eye diagram UI as the period, and superimpose the segmented eye diagram trajectory frames together to obtain the eye diagram.

[0130] The BT value acquisition unit is used to obtain the probability density curve of the eye diagram generated by the Gaussian band waveform of the transmitter with GMSK modulation, and then use the probability density of the eye diagram to back-calculate the BT value of the GMSK signal.

[0131] The radio frequency fingerprint feature vector construction unit is used to obtain the BT value of the eye diagram, construct the eye diagram model using the BT value, and construct the radio frequency fingerprint feature vector based on the eye diagram differences and the BT value.

[0132] The identity authentication unit is used to input the radio frequency fingerprint feature vector into a support vector machine for device identity authentication.

[0133] The description of the eye-map-based radio frequency fingerprint recognition system in this embodiment is relatively simple because it corresponds to the eye-map-based radio frequency fingerprint recognition method in the above embodiment. For related similarities, please refer to the description of the eye-map-based radio frequency fingerprint recognition method in the above embodiment, which will not be described in detail here.

[0134] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0135] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0136] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0137] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0138] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0139] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A radio frequency fingerprint recognition method based on eye diagram, characterized in that, Includes the following steps: The transmitter signal to be identified is acquired, and the transmitter signal to be identified is demodulated to separate the carrier wave and obtain the Gaussian band waveform of the transmitter signal. Calculate the eye diagram UI, segment the Gaussian band waveform with the eye diagram UI as the period, and superimpose all the segmented eye diagram trajectory frames to obtain the eye diagram; Calculate the probability density of the eye diagram; The BT value of the Gaussian minimum frequency shift keying (GMSK) signal is obtained by back-calculating the probability density of the eye diagram; the BT value is the product of the bandwidth B of the Gaussian filter and the input symbol width T. Construct an eye diagram model of the BT value; The difference between the measured signal eye diagram and the modeled eye diagram under the same BT value condition is calculated, and statistical features are further extracted from the eye diagram differences. The RF fingerprint feature vector is constructed by combining the BT value features. The radio frequency fingerprint feature vector is input into a machine learning algorithm for device authentication; the machine learning algorithm is a support vector machine.

2. The radio frequency fingerprint recognition method based on eye diagram according to claim 1, characterized in that, Calculate the eye diagram UI, including: Find the time when the amplitude of the Gaussian band waveform is 0. In the ideal case of signal transmission, the bit period remains unchanged, the position of the zero crossing should be the bit sequence transition edge, and the relationship between the number of bits in the bit sequence and the corresponding time is linearly related. Calculate the cutoff time less than each bit The time of the nearest data point within the time range and Time difference And calculations in greater than The time of the nearest data point within the time range and Time difference ; if The deadline for this bit is Add a 0 before the next data bit; otherwise, the cutoff time for that bit is... .

3. The radio frequency fingerprint recognition method based on eye diagram according to claim 1, characterized in that, Calculating the probability density of the eye diagram includes: The probability density of the eye diagram trajectory for each state group is obtained by taking the derivative and then the inverse of the derivative. The probability density of the eye diagram trajectory is obtained by multiplying the probability density of each state group by the number of state groups, summing the results, and then dividing by the total number of state groups.

4. The radio frequency fingerprint recognition method based on eye diagram according to claim 1, characterized in that, The BT value of the GMSK signal is obtained by inversely calculating the probability density of the eye diagram, including: Clustering is used to obtain the amplitudes corresponding to the four maxima in the probability density curve, including: large 0 level. , small 0 level 1 level and large 1 level ; For the small 0 level respectively and large 0 level Large 1 level and small 1 level Substituting the difference into the level difference function for the BT value, the smaller BT value is calculated. and large BT value .

5. The radio frequency fingerprint recognition method based on eye diagram according to claim 1, characterized in that, Constructing an eye diagram model for the BT value includes: Construct a mathematical model for the eye diagram trajectory for each state group: ； in: This is the state group of the eye diagram. , For width is The Gaussian impulse response of the input rectangular pulse is specifically expressed as: ； exist In this context, B represents the 3dB bandwidth of the filter. The eye diagram model is obtained by superimposing the eye diagram trajectories of all state groups.

6. A radio frequency fingerprint recognition method based on an eye diagram according to claim 1 or 4, characterized in that, A radio frequency fingerprint feature vector is constructed by combining BT value features and eye diagram difference features, including: Get large BT values Small BT value and the mean of these two BT values ; Obtain the mean, maximum, and minimum values ​​of the eye diagram differences for each state group; Construct an RF fingerprint feature vector based on the acquired values.

7. The radio frequency fingerprint recognition method based on eye diagram according to claim 1, characterized in that, The radio frequency fingerprint feature vector is input into a machine learning algorithm for device authentication, including: Features are randomly divided into training and test sets according to a preset ratio and then input into a support vector machine (SVM) for classification. The label results output by the trained model are compared with the actual labels to obtain the classification accuracy.

8. A radio frequency fingerprint recognition method based on an eye diagram according to claim 1 or 7, characterized in that, Also includes: Principal component analysis was used to reduce the dimensionality of the RF fingerprint feature vectors and then visualize them.

9. A radio frequency fingerprint recognition system based on eye diagram, characterized in that, include: The baseband waveform acquisition unit is used to acquire the transmitter signal to be identified, demodulate the transmitter signal to be identified, separate the carrier, and obtain the Gaussian band waveform of the transmitter signal. The eye diagram acquisition unit is used to segment the Gaussian band waveform and superimpose all the captured eye diagram trajectory frames to obtain an eye diagram; The BT value acquisition unit is used to calculate the probability density of the eye diagram and use the probability density of the eye diagram to back-deduce the BT value of the Gaussian minimum frequency shift keying (GMSK) signal; the BT value is the product of the bandwidth B of the Gaussian filter and the input symbol width T. The radio frequency fingerprint feature vector construction unit is used to construct the eye diagram model of the BT value, calculate the difference between the measured signal eye diagram and the modeled eye diagram under the same BT value conditions, further extract statistical features from the eye diagram differences, and construct the radio frequency fingerprint feature vector by combining the BT value features. The identity authentication unit is used to input the radio frequency fingerprint feature vector into a machine learning algorithm for device identity authentication; the machine learning algorithm is a support vector machine.

Images