CN-121980493-A - Radio frequency fingerprint identification method and device based on multidimensional time sequence feature fusion

CN121980493ACN 121980493 ACN121980493 ACN 121980493ACN-121980493-A

Abstract

The invention discloses a radio frequency fingerprint identification method and a device based on multi-dimensional time sequence feature fusion, comprising the steps of intercepting sampling data corresponding to signal data of wireless communication equipment, performing standardized processing, extracting multi-dimensional signal features according to the processed sampling data, and constructing multi-dimensional feature vectors; the method comprises the steps of constructing a multi-dimensional feature sequence and a corresponding label according to multi-dimensional feature vectors of continuous frames to obtain a data set for identity authentication of wireless communication equipment, training a classifier by utilizing the data set to obtain a trained classifier, inputting the signal feature sequence to be identified into the trained classifier, determining the optimal sequence length according to the obtained initial identification result, taking the identification result with the optimal sequence length corresponding to the initial identification result as a target identification result, and carrying out multi-group authentication through the target identification result to obtain a final identification result. The invention improves the fingerprint identification accuracy of the equipment by fully mining the sequence characteristics of the signals in the time dimension.

Inventors

Zhao Shuangrui
LIU YANG
GAO YUE
WANG ZHAOBO
LI SONGYAN
ZHANG YUANYU
SHEN YULONG

Assignees

西安电子科技大学

Dates

Publication Date: 20260505
Application Date: 20251226

Claims (10)

1. A radio frequency fingerprint identification method based on multi-dimensional time series feature fusion, the method comprising: Intercepting sampling data corresponding to a frame preamble of signal data in a communication process of wireless communication equipment, and performing standardization processing to obtain processed sampling data; extracting multidimensional signal features of the wireless communication device according to the processed sampling data; Constructing a multidimensional feature vector according to the multidimensional signal features; constructing a multi-dimensional feature sequence based on time sequence and a corresponding label according to the multi-dimensional feature vectors of continuous frames to obtain a data set for identity authentication of wireless communication equipment, wherein the label represents the serial number of the wireless communication equipment; Training the classifier by using the data set to obtain the trained classifier; inputting the signal characteristic sequence to be identified into the classifier after training is completed, obtaining an initial identification result, determining the optimal sequence length according to the initial identification result, taking a fingerprint identification result with the optimal sequence length corresponding to the initial identification result as a target identification result, and carrying out multi-group authentication through the target identification result to obtain a final identification result.
2. The method for radio frequency fingerprinting based on multi-dimensional time series feature fusion as set forth in claim 1, wherein the multi-dimensional signal features include carrier frequency offset, IQ imbalance, amplitude error and phase error.
3. The method for identifying the radio frequency fingerprint based on the multi-dimensional time series feature fusion according to claim 1, wherein the processed sampling data are represented as follows: ; Wherein, the For the processed sample data, In order to be able to sample the data in question, As an average value of the sampled data, Is the variance of the sampled data.
4. The method for identifying the radio frequency fingerprint based on the multi-dimensional time series feature fusion according to claim 2, wherein the carrier frequency offset is expressed as follows: ; Wherein, the Representing the carrier frequency offset in question, Indicating that the phase angle of the complex number is taken, For the number of sample points of the processed sample data, Is the first of the processed sampling data The co-directional components of the individual sample points, Is the first of the processed sampling data The orthogonal components of the individual sample points, In imaginary units.
5. The method of claim 4, wherein the IQ imbalance includes IQ amplitude imbalance parameters and IQ phase imbalance parameters, the IQ amplitude imbalance parameters being expressed as follows: ; Wherein, the Representing the IQ amplitude imbalance parameter; the IQ phase imbalance parameters are expressed as follows: ; ; Wherein, the Representing the IQ phase imbalance parameter; The phase error is expressed as follows: ; Wherein, the Which is indicative of the phase error of the optical fiber, For adjacent sampling points in the processed sampling data Sum point A phase difference between them; the amplitude error is expressed as follows: ; Wherein, the Representing the magnitude error of the signal in question, Is obtained after fast Fourier transform The amplitude value of the individual frequency points, Is the average of the amplitudes of all frequency bins.
6. The method for identifying the radio frequency fingerprint based on the multi-dimensional time series feature fusion according to claim 5, wherein the normalization process of the multi-dimensional signal features is represented as follows: ; Wherein, the Is the characteristic value of the normalized multidimensional signal characteristic, Is a feature value of the original multidimensional signal feature, Is the mean of the eigenvalues of the original multidimensional signal feature, Is the standard deviation of the feature value of the original multidimensional signal feature; the multi-dimensional feature vector is represented as follows: ; Wherein, the Representing the normalized carrier frequency offset, Representing the normalized IQ amplitude imbalance parameter, Representing the normalized IQ phase imbalance parameter, Indicating the phase error after normalization, Representing the normalized amplitude error.
7. The method for identifying the radio frequency fingerprint based on multi-dimensional time series feature fusion according to claim 1, wherein the data set is represented as follows: ; Wherein, the The data set is represented by a set of data, Indicating the number of wireless communication devices, Represent the first The data of the individual wireless communication devices, Represent the first A multi-dimensional signature sequence for each wireless communication device, Represent the first Labels corresponding to the multidimensional feature sequences of the wireless communication devices; Construction of a time-sequential-based multidimensional feature sequence from multidimensional feature vectors of successive frames The expression is as follows: ; ; ; ; wherein the multidimensional feature sequence , Representing the number of multi-dimensional feature vectors extracted for each wireless communication device, Representing the number of consecutive frames.
8. The method for identifying the radio frequency fingerprint based on multi-dimensional time series feature fusion according to claim 1, wherein the initial identification result is expressed as follows: ; Wherein, the Is the first The initial recognition result of the individual multi-dimensional feature sequences, And obtaining the optimal sequence length according to the initial recognition result.
9. The method for identifying the radio frequency fingerprint based on the multi-dimensional time series feature fusion according to claim 8, wherein the process of multi-group authentication comprises the following steps: obtaining the occurrence number of the fingerprint identification result with the largest occurrence number according to each fingerprint identification result in the target identification results Calculating reliability of multiple groups of authentications ; When the reliability is the same When the number of the fingerprint recognition results is larger than a preset authentication threshold lambda, changing each fingerprint recognition result in the target recognition results into the maximum occurrence number Corresponding fingerprint identification result And if not, taking the target recognition result as the final recognition result.
10. A radio frequency fingerprint identification device based on multi-dimensional time series feature fusion, the device comprising: the sampling data processing module is used for intercepting corresponding sampling data of signal data of the wireless communication equipment in the communication process and carrying out standardized processing to obtain processed sampling data; the feature extraction module is used for extracting multidimensional signal features of the wireless communication equipment according to the processed sampling data; the feature vector construction module is used for constructing a multi-dimensional feature vector according to the multi-dimensional signal features; The system comprises a data set construction module, a data set identification module and a data set identification module, wherein the data set construction module is used for constructing a multi-dimensional characteristic sequence based on a time sequence and a corresponding label according to multi-dimensional characteristic vectors of continuous frames to obtain a data set for the identity authentication of wireless communication equipment; The classifier training module is used for training the classifier by utilizing the data set to obtain the classifier after training; The identification module is used for inputting the signal characteristic sequence to be identified into the classifier after training is completed, obtaining an initial identification result, determining the optimal sequence length according to the initial identification result, taking the fingerprint identification result with the optimal sequence length corresponding to the initial identification result as a target identification result, and carrying out multi-group authentication through the target identification result to obtain a final identification result.

Description

Radio frequency fingerprint identification method and device based on multidimensional time sequence feature fusion Technical Field The invention belongs to the technical field of wireless information security, and particularly relates to a radio frequency fingerprint identification method and device based on multi-dimensional time sequence feature fusion. Background The wide popularization of wireless equipment promotes communication convenience and simultaneously brings the challenges of mass equipment identity authentication, so that the corresponding network security threat is more and more serious. The equipment identity authentication has become a key link to be enhanced in the current network security system. Compared with the traditional cryptography method, the physical layer authentication technology provides a quicker and more efficient solving path for dealing with large-scale wireless equipment access. The radio frequency fingerprint is used as a key physical layer characteristic, and has unique advantages in the aspects of realizing equipment classification and authentication. At present, research on radio frequency fingerprints mainly extends around two types of methods, namely a scheme based on feature engineering and a scheme based on deep learning. However, most of the two directions still depend on signal characteristics at a single moment, and sequence characteristics of signals in a time dimension cannot be fully mined, so that the recognition accuracy still has room for improvement. Disclosure of Invention In order to solve the problems in the prior art, the invention provides a radio frequency fingerprint identification method and device based on multi-dimensional time sequence feature fusion. The technical problems to be solved by the invention are realized by the following technical scheme: In a first aspect, the present invention provides a radio frequency fingerprint identification method based on multi-dimensional time series feature fusion, the method comprising: Intercepting sampling data corresponding to a frame preamble of signal data in a communication process of wireless communication equipment, and performing standardization processing to obtain processed sampling data; extracting multidimensional signal features of the wireless communication device according to the processed sampling data; Constructing a multidimensional feature vector according to the multidimensional signal features; constructing a multi-dimensional feature sequence based on time sequence and a corresponding label according to the multi-dimensional feature vectors of continuous frames to obtain a data set for identity authentication of wireless communication equipment, wherein the label represents the serial number of the wireless communication equipment; Training the classifier by using the data set to obtain the trained classifier; inputting the signal characteristic sequence to be identified into the classifier after training is completed, obtaining an initial identification result, determining the optimal sequence length according to the initial identification result, taking a fingerprint identification result with the optimal sequence length corresponding to the initial identification result as a target identification result, and carrying out multi-group authentication through the target identification result to obtain a final identification result. Optionally, the constructing a multidimensional feature vector according to the processed frame preamble includes: extracting multidimensional signal features of the wireless communication device according to the processed frame preamble; And constructing a multidimensional feature vector according to the multidimensional signal features. Optionally, the multi-dimensional signal characteristics include carrier frequency offset, IQ imbalance, amplitude error, and phase error. Optionally, the processed sample data is represented as follows: ; Wherein, the For the processed sample data,In order to be able to sample the data in question,As an average value of the sampled data,Is the variance of the sampled data. Optionally, the carrier frequency offset is expressed as follows: ; Wherein, the Representing the carrier frequency offset in question,Indicating that the phase angle of the complex number is taken,For the number of sample points of the processed sample data,Is the first of the processed sampling dataThe co-directional components of the individual sample points,Is the first of the processed sampling dataThe orthogonal components of the individual sample points,In imaginary units. Optionally, the IQ imbalance includes an IQ amplitude imbalance parameter and an IQ phase imbalance parameter, the IQ amplitude imbalance parameter being expressed as follows: ; Wherein, the Representing the IQ amplitude imbalance parameter; the IQ phase imbalance parameters are expressed as follows: ; ; Wherein, the Representing the IQ phase imbalance parameter; The ph