ADS-B signal radio frequency fingerprint open set identification method

Through Gaussian hybrid model and frequency bias layer processing, combined with Swin Transformer model, the problem of low accuracy of RF fingerprint recognition of ADS-B signals is solved, and high accuracy recognition under time and environment changes are achieved.

CN119004201AActive Publication Date: 2024-11-22BEIHANG UNIV
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411011094.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-22
Estimated Expiration
2044-07-26

AI Technical Summary

Technical Problem

The existing ADS-B signal radio frequency fingerprint recognition technology has low recognition accuracy in practical applications, especially in the case of time and environment changes, resulting in invalid models and inability to effectively identify newly acquired data.

Method used

The Gaussian hybrid model and the expected maximization algorithm are used to reconstruct the ideal signal, combine the time-frequency feature map and frequency bias layer processing, and the ADS-B signal RF fingerprint open set recognition is performed through the Swin Transformer model, taking into account the signal's own attributes and frequency bias characteristics.

Benefits of technology

The recognition accuracy is significantly improved in closed and open set time tests, and can effectively identify newly acquired ADS-B signals to adapt to time and environmental changes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004201A_ABST
    Figure CN119004201A_ABST
Patent Text Reader

Abstract

The invention provides an ADS-B (Automatic Dependent Surveillance-Broadcast) signal radio frequency fingerprint open set identification method, which comprises the following steps of: constructing a Gaussian mixture model according to an electromagnetic wave signal, and reconstructing an ideal signal; performing time shift and frequency shift on the ideal signal, and converting the electromagnetic wave signal and the ideal signal subjected to time shift and frequency shift to obtain a time-frequency conversion characteristic matrix; establishing a time-frequency characteristic graph according to the time-frequency transformation characteristic matrix; and inputting the time-frequency characteristic pattern into a radio frequency fingerprint identification model to obtain a radio frequency fingerprint identification result, the radio frequency fingerprint identification model including a frequency offset layer of a frequency offset probability matrix of the transmitter. According to the method, radio frequency fingerprint features possibly existing in signals are visually represented through a time-frequency feature map, a frequency offset layer structure is introduced into a radio frequency fingerprint identification model based on Swin Transformer to improve the identification accuracy, the characteristics of a transmitter are combined with an identification network through a frequency offset layer, and the identification accuracy is improved in a closed set time test and an open set time test. And better identification accuracy is obtained.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of radio frequency fingerprint recognition, and in particular relates to an open set recognition method for radio frequency fingerprint of an ADS-B signal. Background Art

[0002] The Automatic Dependent Surveillance-Broadcast (ADS-B) system uses the global satellite positioning system, air-to-ground and air-to-air data links for traffic monitoring and information dissemination. Each aircraft uses a unique 24-bit address code as its identity information in the ADS-B signal. Aircraft equipped with ADS-B equipment will periodically broadcast their status information, allowing airspace management departments to comprehensively monitor the status of all aircraft in the area, thereby ensuring the normal operation of air traffic. However, due to the unencrypted nature of the ADS-B system and the widespread use of software-defined radio equipment, any user can send ADS-B signals. Malicious users can use known 24-bit address codes to impersonate legitimate users, resulting in airspace security failure. Although traditional key encryption methods can solve the problem of malicious intrusion, it undermines the openness principle of the ADS-B system and is not suitable for practical applications.

[0003] RF fingerprinting technology uses subtle differences in signal performance caused by hardware deviations during the manufacturing process of the device to confirm the identity of the radiation source. These deviations arise from various factors, such as materials, processes, equipment and environmental conditions, and even exist within the same batch of equipment. Signal distortion is mainly caused by components in analog circuits such as oscillators, amplifiers and mixers, which will cause DC bias, I / Q orthogonal bias errors, phase noise, frequency offset, etc. in the signal. The expression of these deviations in the signal is typically universal and unique. Specifically, deviations are common between devices, but are unique in each device. This forms the physical basis for identifying the identity of the radiation source through RF signals.

[0004] In the fingerprint recognition research, ADS-B has been widely discussed. The feasibility of the radio frequency fingerprint method of ADS-B described by American universities and research institutions has been confirmed through experiments. It is worth noting that when the system is deployed outdoors for real-time data reception and recognition using the existing algorithm, the recognition accuracy of the system will drop below 50%. This type of problem corresponds to the problem of open set time test and closed set time test. Closed set time test means that the data for training model and the data for verification model are from the same time period. This method is usually used in scientific research experiments. The collected data is randomly selected, and one part is selected as the training set and the other part is selected as the verification set. Open set time test means that the data during the verification test does not belong to the same time period as the data during the training test. Due to the changes in time and environment, the data collected in a fixed time cannot contain the future state and its impact on the circuit. Therefore, the model trained with data within a specific time range is valid for the data collected within the time period, but invalid for new data beyond the time distribution of the training data, even if the signals come from the same transmitter. Because the circuit performance is affected by power supply fluctuations, ambient temperature and other radiation sources, some characteristics of the circuit will change with the changes in time and environment. For newly acquired data in actual scenarios, changes in circuit performance due to various factors may render the trained model invalid. Therefore, the existing technology in actual applications will cause a significant drop in recognition accuracy. Summary of the invention

[0005] In view of this, an object of the present invention is to provide an ADS-B signal radio frequency fingerprint open set recognition method to meet the demand for improving the recognition rate.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] According to a first aspect, an embodiment of the present invention provides an ADS-B signal radio frequency fingerprint open set identification method, including: receiving and preprocessing an electromagnetic wave signal emitted by an ADS-B transmitter antenna; constructing a Gaussian mixture model according to the preprocessed electromagnetic wave signal, updating the parameters of the Gaussian mixture model using an expectation maximization algorithm, and reconstructing an ideal signal; time-shifting and frequency-shifting the ideal signal, transforming the preprocessed electromagnetic wave signal with the ideal signal that has undergone time shift and frequency shift to obtain a time-frequency transformation feature matrix; establishing a time-frequency feature graph according to the time-frequency transformation feature matrix; inputting the time-frequency feature graph into an ADS-B radio frequency fingerprint recognition model to obtain an ADS-B signal radio frequency fingerprint open set identification result, wherein the ADS-B radio frequency fingerprint recognition model is constructed based on a modified Swin Transformer model, and a frequency deviation layer processing structure is added, wherein the frequency deviation layer is constructed using the frequency deviation characteristics inherent in the ADS-B transmitter as a priori conditions.

[0008] Optionally, a Gaussian mixture model is constructed based on the preprocessed electromagnetic wave signal, and the parameters of the Gaussian mixture model are updated using an expectation-maximization algorithm to reconstruct an ideal signal, including: calculating the posterior probability that the pulse amplitude of the preprocessed electromagnetic wave signal data point belongs to each Gaussian distribution, and updating the parameters of the Gaussian mixture model based on the posterior probability until convergence to obtain multiple high pulse amplitude parameters and multiple low pulse amplitude parameters; calculating the average value of the multiple high pulse amplitude parameters, calculating the average value of the multiple low pulse amplitude parameters, and obtaining the ideal signal based on the average values ​​of the high and low pulse amplitude parameters and the average value of the low pulse amplitude parameters.

[0009] Optionally, the ideal signal is time-shifted and frequency-shifted, and the preprocessed electromagnetic wave signal is used to transform the ideal signal with time-shifted and frequency-shifted to obtain a time-frequency transformation feature matrix, including:

[0010]

[0011] Among them, |χ(τ,f d )∣ represents the result of transforming the ideal signal with the electromagnetic wave signal demodulated to the baseband, De represents the demodulation of the signal to the baseband, λ represents the spatial propagation attenuation coefficient, G is the nonlinear effect of the power amplifier on the signal after passing through the power amplifier, and k I and k Q is the I / Q gain factor, g I (t), g Q (t) represents the output of I / Q signal through DAC, D I and D Q represents the I / Q DC offset, Δf represents the frequency offset error caused by the local oscillator, and f d represents the Doppler frequency shift caused by the channel, f0 is the propagation frequency of the signal in space, c is the speed of light, v is the speed of the aircraft, θ is the angle between the speed direction and the line connecting the aircraft and the ground receiving station, φ is the phase error, is the phase noise generated by the local oscillator, It is an ideal signal to add time and frequency shift.

[0012] Optionally, the time-frequency feature map is input into an ADS-B radio frequency fingerprint recognition model to obtain an open set recognition result of the ADS-B signal radio frequency fingerprint, including: performing patch partition processing on the time-frequency feature map to divide it into multiple image blocks; the linear embedding layer maps the divided image blocks and passes them into the Swin Transformer to extract feature information; the feature information is input into the normalization layer and the average pooling layer for normalization and average pooling, and then input into the post-processing layer, the post-processing layer includes a fully connected layer, a frequency offset layer and a Softmax layer, the output of the fully connected layer is used as the input of the frequency offset layer, the frequency offset layer contains the frequency offset probability matrix of various ADS-B transmitters, and the Softmax layer uses the frequency offset probability matrix of various ADS-B transmitters as a priori conditions to obtain the open set recognition result of the ADS-B signal radio frequency fingerprint.

[0013] Optionally, the received electromagnetic wave signal emitted by the ADS-B transmitter antenna is:

[0014]

[0015] Among them, X Spa (t) is the electromagnetic wave signal received from the antenna of the ADS-B transmitter, λ represents the spatial propagation attenuation coefficient, G is the nonlinear effect of the power amplifier on the signal after passing through the power amplifier, and k I and k Q is the I / Q gain factor, g I (t), g Q (t) represents the output of I / Q signal through DAC, D I and D Q represents the I / Q DC offset, Δf represents the frequency offset error caused by the local oscillator, and f d represents the Doppler frequency shift caused by the channel, f0 is the propagation frequency of the signal in space, c is the speed of light, v is the speed of the aircraft, θ is the angle between the speed direction and the line connecting the aircraft and the ground receiving station, φ is the phase error, is the phase noise produced by the local oscillator.

[0016] Optionally, according to the posterior probability, the parameters of the Gaussian mixture model are updated, including:

[0017]

[0018] Among them, N k represents the number of k-th Gaussian function data points, I represents the number of parameters, γ(ik) represents the posterior probability of the pulse amplitude that the i-th data point belongs to the k-th Gaussian distribution, and λ k represents the weight of the k-th Gaussian model, μ krepresents the mean of the Gaussian component, σ k represents the Gaussian component variance, A i Represents the pulse amplitude of the ith data point.

[0019] Optionally, calculating an average value of a plurality of high pulse amplitude parameters and obtaining an ideal signal according to the average value of the high pulse amplitude parameters, or calculating an average value of a plurality of low pulse amplitude parameters and obtaining an ideal signal according to the average value of the low pulse amplitude parameters, comprises:

[0020]

[0021] in, represents the ideal signal for reconstruction, A HL Indicates that the pulse amplitude is A H or A L , A H Represents the average value of the high pulse amplitude parameter, A L represents the average value of the low pulse amplitude parameter, N represents the total number of data points, and p(t-iT) represents the pulse of the i-th point at time iT.

[0022] Optionally, the modified SwinTransformer includes: a SW-MSA module and a W-MSA module.

[0023] According to a second aspect, an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the steps of the ADS-B signal radio frequency fingerprint open set identification method described in the first aspect or any embodiment of the first aspect.

[0024] According to a third aspect, an embodiment of the present invention provides a computer storage medium having computer instructions stored thereon, which, when executed by a processor, implement the steps of the ADS-B signal radio frequency fingerprint open set identification method described in the first aspect or any embodiment of the first aspect.

[0025] This method provides an open set identification method for radio frequency fingerprint of ADS-B signal. In this method, various parameters such as phase noise, frequency shift, and time delay in the signal are reflected through the signal time-frequency characteristic diagram. Compared with the traditional method that only uses signal I / Q data or simple spectrum data as the input of the neural network, the scheme of the embodiment of the present invention considers the study of the properties of the signal itself, more intuitively represents the radio frequency fingerprint features that may exist in the signal, and introduces a post-processing layer in the ADS-B radio frequency fingerprint recognition model to replace the original SoftMax layer output. The post-processing layer includes a fully connected layer, a frequency deviation layer and a softmax layer. The characteristics of the transmitter are combined with the recognition network through the frequency deviation layer. Compared with the traditional method, better recognition accuracy is achieved in both the closed set time test and the open set time test.

[0026] Other advantages, objectives and features of the present invention will be described in the following description and will be apparent to those skilled in the art to some extent, or those skilled in the art may be taught from the practice of the present invention. The objectives and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to make the purpose, technical solution and beneficial effects of the present invention clearer, the present invention provides the following drawings for illustration:

[0028] Figure 1 Design a flow chart for a typical RF fingerprint system;

[0029] Figure 2 A specific example flow chart of an ADS-B signal radio frequency fingerprint open set identification method according to the present invention;

[0030] Figure 3 The basic circuit diagram of the ADS-B transmitter of the present invention is shown in FIG.

[0031] Figure 4 It is a schematic diagram of the overall process of the method for constructing a time-frequency feature map of the present invention;

[0032] Figure 5a This is a diagram of electromagnetic wave signals received by the present invention;

[0033] Figure 5b The ideal signal diagram reconstructed by the present invention;

[0034] Figure 6 The time-frequency characteristic diagrams of different types of transmitters established by the present invention through the time-frequency transformation characteristic matrix;

[0035] Figure 7 This is a schematic diagram of the structure of the ADS-B radio frequency fingerprint recognition model of the present invention;

[0036] Figure 8 This is a schematic diagram of the model structure of two consecutive Swin Transformers connected in series in the present invention;

[0037] Fig. 9 This is a framework diagram of the overall process of ADS-B signal transmission and processing of the present invention;

[0038] Fig.10 It is a principle block diagram of a specific example of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, it can also be the internal connection of two components, it can be a wireless connection, or it can be a wired connection. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0041] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0042] Through further research and experiments, it was found that the previous research was conducted in a closed time set, that is, the model training and test data were in the same period. However, the circuit performance will be affected by many factors such as power supply fluctuations, ambient temperature and interference from other radiation sources. As time and environment change, some circuit characteristics will also change accordingly. Previous studies on specified data sets basically randomly selected part of the data for training and the rest of the data for testing. The training set and the test set have the same time distribution and are in the same period. However, for newly acquired data in real scenarios, various factors will cause changes in circuit performance, ultimately making the trained model invalid. Therefore, the research on RF fingerprint technology based on closed time is no longer valid.

[0043] In the relevant research, there are few studies on the open set time, which mainly focus on the research on the open set of data types. How to enable the model to recognize newly arrived categories usually involves related research such as transfer learning, lifelong learning and zero-shot learning, which are classified as research on the openness and non-stationarity of real-world scenarios. The same applies to the study of the open set time of ADS-B RF fingerprint. As time and environment change, data collected in a fixed period cannot cover the state of the circuit in the future and the impact it will receive. Therefore, the model trained based on data from a specific time range is valid for data collected in that time period, but even if the signal comes from the same transmitter, if the time distribution exceeds the training data, then such new data is invalid. In this case, simply using signal I / Q data or simple spectrum data as the input of the neural network is not enough to solve the actual problem. Research on the properties of the signal itself should also be considered in model development.

[0044] Typical RF fingerprint system design is as follows: Figure 1 As shown in the figure, the research on the signal mainly involves the research on the characteristics of the transmitter circuit. By analyzing the main RF features affected by the hardware characteristics and combining the signal transmission method, the signal is modeled. The system design needs to meet the application requirements of the RF fingerprint recognition system and incorporate the functions of signal reception, processing, recognition, storage and algorithm optimization into the entire system. Feature extraction and clustering algorithm design need to use the collected high signal-to-noise ratio data. The clustering algorithm groups signals with similar RF features to identify different emission sources. Then, the algorithm performance is verified with data with the lowest possible signal-to-noise ratio, and high recognition accuracy is maintained under open collection time conditions. Algorithm verification requires processing real data under the above conditions and in combination with the actual application environment to verify the effectiveness of the algorithm.

[0045] Based on the above analysis and description, an embodiment of the present invention provides an ADS-B signal radio frequency fingerprint open set recognition method, such as Figure 2 As shown, including:

[0046] S101, receiving and pre-processing electromagnetic wave signals transmitted by an ADS-B transmitter antenna;

[0047] For example, the electromagnetic wave signal emitted by the antenna of the ADS-B transmitter may be an electromagnetic wave signal obtained after being affected by the DC offset, I / Q imbalance, frequency offset error and phase error in the basic circuit of the ADS-B transmitter. After receiving the electromagnetic wave signal, pre-processing may be performed, such as down-conversion, to obtain the down-converted signal, and demodulation may be performed to extract the original baseband signal from the down-converted signal to restore the original information of the transmitter.

[0048] ADS-B transmitter basic circuit Figure 3As shown, the I / Q signals are output through the DAC and are represented as g I (t), g Q (t). Considering the possible DC offset and I / Q imbalance in the circuit, the DAC output G DAC-OUT This effect is introduced in (t):

[0049] G DAC-OUT (t) = k I (g I (t)+D I )+k Q (g Q (t)+D Q )

[0050] where k I and k Q is the I / Q gain factor, D I and D Q Indicates the I / Q DC bias.

[0051] For the output signal, the carrier generated by the local oscillator is up-converted in the mixer. In reality, the carrier frequency is not always the same as the preset value. This is due to the differences in materials and processes during the manufacturing process of semiconductor devices, which makes the local oscillator frequency of each device slightly different. The deviation between the actual carrier frequency and the expected frequency caused by these changes is called carrier frequency offset (CFO). After passing through the mixer, we introduce frequency offset error and phase error.

[0052]

[0053] Among them G mix-OUT (t) represents the output of the mixer, Δf represents the frequency offset error, φ is the phase error, which is a random process that varies with time, and the nonlinear effect of the power amplifier on the signal after passing through the power amplifier is summarized as G.

[0054]

[0055] Where X Tr (t) is the electromagnetic wave signal emitted from the ADS-B transmitter antenna.

[0056] The above is the circuit model of the entire transmitter. For RF fingerprint research, the main considerations should include the above effects, while other effects are not fully discussed here.

[0057] For an aircraft with a speed v and an altitude of H, assume that θ is the angle between the direction of the speed and the line between the aircraft and the ground receiving station.

[0058]

[0059] Among them, f d represents the Doppler frequency shift caused by the channel, f0 is the propagation frequency of the signal in space, and c represents the speed of light. Because the receiver is usually located at an open airport or on a mountain top, the multipath effect can be ignored. Therefore, considering the impact of the channel on the transmission signal, the electromagnetic wave signal emitted by the ADS-B transmitter antenna is received as:

[0060]

[0061] Among them, X Spa (t) is the electromagnetic wave signal received from the antenna of the ADS-B transmitter, λ represents the spatial propagation attenuation coefficient, G is the nonlinear effect of the power amplifier on the signal after passing through the power amplifier, and k I and k Q is the I / Q gain factor, g I (t), g Q (t) represents the output of I / Q signal through DAC, D I and D Q represents the I / Q DC offset, Δf represents the frequency offset error caused by the local oscillator, and f d represents the Doppler frequency shift caused by the channel, f0 is the propagation frequency of the signal in space, c is the speed of light, v is the speed of the aircraft, θ is the angle between the speed direction and the line connecting the aircraft and the ground receiving station, φ is the phase error, is the phase noise produced by the local oscillator.

[0062] S102, constructing a Gaussian mixture model according to the preprocessed electromagnetic wave signal, updating the parameters of the Gaussian mixture model using an expectation maximization algorithm, and reconstructing an ideal signal;

[0063] Exemplarily, the Gaussian mixture model is:

[0064]

[0065] Among them, GMM is a Gaussian mixture model, λ k represents the weight of the k-th Gaussian model, N(μ k ,σ k ) represents the distribution of the k-th Gaussian model, where μ k represents the mean and σ k Represents variance.

[0066] The process of reconstructing the ideal signal is as follows Figure 4As shown, first determine the number of classification clusters K, K corresponds to the number of classification categories, there are only two levels of high and low amplitude in the ideal ADS-B signal, so K = 2, then initialize the Gaussian distribution, select an initial parameter set, such as the mean and variance of the Gaussian distribution. Then, use the expectation maximization (EM) iterative algorithm to solve which Gaussian distribution the data point comes from, where the data point can be the signal pulse amplitude of the preprocessed electromagnetic wave signal at a certain moment in the time series. Expectation maximization (EM) includes two steps, namely E-step (expectation step), calculating the posterior probability of the pulse amplitude of each data point belonging to each Gaussian distribution, and M-step (maximization step), according to the posterior probability of the pulse amplitude of the data point belonging to each Gaussian distribution, update the mean and variance parameters of the Gaussian mixture model, so that the log-likelihood function is maximized, and repeat the above steps until the convergence condition is met.

[0067] The specific process of calculating the posterior probability of the pulse amplitude of each data point belonging to each Gaussian distribution in the E-step (expectation step) is as follows:

[0068] By using Bayes' theorem, we can calculate:

[0069]

[0070] Where γ(ik) indicates that the i-th data point belongs to the k-th Gaussian distribution.

[0071] The specific process of updating the mean and variance parameters of the Gaussian mixture model is as follows:

[0072]

[0073] Among them, N k represents the number of data points of the k-th Gaussian function, I represents the number of parameters, γ(ik) represents the posterior probability of the pulse amplitude that the ith data point belongs to the k-th Gaussian distribution, and λ k represents the weight of the k-th Gaussian model, μ k represents the mean of the Gaussian component, σ k represents the Gaussian component variance, A i Represents the pulse amplitude of the ith data point.

[0074] If convergence is not reached, the parameters of the Gaussian distribution are updated and the EM algorithm is repeated for iteration. After convergence, clustering is performed to obtain two clusters of multiple high pulse amplitude parameters and multiple low pulse amplitude parameters, and the average value is taken to obtain A H and A L . Calculate the average value of multiple high pulse amplitude parameters, calculate the average value of multiple low pulse amplitude parameters, and obtain the ideal signal according to the average value of the high pulse amplitude parameter and the average value of the low pulse amplitude parameter:

[0075]

[0076] in, represents the ideal reconstructed signal, A HL Indicates that the pulse amplitude is A H or A L , A H Represents the average value of the high pulse amplitude parameter, A L represents the average value of the low pulse amplitude parameter, N represents the total number of data points, and p(t-iT) represents the pulse of the i-th point at time iT.

[0077] The received electromagnetic wave signal is Figure 5a As shown, the reconstructed ideal signal is Figure 5b shown.

[0078] In this embodiment, a method for constructing an ideal signal based on a received signal is proposed. Compared with the traditional method, since the Gaussian mixture model can fit the probability distribution of the data, it can be used for anomaly detection to eliminate erroneous sampling points that may have large deviations in the signal, making the reconstructed signal more accurate.

[0079] S103, performing time shift and frequency shift on the ideal signal, transforming the preprocessed electromagnetic wave signal with the time-shifted and frequency-shifted ideal signal to obtain a time-frequency transformation feature matrix;

[0080] Exemplarily, after reconstructing the ideal signal, Figure 4 As shown, the ideal signal is time-shifted and frequency-shifted, and the ideal signal is transformed into: The preprocessed electromagnetic wave signal is the electromagnetic wave signal demodulated to the baseband, which can be expressed as: De[X Spa (t)], De means that the transmission signal is demodulated to the baseband, and the ideal signal for time shift and frequency shift is transformed by the electromagnetic wave signal demodulated to the baseband. The specific formula of the transformation is as follows:

[0081]

[0082] Among them, |χ(τ,f d )∣ represents the result of transforming the ideal signal with the electromagnetic wave signal demodulated to the baseband, De represents the demodulation of the signal to the baseband, λ represents the spatial propagation attenuation coefficient, G is the nonlinear effect of the power amplifier on the signal after passing through the power amplifier, and k I and k Q is the I / Q gain factor, g I (t), g Q (t) represents the output of I / Q signal through DAC, D I and D Qrepresents the I / Q DC offset, Δf represents the frequency offset error caused by the local oscillator, and f d represents the Doppler frequency shift caused by the channel, f0 is the propagation frequency of the signal in space, c is the speed of light, v is the speed of the aircraft, θ is the angle between the speed direction and the line connecting the aircraft and the ground receiving station, φ is the phase error, is the phase noise generated by the local oscillator, It is an ideal signal to add time and frequency shift.

[0083] The ranges of time offset and rate offset are defined as (τ1, τ2...τ K )∈[-T,T] and (f1,f2...f K )∈[-f,f], we can get the time-frequency transform feature matrix:

[0084]

[0085] Wherein, W represents the time-frequency transform feature matrix constructed by the signal, and K is a positive integer.

[0086] S104, establishing a time-frequency feature map according to the time-frequency transformation feature matrix;

[0087] For example, the time-frequency feature graph established according to the time-frequency transformation feature matrix can be drawn into an image using an appropriate visualization tool, such as Matplotlib, Seaborn, etc. The horizontal axis of the feature graph is usually time, the vertical axis is frequency, and the color represents energy or amplitude. The time-frequency feature graphs of different types of transmitters established by the time-frequency transformation feature matrix are as follows: Figure 6 As shown, there are 6 categories, each of which has different image features. This embodiment does not limit the method of establishing the time-frequency feature map according to the time-frequency transformation feature matrix, and those skilled in the art can determine it according to needs.

[0088] For different time and frequency shifts, there must be τ and f τ The coherence between the two signals |χ(τ,f d )| reaches the maximum value. This means that the time shift and frequency shift added to the ideal signal simulate the signal frequency change caused by hardware reasons in the RF circuit, which is reflected in the two-dimensional time-frequency characteristic diagram.

[0089] S105, input the time-frequency feature map into the ADS-B radio frequency fingerprint recognition model to obtain the ADS-B signal radio frequency fingerprint open set recognition result. The ADS-B radio frequency fingerprint recognition model is constructed based on the modified Swin Transformer model, and a frequency offset layer processing structure is added. The frequency offset layer is constructed using the inherent frequency offset characteristics of the ADS-B transmitter as a priori conditions.

[0090] For example, the ADS-B radio frequency fingerprint recognition model can be as follows: Figure 7 The Swin Transformer shown is a network with the main architecture. The time-frequency feature map is divided into multiple image blocks by performing patch partition processing. The linear embedding layer maps the divided image blocks and passes them into the Swin Transformer to extract feature information. The feature information is input into the normalization layer and the average pooling layer for normalization and average pooling, and then input into the post-processing layer. The post-processing layer includes a fully connected layer, a frequency offset layer, and a Softmax layer. The output of the fully connected layer is used as the input of the frequency offset layer. The frequency offset layer contains the frequency offset probability matrix of various ADS-B transmitters. The Softmax layer uses the frequency offset probability matrix of various ADS-B transmitters as a priori conditions to obtain the open set recognition result of the ADS-B signal RF fingerprint.

[0091] Specifically, the linear embedding layer maps the divided image blocks and passes them into the Swin Transformer block to extract feature information. By repeatedly stacking the Swin Transformer blocks, the feature information in the image is further extracted. For classification and recognition problems, the output layer includes layer normalization (LN), average pooling (AP), post-processing (BP) and output layer. The BP layer contains a fully connected (FC) layer, a frequency offset layer and a Softmax layer. Based on the above-mentioned characteristic that the signal has frequency offset, the frequency offset probability matrix of the signal is constructed as a priori condition for recognition. When identifying Class A transmitters, it is assumed that there are L different frequency offsets.

[0092]

[0093] Where W f represents the frequency offset probability matrix, P AL It represents the probability that the frequency offset of the type A transmitter is L.

[0094] For the output K of the FC layer:

[0095] K=[K1 K2 ··· K A ]

[0096] Among them, K represents the output of the FC layer, K A Represents the Ath element of the FC layer output, corresponding to the Ath category.

[0097] Taking the frequency offset into account, the corresponding SoftMax output is:

[0098]

[0099] in Indicates that the parameters (z1, z2, ···, zA ) is the vector input of the SoftMax function, W f-i (n) indicates the frequency deviation is W f The element in the i-th row of the n-th column, K i is the i-th element corresponding to the output of the FC layer.

[0100] By inputting the constructed time-frequency feature map into the optimized Swin-Transformer recognition network architecture, the structure fully utilizes the characteristics of the signal, introduces a frequency offset processing layer, and improves the recognition accuracy. High-accuracy video fingerprint recognition is achieved in open-set time test and closed-set time test.

[0101] The ADS-B signal radio frequency fingerprint open set identification method provided by the embodiment of the present invention reflects various parameters such as phase noise, frequency shift, and time delay in the signal through the signal time-frequency characteristic diagram. Compared with the traditional method that only uses signal I / Q data or simple spectrum data as the input of the neural network, the solution of the embodiment of the present invention considers the study of the signal's own properties, more intuitively represents the radio frequency fingerprint features that may exist in the signal, and introduces a post-processing layer in the ADS-B radio frequency fingerprint recognition model to replace the original SoftMax layer output. The post-processing layer includes a fully connected layer, a frequency deviation layer and a softmax layer. The characteristics of the transmitter are combined with the recognition network through the frequency deviation layer. Compared with the traditional method, better recognition accuracy is achieved in both the closed set time test and the open set time test.

[0102] As an optional implementation, an ADS-B signal radio frequency fingerprint open set recognition method, the modified SwinTransformer includes: a SW-MSA module and a W-MSA module.

[0103] For example, Figure 8 As shown, two consecutive Swin Transformers are connected in series, and each Swin Transformer is connected using a SW-MSA module and a W-MSA module.

[0104] Two consecutive Swin Transformer modules are usually represented as:

[0105]

[0106] in and z l is the output feature of SW-MSA / W-MSA and MLP in the lth block.

[0107] In this embodiment, the information on the local window is processed by connecting the SW-MSA module and the W-MSA module, which effectively reduces the number of parameters. In addition, its hierarchical construction method allows efficient parallel computing while effectively extracting global features.

[0108] According to the scheme of the above embodiment, the overall process framework of ADS-B signal transmission and processing is shown in Fig. 9 The electromagnetic wave signal is transmitted through the transmitter circuit and transmitted to the receiver through the channel. The receiver can be an aircraft or other. After receiving the electromagnetic wave signal, the receiver pre-processes the electromagnetic wave signal, including down-conversion and demodulation, extracts the original baseband signal, and performs signal analysis on the extracted baseband signal and stores it for subsequent training and verification of the ADS-B radio frequency fingerprint recognition model. At the same time, the demodulated electromagnetic wave signal is used to reconstruct the ideal signal, and the ideal signal is time-shifted and frequency-shifted. The pre-processed electromagnetic wave signal is used to transform the ideal signal with time-shifted and frequency-shifted to obtain a time-frequency transformation feature matrix, thereby constructing a time-frequency feature map, and the time-frequency feature map is input into the ADS-B radio frequency fingerprint recognition model for recognition to obtain a recognition result.

[0109] The present application also provides an electronic device, such as Fig.10 As shown, a processor 501 and a memory 502, wherein the processor 501 and the memory 502 may be connected via a bus or other means.

[0110] The processor 501 may be a central processing unit (CPU). The processor 501 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above chips.

[0111] The memory 502 is a non-transitory computer-readable storage medium that can be used to store non-transitory software programs, non-transitory computer executable programs and modules, such as the program instructions / modules corresponding to the ADS-B signal radio frequency fingerprint open set identification method in the embodiment of the present invention. The processor executes various functional applications and data processing of the processor by running the non-transitory software programs, instructions and modules stored in the memory.

[0112] The memory 502 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the processor, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 502 may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0113] The one or more modules are stored in the memory 502, and when executed by the processor 501, the following is performed: Figure 1 The ADS-B signal radio frequency fingerprint open set recognition method in the illustrated embodiment.

[0114] For details of the above electronic equipment, please refer to Figure 2 The corresponding related descriptions and effects in the illustrated embodiments can be understood and will not be repeated here.

[0115] This embodiment also provides a computer storage medium, which stores computer executable instructions, and the computer executable instructions can execute the ADS-B signal radio frequency fingerprint open set identification method in any of the above method embodiments. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated: HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above types of memory.

[0116] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made in form and details without departing from the scope defined by the claims of the present invention.

Claims

1. A method for open set identification of radio frequency fingerprint of ADS-B signal. It is characterized in that include: Receive and pre-process the electromagnetic wave signal emitted by the ADS-B transmitter antenna; According to the preprocessed electromagnetic wave signal, a Gaussian mixture model is constructed, and the expectation maximization algorithm is used to update the parameters of the Gaussian mixture model and reconstruct the ideal signal; The ideal signal is time-shifted and frequency-shifted, and the pre-processed electromagnetic wave signal is transformed with the ideal signal with time-shifted and frequency-shifted to obtain a time-frequency transformation feature matrix; Establish a time-frequency feature map according to the time-frequency transformation feature matrix; The time-frequency feature map is input into the ADS-B radio frequency fingerprint recognition model to obtain the open set recognition result of the ADS-B signal radio frequency fingerprint. The ADS-B radio frequency fingerprint recognition model is constructed based on the modified Swin Transformer model, and a frequency offset layer processing structure is added. The frequency offset layer is constructed with the inherent frequency offset characteristics of the ADS-B transmitter as a priori conditions.

2. According to claim 1, a method for open set identification of radio frequency fingerprint of ADS-B signal, It is characterized in that According to the preprocessed electromagnetic wave signal, a Gaussian mixture model is constructed, and the expectation maximization algorithm is used to update the parameters of the Gaussian mixture model and reconstruct the ideal signal, including: Calculate the posterior probability of the pulse amplitude of the electromagnetic wave signal data point after preprocessing belonging to each Gaussian distribution, and update the parameters of the Gaussian mixture model according to the posterior probability until convergence, so as to obtain multiple high pulse amplitude parameters and multiple low pulse amplitude parameters; The average values ​​of the multiple high pulse amplitude parameters are calculated, the average values ​​of the multiple low pulse amplitude parameters are calculated, and the ideal signal is obtained according to the average values ​​of the high pulse amplitude parameters and the average values ​​of the low pulse amplitude parameters.

3. According to claim 1, a method for open set identification of radio frequency fingerprint of ADS-B signal, It is characterized in that The ideal signal is time-shifted and frequency-shifted, and the pre-processed electromagnetic wave signal is transformed with the ideal signal with time-shifted and frequency-shifted to obtain the time-frequency transformation feature matrix, including: Among them, |χ(τ,f d )∣ represents the result of transforming the ideal signal with the electromagnetic wave signal demodulated to the baseband, De represents the demodulation of the signal to the baseband, λ represents the spatial propagation attenuation coefficient, G is the nonlinear effect of the power amplifier on the signal after passing through the power amplifier, k I and k Q is the I / Q gain factor, g I (t), g Q (t) represents the output of I / Q signal through DAC, D I and D Q represents the I / Q DC offset, Δf represents the frequency offset error produced by the local oscillator, and f d represents the Doppler frequency shift caused by the channel, f 0 is the propagation frequency of the signal in space, c represents the speed of light, v is the speed of the aircraft, θ is the angle between the speed direction and the line connecting the aircraft and the ground receiving station, φ is the phase error, is the phase noise generated by the local oscillator, It is an ideal signal to add time and frequency shift.

4. According to claim 1, a method for open set identification of radio frequency fingerprint of ADS-B signal, It is characterized in that The time-frequency feature graph is input into the ADS-B radio frequency fingerprint recognition model to obtain the ADS-B signal radio frequency fingerprint open set recognition results, including: Perform patchpartition processing on the time-frequency feature map and divide it into multiple image blocks; The linear embedding layer maps the divided image blocks and passes them into the Swin Transformer to extract feature information; The feature information is input into the normalization layer and the average pooling layer for normalization and average pooling, and then input into the post-processing layer. The post-processing layer includes a fully connected layer, a frequency offset layer, and a Softmax layer. The output of the fully connected layer is used as the input of the frequency offset layer. The frequency offset layer contains the frequency offset probability matrix of various ADS-B transmitters. The Softmax layer uses the frequency offset probability matrix of various ADS-B transmitters as a priori conditions to obtain the open set recognition result of the ADS-B signal RF fingerprint.

5. According to claim 1, an ADS-B signal radio frequency fingerprint open set recognition method, It is characterized in that The electromagnetic wave signal received from the ADS-B transmitter antenna is: Among them, X Spa (t) is the electromagnetic wave signal received from the antenna of the ADS-B transmitter, λ represents the spatial propagation attenuation coefficient, G is the nonlinear effect of the power amplifier on the signal after passing through the power amplifier, and k I and k Q is the I / Q gain factor, g I (t), g Q (t) represents the output of I / Q signal through DAC, D I and D Q represents the I / Q DC offset, Δf, produced by the local oscillator, and represents the frequency offset error, f d represents the Doppler frequency shift caused by the channel, f 0 is the propagation frequency of the signal in space, c represents the speed of light, v is the speed of the aircraft, θ is the angle between the speed direction and the line connecting the aircraft and the ground receiving station, φ is the phase error, is the phase noise produced by the local oscillator.

6. According to claim 2, an ADS-B signal radio frequency fingerprint open set recognition method, It is characterized in that According to the posterior probability, update the parameters of the Gaussian mixture model, including: Among them, N k represents the number of data points of the k-th Gaussian function, I represents the number of parameters, γ(ik) represents the posterior probability of the pulse amplitude that the ith data point belongs to the k-th Gaussian distribution, and λ k represents the weight of the k-th Gaussian model, μ k represents the mean of the Gaussian component, σ k represents the Gaussian component variance, A i Represents the pulse amplitude of the ith data point.

7. According to claim 2, an ADS-B signal radio frequency fingerprint open set recognition method, It is characterized in that Calculating an average value of a plurality of high pulse amplitude parameters, calculating an average value of a plurality of low pulse amplitude parameters, and obtaining an ideal signal according to the average value of the high pulse amplitude parameters and the average value of the low pulse amplitude parameters, including: in, represents the ideal signal for reconstruction, A HL Indicates that the pulse amplitude is A H or A L , A H Represents the average value of the high pulse amplitude parameter, A L represents the average value of the low pulse amplitude parameter, N represents the total number of data points, and p(t-iT) represents the pulse of the i-th point at time iT.

8. According to claim 4, a method for open set identification of radio frequency fingerprint of ADS-B signal, It is characterized in that The modified SwinTransformer includes: SW-MSA module, W-MSA module.

9. An electronic device, the device include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the steps of the ADS-B signal radio frequency fingerprint open set identification method according to any one of claims 1 to 8.

10. A computer storage medium having computer instructions stored thereon, It is characterized in that When the instruction is executed by the processor, the steps of the ADS-B signal radio frequency fingerprint open set identification method described in any one of claims 1-8 are implemented.

Citation Information

Patent Citations

  • Method for identifying radio frequency fingerprints in civil aviation field based on dynamic fusion model

    CN113449682A

  • Communication radiation source open set identification method and system

    CN115169415A

  • LoRa non-stationary radio frequency fingerprint feature extraction method and related device

    CN116386092A

  • Anti-Doppler radio frequency fingerprint identification method based on lightweight

    CN117851912A

  • Radio frequency fingerprint authentication and identification method based on multi-domain feature fusion and deep learning

    CN118354322A