Full-duplex ethernet device fingerprint extraction method and apparatus, and electronic device

Abstract

The application provides a full-duplex Ethernet device fingerprint extraction method and device and electronic equipment, and the method comprises the following steps: obtaining a training mode signal sent by a master device; performing cross-correlation operation between the training mode signal and a fixed training mode sequence, and obtaining a synchronization signal by intercepting the training mode signal according to the cross-correlation operation result and the length of the fixed training mode sequence; and extracting the device fingerprint of the master device according to the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training mode sequence. The training mode signal sent by the master device is obtained, the synchronization signal is obtained through synchronization, and the device fingerprint of the master device is extracted according to the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training mode sequence, so that the device fingerprint of the full-duplex Ethernet master device is stably extracted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information security, and in particular to a full-duplex Ethernet device fingerprint extraction method and device and electronic equipment. BACKGROUND

[0002] Device fingerprint information (DFI) is an identity information generated by the physical characteristics of the device itself and parasitic in the transmitted signal, which has the characteristics of being unclonable. The transmitted signal of wired devices or wireless devices contains device fingerprints, and the access authentication problem can be effectively solved through device fingerprints.

[0003] The prior art discloses a method for obtaining adaptive filter parameters as wired network card device fingerprints by restoring ideal signals. For half-duplex 100BASE-T consumer Ethernet, this method can be applicable, but for full-duplex 100BASE-T1 devices, such as 100BASE-T1 devices, in the stable communication process, the full-duplex signals transmitted by the devices on both sides of the line are superimposed together, and it is difficult to restore the correct ideal signal, resulting in unstable device fingerprint information extracted. SUMMARY

[0004] In view of the problems in the prior art, the present application provides a full-duplex Ethernet device fingerprint extraction method, device and electronic equipment.

[0005] In a first aspect, the present application provides a full-duplex Ethernet device fingerprint extraction method, comprising:

[0006] Obtaining a training mode signal sent by a master device;

[0007] Correlating the training mode signal and a fixed training mode sequence, and according to the correlation result and the length of the fixed training mode sequence, the training mode signal is intercepted to obtain a synchronization signal;

[0008] According to the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training mode sequence, the device fingerprint of the master device is extracted.

[0009] Optionally, the device fingerprint of the master device is extracted according to the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training mode sequence, comprising:

[0010] The synchronization signal is normalized to obtain a normalized synchronization signal;

[0011] The normalized synchronization signal is divided into Q sub-signals with same length, and an eigenvalue vector of the synchronization signal is determined according to eigenvalues of Wigner-Ville distribution matrix of each of the sub-signals, wherein Q is an integer greater than or equal to 1.

[0012] The fixed training mode sequence is divided into Q sub-sequences with same length, and an eigenvalue vector of the fixed training mode sequence is determined according to eigenvalues of Wigner-Ville distribution matrix of each of the sub-sequences.

[0013] A device fingerprint of the master device is extracted according to the eigenvalue vector of the synchronization signal and the eigenvalue vector of the fixed training mode sequence.

[0014] Optionally, the eigenvalue vector of the synchronization signal is determined according to the eigenvalues of Wigner-Ville distribution matrix of each of the sub-signals, including:

[0015] The eigenvalue vector of the synchronization signal is obtained by averaging the eigenvalues of Wigner-Ville distribution matrix of each of the sub-signals.

[0016] The eigenvalue vector of the fixed training mode sequence is determined according to the eigenvalues of Wigner-Ville distribution matrix of each of the sub-sequences, including:

[0017] The eigenvalue vector of the fixed training mode sequence is obtained by averaging the eigenvalues of Wigner-Ville distribution matrix of each of the sub-sequences.

[0018] Optionally, the device fingerprint of the master device is extracted according to the eigenvalue vector of the synchronization signal and the eigenvalue vector of the fixed training mode sequence, including:

[0019] The device fingerprint of the master device is obtained by dividing corresponding elements of the eigenvalue vector of the synchronization signal and the eigenvalue vector of the fixed training mode sequence.

[0020] Optionally, the training mode signal sent by the master device is obtained, including:

[0021] A handshake process signal between the master device and the slave device is obtained.

[0022] The training mode signal is obtained by intercepting the handshake process signal according to a signal envelope threshold.

[0023] Optionally, the training mode signal is obtained by intercepting the handshake process signal according to a signal envelope threshold, including:

[0024] A first signal envelope threshold y th1 and a second signal envelope threshold yth2 determining a start position n1 and an end position n2 of the training mode signal;

[0025] obtaining the training mode signal from the handshake procedure signal according to the start position n1 and the end position n2;

[0026] wherein the start position n1 and the end position n2 satisfy:

[0027]

[0028] wherein, respectively represent the envelope amplitudes of the envelope of the handshake procedure signal at n1-1, n1, n2, n2+1.

[0029] Optionally, the obtaining the synchronization signal from the training mode signal according to the result of the cross-correlation operation and the length of the fixed training mode sequence comprises:

[0030] determining a start position of the synchronization signal according to the maximum cross-correlation value between the training mode signal and the fixed training mode sequence;

[0031] obtaining the synchronization signal from the training mode signal according to the start position of the synchronization signal, and the length of the synchronization signal is equal to the length of the fixed training mode sequence.

[0032] In a second aspect, the present application provides a full-duplex Ethernet device fingerprint extraction apparatus, comprising:

[0033] a signal obtaining module, configured to obtain a training mode signal sent by a master device;

[0034] a synchronization signal determining module, configured to perform a cross-correlation operation between the training mode signal and a fixed training mode sequence, and obtain a synchronization signal from the training mode signal according to a result of the cross-correlation operation and a length of the fixed training mode sequence;

[0035] a device fingerprint extraction module, configured to extract a device fingerprint of the master device according to a Wigner-Ville distribution feature of the synchronization signal and a Wigner-Ville distribution feature of the fixed training mode sequence.

[0036] In a third aspect, the present application 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 implements the full-duplex Ethernet device fingerprint extraction method of the first aspect when executing the program.

[0037] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the full-duplex Ethernet device fingerprint extraction method described in the first aspect above.

[0038] The present invention provides a method, apparatus, and electronic device for extracting full-duplex Ethernet device fingerprints. By acquiring the training mode signal emitted by the master device, synchronizing it to obtain a synchronization signal synchronized with a fixed training mode sequence, and extracting the device fingerprint of the master device based on the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training mode sequence, the device fingerprint of the master device is achieved stably. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in this 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 this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0040] Figure 1 This is a flowchart illustrating the fingerprint extraction method for full-duplex Ethernet devices provided by the present invention;

[0041] Figure 2 This is a schematic diagram of the signal acquisition method provided by the present invention;

[0042] Figure 3 This is a schematic diagram of the handshake process signals provided by the present invention;

[0043] Figure 4 This is a schematic diagram of the envelope signal obtained by performing a Hilbert transform on the handshake process signal and taking its amplitude, as provided by the present invention.

[0044] Figure 5 This is a schematic diagram of the training mode signal waveform emitted by the master device after coarse synchronization, provided by the present invention.

[0045] Figure 6 This is a schematic diagram of the fixed training mode sequence provided by the present invention;

[0046] Figure 7 This is a schematic diagram of the cross-correlation results between the training mode signal and the fixed training mode sequence provided by the present invention;

[0047] Figure 8 This is a schematic diagram of the synchronization signal after fine synchronization provided by the present invention;

[0048] Figure 9 This is a schematic diagram of the normalized synchronization signal provided by the present invention;

[0049] Figure 10 This is a schematic diagram of the four sub-signals of the normalized synchronization signal provided by the present invention;

[0050] Figure 11 This is a schematic diagram of the feature value vectors of the synchronization signal and the fixed training mode sequence provided by the present invention;

[0051] Figure 12 These are schematic diagrams of fingerprint curves from different devices provided by this invention;

[0052] Figure 13 This is a schematic diagram of the fingerprint extraction device for full-duplex Ethernet devices provided by the present invention;

[0053] Figure 14 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0055] Figure 1 This is a schematic diagram of the fingerprint extraction method for full-duplex Ethernet devices provided by the present invention, as shown below. Figure 1 As shown, the method includes the following steps:

[0056] Step 100: Obtain the training mode signal sent by the main device.

[0057] Step 101: Perform cross-correlation between the training mode signal and the fixed training mode sequence, and extract the synchronization signal from the training mode signal based on the result of the cross-correlation operation and the length of the fixed training mode sequence.

[0058] Step 102: Extract the device fingerprint of the master device based on the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training mode sequence.

[0059] Specifically, in response to the problem that the full-duplex signals sent by devices on both sides of the line overlap during stable communication of full-duplex Ethernet devices, resulting in unstable extracted device fingerprint information, this invention proposes a method based on Wigner-Ville distribution to extract features from the training mode signal sent by the master device to achieve stable extraction of the master device fingerprint.

[0060] First, it is necessary to obtain the training mode signal emitted by the master device. During the second phase of the link handshake process in a full-duplex Ethernet connection, the master device is in training mode, while the slave device is in silent mode. At this time, the slave device emits a zero signal, so the training mode signal emitted by the master device will not alias with the signal emitted by the slave device on the twisted pair. Therefore, obtaining the training mode signal emitted by the master device during the second phase of the link handshake process for device fingerprint extraction can avoid interference from full-duplex signal aliasing in the extraction of the master device's device fingerprint.

[0061] Optionally, the training mode signal sent by the master device can be obtained by directly sampling the signal within the corresponding time period of the second stage based on the time information of the handshake process, or by collecting the complete handshake process signal and extracting the training mode signal of the master device from the complete handshake process signal. The specific acquisition method is not limited here.

[0062] The training mode signal emitted by the master device is a fixed signal. Based on this fixed training mode signal, a fixed training mode sequence can be determined, which is used for feature comparison with the acquired training mode signal carrying device fingerprint information. Since the acquired training mode signal and the determined fixed training mode sequence may be out of sync or have inconsistent lengths, a cross-correlation operation is performed between the training mode signal and the fixed training mode sequence to facilitate feature comparison. Based on the result of the cross-correlation operation and the length of the fixed training mode sequence, a synchronization signal is obtained by truncating the training mode signal to achieve synchronization with the fixed training mode sequence and having the same signal length.

[0063] The Wigner-Ville distribution is defined as the Fourier transform of the instantaneous autocorrelation function of a signal. It reflects the instantaneous time-frequency relationship of the signal and has good time-frequency resolution. Compared with time-frequency methods such as short-time Fourier transform and wavelet transform, the Wigner-Ville distribution can better describe the time-varying characteristics of a signal. Therefore, after obtaining the synchronization signal, the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training pattern sequence can be calculated. Based on the Wigner-Ville distribution characteristics of the synchronization signal and the fixed training pattern sequence, the device fingerprint of the master device can be calculated.

[0064] The full-duplex Ethernet device fingerprint extraction method provided by this invention obtains the training mode signal sent by the master device, synchronizes it to obtain a synchronization signal that is synchronized with the fixed training mode sequence, and extracts the device fingerprint of the master device based on the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training mode sequence, thereby achieving stable extraction of the device fingerprint of the full-duplex Ethernet master device.

[0065] It should be noted that although the present invention uses a 100BASE-T1 device as an example for illustration, the full-duplex Ethernet device described in the present invention is not limited to a 100BASE-T1 device.

[0066] To facilitate a clearer understanding of the technical solutions of the various embodiments of the present invention, the working mode and handshake process of the 100BASE-T1 device are briefly described below.

[0067] (1) The three working modes of the full-duplex vehicle-mounted 100Mbps Ethernet 100BASE-T1 device and the corresponding transmitted signals

[0068] The 100BASE-T1 master device physical layer and the 100BASE-T1 slave device physical layer need to go through a link handshake to start the device. This process uses three different signals:

[0069] Quiet mode (Tx_mode = SEND_Z): Used to send a zero signal, that is, the link is inactive or in a quiet state.

[0070] Training mode (Tx_mode = SEND_I): Used to send PAM3 idle symbols.

[0071] Data mode (Tx_mode = SEND_N): Used for normal data transmission.

[0072] Where Tx_mode represents the device mode, and SEND_Z, SEND_I, and SEND_N represent quiet mode, training mode, and data mode, respectively.

[0073] (2) The handshake process between a full-duplex vehicle-mounted 100Mbps Ethernet 100BASE-T1 master and 100BASE-T1 slave device includes four stages.

[0074] Phase 1: Both the master device (M) and the slave device (S) are in quiet mode (Tx_mode = SEND_Z).

[0075] Phase 2: The master device (M) is in training mode (Tx_mode = SEND_I), and the slave device (S) is in quiet mode (Tx_mode = SEND_Z).

[0076] Phase 3: Both the master device (M) and the slave device (S) are in training mode (Tx_mode = SEND_I).

[0077] Phase 4: Both the master device (M) and the slave device (S) are in data mode (Tx_mode = SEND_N).

[0078] Optionally, based on the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training pattern sequence, the device fingerprint of the master device is extracted, including:

[0079] The synchronization signal is normalized to obtain a normalized synchronization signal;

[0080] The normalized synchronization signal is divided into Q sub-signals of equal length, and the eigenvalue vector of the synchronization signal is determined according to the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal; where Q is an integer greater than or equal to 1.

[0081] The fixed training pattern sequence is divided into Q subsequences of equal length, and the feature vector of the fixed training pattern sequence is determined based on the feature values ​​of the Wigner-Ville distribution matrix of each subsequence.

[0082] The device fingerprint of the master device is extracted based on the feature vector of the synchronization signal and the feature vector of the fixed training mode sequence.

[0083] Specifically, after obtaining the synchronization signal by extracting the training mode signal, the synchronization signal is first normalized to obtain the normalized synchronization signal. The normalized synchronization signal can be determined according to the following formula:

[0084]

[0085] Where n can represent the signal sampling point or time, normalize() represents the normalization method, and y IC (n) represents the synchronization signal. This represents the normalized synchronization signal.

[0086] Optionally, the normalization method represented by normalize() can be an amplitude normalization method, an energy normalization method, etc.

[0087] After obtaining the normalized synchronization signal, the normalized synchronization signal is divided into Q sub-signals, each of length L, as follows:

[0088]

[0089] Where n can represent the signal sampling point or time. This represents the sub-signal of the q-th normalized synchronization signal, where q can be a positive integer from 1 to Q.

[0090] Since the dimension of the Wigner-Ville distribution matrix depends on the length of the signal, dividing the normalized synchronization signal into Q sub-signals can reduce the computational complexity of the subsequent calculation of the Wigner-Ville distribution matrix. The normalized synchronization signal can be divided according to the length of the signal and the desired computational complexity. The specific number of sub-signals is not limited here.

[0091] Furthermore, the Wigner-Ville distribution matrix is ​​calculated for each of the Q sub-signals of the normalized synchronization signal.

[0092] The Wigner-Ville distribution matrix corresponding to the q-th sub-signal is calculated as follows:

[0093]

[0094] Where n can represent the signal sampling point or time, m can represent the sampling point offset or time shift, k represents the frequency point obtained after signal frequency discretization, L represents the sub-signal length, and j represents the imaginary unit. Indicates the q-th sub-signal in The signal value at that location, Indicates the q-th sub-signal in The conjugate value corresponding to the signal value at that location, WVD IC-q (n, k) represents the Wigner-Ville distribution matrix corresponding to the q-th sub-signal.

[0095] It should be noted that n and m represent corresponding values. When n represents the signal sampling point, m represents the sampling point offset; when n represents the time, m represents the time shift.

[0096] Then, singular value decomposition (SVDe) is performed on the Wigner-Ville distribution matrices corresponding to the Q sub-signals, reducing the signal features from two dimensions to one dimension. According to SVDe theory, for the Wigner-Ville distribution matrix WVD corresponding to the q-th sub-signal... IC-q For (n, k), there exist two corresponding orthogonal matrices U. IC-q and V IC-q and a corresponding diagonal matrix Λ IC-q ,satisfy:

[0097]

[0098] Among them, diag(λ1, λ2,...λ L ) represents the eigenvalues ​​as λ1, λ2, ..., λ. L A diagonal matrix.

[0099] diagonal matrix Λ IC-qeigenvalues ​​λ1, λ2, ..., λ L This is the eigenvalue of the Wigner-Ville distribution matrix corresponding to the q-th sub-signal that we are trying to solve.

[0100] After obtaining the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to the Q sub-signals, the eigenvalue vector of the synchronization signal is calculated by solving the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to these Q sub-signals.

[0101] Optionally, the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to the Q sub-signals can be summed, meanped, summed squared, differenced squared, or subtracted according to their corresponding positions in the diagonal matrix. The calculated results can then be rearranged according to their positions in the diagonal matrix to obtain the eigenvalue vector of the synchronization signal. The specific method is not limited here.

[0102] Similarly, the fixed training pattern sequence is divided into Q subsequences, each of length L, as follows:

[0103] {y C-q (n), q = 1, 2, ..., Q}

[0104] Where n can represent the signal sampling point or time, y C-q (n) represents the subsequence of the q-th fixed training pattern sequence, where q can be a positive integer from 1 to Q.

[0105] Since the dimension of the Wigner-Ville distribution matrix depends on the length of the signal, dividing the fixed training pattern sequence into Q subsequences can reduce the computational complexity of the subsequent calculation of the Wigner-Ville distribution matrix. The fixed training pattern sequence can be divided according to the length of the signal and the desired computational complexity. The specific number of subsequences is not limited here.

[0106] Furthermore, the Wigner-Ville distribution matrix is ​​calculated for each of the Q subsequences.

[0107] The Wigner-Ville distribution matrix corresponding to the qth subsequence is calculated as follows:

[0108]

[0109] Where n can represent the signal sampling point or time, m can represent the sampling point offset or time shift, k represents the frequency point obtained after signal frequency discretization, L represents the subsequence length, and j represents the imaginary unit. Indicates that the q-th subsequence is in The signal value at that location, Indicates that the q-th subsequence is in The conjugate value corresponding to the signal value at that location, WVD C-q (n, k) represents the Wigner-Ville distribution matrix corresponding to the q-th subsequence.

[0110] It should be noted that n and m represent corresponding values. When n represents the signal sampling point, m represents the sampling point offset; when n represents the time, m represents the time shift.

[0111] Then, singular value decomposition (SVDe) is performed on the Wigner-Ville distribution matrices corresponding to the Q subsequences, reducing the signal features from two dimensions to one dimension. According to SVDe theory, for the Wigner-Ville distribution matrix WVD corresponding to the q-th subsequence... C-q For (n, k), there exist two corresponding orthogonal matrices U. C-q and V C-q and a corresponding diagonal matrix Λ C-q ,satisfy:

[0112]

[0113] Among them, diag(λ1', λ2',...λ L ') represents the eigenvalues ​​as λ1', λ2', ..., λ L A diagonal matrix of '.

[0114] diagonal matrix Λ C-q The eigenvalues ​​λ1', λ2', ... λ L 'That is, the eigenvalues ​​of the Wigner-Ville distribution matrix corresponding to the q-th subsequence to be solved.

[0115] After obtaining the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to the Q subsequences, the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to these Q subsequences are calculated to obtain the eigenvalue vector of the fixed training pattern sequence.

[0116] Optionally, the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to the Q subsequences can be summed, meanped, summed squared, differenced squared, or subtracted according to their corresponding positions in the diagonal matrix. The calculated results can then be arranged according to their positions in the diagonal matrix to obtain the eigenvalue vector of the fixed training pattern sequence. The specific calculation method is not limited here.

[0117] It should be noted that the methods for dividing the normalized synchronization signal and the fixed training mode sequence are the same, and the methods for calculating the feature vectors of the synchronization signal and the fixed training mode sequence are also the same.

[0118] By normalizing the synchronization signal, a normalized synchronization signal is obtained, which reduces the eigenvalue vector error caused by potential signal deviations in the synchronization signal, making the extracted master device fingerprint more accurate. Furthermore, since the Wigner-Ville distribution can effectively describe the time-varying characteristics of a signal, and the dimension of the Wigner-Ville distribution matrix depends on the signal length, the normalized synchronization signal and the fixed training pattern sequence are respectively divided into sub-signals and sub-sequences. Based on the eigenvalues ​​of the Wigner-Ville distribution matrices of each sub-signal and sub-sequence, the eigenvalue vectors of the normalized synchronization signal and the fixed training pattern sequence are determined, and the master device fingerprint is further extracted. This reduces the computational complexity of device fingerprint extraction and improves the accuracy and stability of the device fingerprint.

[0119] Optionally, the eigenvalue vector of the synchronization signal is determined based on the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal, including:

[0120] The average value of the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal is calculated to obtain the eigenvalue vector of the synchronization signal;

[0121] Based on the eigenvalues ​​of the Wigner-Ville distribution matrix of each subsequence, the eigenvalue vector of the fixed training pattern sequence is determined, including:

[0122] The eigenvalues ​​of the Wigner-Ville distribution matrix of each subsequence are averaged to obtain the eigenvalue vector of the fixed training pattern sequence.

[0123] Specifically, after obtaining the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to the Q sub-signals divided by the normalized synchronization signal, the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to the Q sub-signals are summed and averaged according to their corresponding positions in the diagonal matrix. The averages calculated at each position are then rearranged according to their positions in the diagonal matrix to obtain the eigenvalue vector Λ of the synchronization signal. The specific calculation method is as follows:

[0124]

[0125] in, This involves summing the Q diagonal matrices corresponding to the Q sub-signals using matrix addition, and then multiplying the summed diagonal matrix by a coefficient. This means taking the diagonal elements of the matrix from left to right to form a vector, and Q represents the number of sub-signals.

[0126] For example, a normalized synchronization signal is divided into three sub-signals: sub-signal A, sub-signal B, and sub-signal C. The diagonal matrix corresponding to sub-signal A is diag(a1, a2, a3, a4), the diagonal matrix corresponding to sub-signal B is diag(b1, b2, b3, b4), and the diagonal matrix corresponding to sub-signal C is diag(c1, c2, c3, c4). Then, the eigenvector of the synchronization signal is...

[0127] Similarly, the eigenvalues ​​of the Wigner-Ville distribution matrices corresponding to the Q subsequences of the fixed training pattern sequence are summed and averaged according to their corresponding positions in the diagonal matrix. The averages calculated at each position are then rearranged according to their positions in the diagonal matrix to obtain the eigenvalue vector Λ of the fixed training pattern sequence. C The specific calculation method is as follows:

[0128]

[0129] in, This represents performing matrix addition on the Q diagonal matrices corresponding to the Q subsequences, and then multiplying the resulting diagonal matrix by a coefficient. This means taking the diagonal elements of the matrix from left to right to form a vector, where Q represents the number of subsequences.

[0130] For example, a fixed training pattern sequence is divided into three subsequences: subsequence D, subsequence E, and subsequence F. The diagonal matrix corresponding to subsequence D is diag(d1, d2), the diagonal matrix corresponding to subsequence E is diag(e1, e2), and the diagonal matrix corresponding to subsequence F is diag(f1, f2). Then, the feature vector of the fixed training pattern sequence is...

[0131] The present invention does not restrict the time order of the steps of calculating the feature vector of the fixed training mode sequence and the steps of calculating the feature vector of the synchronization signal; the two can be performed in a specific order or simultaneously.

[0132] By averaging the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal and sub-sequence, the eigenvalue vectors of the normalized synchronization signal and the fixed training mode sequence can be obtained, which can reduce the eigenvalue vector error caused by possible deviations in some sub-signals and sub-sequences.

[0133] Optionally, based on the feature vector of the synchronization signal and the feature vector of the fixed training pattern sequence, the device fingerprint of the master device is extracted, including:

[0134] The device fingerprint of the master device is obtained by dividing the corresponding element of the feature value vector of the synchronization signal by the corresponding element of the feature value vector of the fixed training mode sequence.

[0135] Specifically, the feature vector of the synchronization signal has the same dimension as the feature vector of the fixed training pattern sequence. The elements of the same dimension in both the synchronization signal and fixed training pattern sequence feature vectors are divided (i.e., the Kth dimension feature value of the synchronization signal feature vector is divided by the Kth dimension feature value of the fixed training pattern sequence feature vector). The results of these divisions are then rearranged according to their order in the feature vector to form the device fingerprint (DF). The specific calculation method is as follows:

[0136] DF = Λ. / Λ C

[0137] Where ". / " indicates element-wise division of vectors, and Λ represents the eigenvector of the synchronization signal. C This represents the feature value vector of a fixed training pattern sequence.

[0138] It should be noted that if there is a feature value of 0 in the feature value vector of a fixed training pattern sequence, then the result of dividing the element corresponding to the position with feature value of 0 is defined as 0.

[0139] For example, if the feature vector of the synchronization signal is (3, 2, 6) and the feature vector of the fixed training pattern sequence is (1, 0, 3), then the device fingerprint obtained by dividing the corresponding elements is: That is, (3, 0, 2).

[0140] The device fingerprint of the master device is obtained by dividing the feature value vector of the synchronization signal by the corresponding element of the feature value vector of the fixed training pattern sequence. This allows the device fingerprints of different devices to be easily distinguished based on the feature value vector of the fixed training pattern sequence.

[0141] Optionally, the training mode signal emitted by the master device is acquired, including:

[0142] Acquire handshake process signals between the master and slave devices;

[0143] The training mode signal is obtained by extracting the handshake process signal based on the signal envelope threshold.

[0144] Specifically, to obtain the training mode signal sent by the master device, the handshake process signal between the master and slave devices can be obtained first. The handshake process signal contains the training mode signal sent by the master device in the second phase of the link handshake process, so the training mode signal can be obtained by intercepting the handshake process signal.

[0145] Optionally, obtaining the handshake process signal between the master and slave devices can be achieved by sampling the transmission signal on the twisted pair. For example, the symbol rate of the full-duplex Ethernet handshake process signal is 66.6MHz, and the transmission signal can be sampled at a sampling rate of 1332MHz (20 times oversampling) to obtain the handshake process signal. The specific sampling rate is not limited here.

[0146] After acquiring the handshake process signal, due to the signal superposition factor, the signal envelope amplitudes of the four stages in the link handshake process are different. Therefore, the signal envelope threshold can be determined based on the envelope amplitude corresponding to the handshake process signal, and the training mode signal can be obtained by further extracting the handshake process signal based on the determined envelope threshold.

[0147] By directly acquiring the handshake process signal between the master and slave devices, and extracting the handshake process signal according to the signal envelope threshold to obtain the training mode signal, the training mode signal can be conveniently and effectively acquired when the master-slave link handshake process is triggered.

[0148] Optionally, the training mode signal is obtained by extracting the handshake process signal based on the signal envelope threshold, including:

[0149] According to the first signal envelope threshold y th1 Second signal envelope threshold y th2 Determine the start position n1 and end position n2 of the training mode signal;

[0150] Based on the start position n1 and the end position n2, the handshake process signal is extracted to obtain the training mode signal;

[0151] Wherein, the starting position n1 and the ending position n2 satisfy:

[0152]

[0153] In the formula, These represent the envelope amplitudes of the handshake process signal at points n1-1, n1, n2, and n2+1, respectively.

[0154] Specifically, the first signal envelope threshold y th1 Second signal envelope threshold y th2 It is determined based on the envelope amplitude corresponding to the training mode signal, the first signal envelope threshold y th1 Less than the second signal envelope threshold y th2 Only when the envelope amplitude of the training mode signal is within the first signal envelope threshold y th1 Second signal envelope threshold y th2 between.

[0155] Therefore, it is possible to determine the envelope corresponding to the handshake process signal. The envelope amplitude at n1-1 Less than the first signal envelope threshold y th1 The envelope amplitude at n1 Greater than or equal to the first signal envelope threshold y th1 The starting position n1 is uniquely determined.

[0156] Similarly, the envelope corresponding to the handshake signal can be used as a basis. The envelope amplitude at n2 Less than the second signal envelope threshold y th2 The envelope amplitude at n2+1 Greater than or equal to the second signal envelope threshold y th2 The termination position n2 is uniquely determined.

[0157] Where n1 and n2 can be signal sampling points or times.

[0158] Alternatively, the envelope of the handshake process signal can be obtained by performing a Hilbert transform and taking its amplitude. The specific calculation method is as follows:

[0159]

[0160] Where n can represent the signal sampling point or time. Let |·| denote the Hilbert transform, |·| denote the amplitude, and y(n) denote the handshake signal. y(-2n-1) represents the envelope corresponding to the handshake process signal, and y(-2n-1) represents the signal value of the handshake process signal at -2n-1.

[0161] After determining the start position n1 and end position n2, the handshake process signal can be directly extracted from the start position n1 and end position n2 to obtain the training mode signal. The training mode signal y sent by the master device... I (n), satisfying:

[0162] y I (n)=y(n), n1≤n≤n2

[0163] By setting the first signal envelope threshold y th1 Second signal envelope threshold y th2 This allows us to determine the start position n1 and end position n2 of the training mode signal, and thus accurately extract the handshake process signal based on the start position n1 and end position n2 to obtain the complete training mode signal.

[0164] Optionally, based on the result of the cross-correlation operation and the length of the fixed training pattern sequence, a synchronization signal is obtained by truncating the training pattern signal, including:

[0165] The starting position of the synchronization signal is determined based on the maximum cross-correlation value between the training mode signal and the fixed training mode sequence.

[0166] Based on the starting position of the synchronization signal, the training mode signal is extracted to obtain the synchronization signal. The length of the synchronization signal is equal to the length of the fixed training mode sequence.

[0167] Specifically, to synchronize the training mode signal and the fixed training mode sequence, a cross-correlation operation is performed between the training mode signal and the fixed training mode sequence to find the position 'start' with the highest correlation during the shifting process. The calculation method for 'start' is as follows:

[0168]

[0169] Where n can represent the signal sampling point or time, m can represent the sampling point offset or time shift, N represents the length of the training mode signal, and y C (n+m) represents the signal value at position n+m of the fixed training pattern sequence, y I * (n) represents the conjugate value of the training mode signal at position n, R CM (m) represents the cross-correlation value.

[0170] It should be noted that n and m represent corresponding values. When n represents the signal sampling point, m represents the sampling point offset; when n represents the time, m represents the time shift.

[0171] When the cross-correlation value reaches its maximum at the start point, it indicates that the training pattern sequence y is fixed at this point. C (n) and training mode signal y I (n) is the closest, thus finding the fixed training pattern sequence y. C (n) in the training mode signal y I The starting position in (n) is used to train the mode signal y. I (n) Extract a segment of the sequence y that corresponds to the fixed training pattern. C (n) Signals of equal length are used to obtain the synchronization signal y. IC (n). Synchronization signal y IC (n) is shown below:

[0172] y IC (n)=y I (n), n∈(start, start+Len-1]

[0173] Here, Len represents the signal length of the fixed training pattern sequence.

[0174] The starting position of the fixed training pattern sequence in the training pattern signal is obtained by cross-correlation operation. Then, starting from the starting position, a signal of the same length as the fixed training pattern sequence is extracted from the training pattern signal to obtain a synchronization signal that is synchronized with the fixed training pattern sequence and of the same length. Thus, the training pattern signal and the fixed training pattern can be conveniently compared in terms of features through the synchronization signal.

[0175] The methods provided in the above embodiments of the present invention will be illustrated below through specific application scenarios.

[0176] Step 1: Connect the full-duplex Ethernet master and slave devices to trigger the link handshake process.

[0177] In this embodiment, the 100BASE-T1 master device and the 100BASE-T1 slave device are connected by a twisted pair cable. Figure 2 This is a schematic diagram of the signal acquisition method provided by the present invention, as shown below. Figure 2 As shown, the signal acquisition device is located near the slave device end, i.e. Figure 2 Signal acquisition is performed at point A. Upon power-up, the master and slave devices initiate synchronization via a link handshake at the physical layer. Alternatively, during normal communication between the master and slave devices, a short-duration interference signal can be inserted to disrupt the communication and re-trigger the training process for synchronization.

[0178] It should be noted that although this embodiment uses a 100BASE-T1 device as an example for illustration, the full-duplex Ethernet device described in this invention is not limited to a 100BASE-T1 device.

[0179] Step 2: Acquire handshake process signals and capture the training mode signals sent by the master device through coarse synchronization.

[0180] The training process signal is acquired at a sampling rate of 1332MHz. Since the symbol rate of the 100BASE-T1 signal is 66.6MHz, this represents a 20x oversampling. The master and slave physical layers need to initiate a link handshake to start the device; this process uses three different signals:

[0181] (1) Quiet mode (Tx_mode = SEND_Z): Used to send zero signal, that is, the link is inactive or in a quiet state.

[0182] (2) Training mode (Tx_mode = SEND_I): Used to send PAM3 idle symbols.

[0183] (3) Data mode (Tx_mode = SEND_N): Used for normal data transmission.

[0184] In this embodiment, the handshake process signal collected is denoted as y(n). Figure 3 This is a schematic diagram of the handshake process signals provided by the present invention, such as... Figure 3 As shown, the handshake process signals include the following four stages:

[0185] (1) Both the master device (M) and the slave device (S) are in quiet mode (Tx_mode=

[0186] SEND_Z).

[0187] (2) The master device (M) is in training mode (Tx_mode = SEND_I), and the slave device (S) is in quiet mode (Tx_mode = SEND_Z).

[0188] (3) Both the master device (M) and the slave device (S) are in training mode (Tx_mode=

[0189] SEND_I).

[0190] (4) Both the master device (M) and the slave device (S) are in data mode (Tx_mode=

[0191] SEND_N).

[0192] Due to signal superposition, the envelope amplitudes of the signals differ across the four stages. The envelope of signal y(n) is obtained by performing a Hilbert transform and taking the amplitude values.

[0193]

[0194] in, represents the Hilbert transform, and |·| represents taking the amplitude. Figure 4 This is a schematic diagram of the envelope signal obtained by performing a Hilbert transform on the handshake process signal provided by the present invention and taking its amplitude. The waveform is as follows Figure 4 As shown. By setting the envelope threshold y th1 =1.5 and y th2 =2.5 for coarse synchronization, intercepting the second-stage handshake process signal, i.e., the training mode signal y sent by the master device. I (n), satisfying:

[0195]

[0196] Figure 4 In this context, n1 is approximately 5.2 × 10⁻⁶. 6 +1, n² is approximately 10 × 10 6 . Figure 5This is a schematic diagram of the training mode signal waveform emitted by the master device after coarse synchronization, as provided by the present invention. The obtained training mode signal y is extracted. I The waveform of (n) is as follows Figure 5 As shown in (a), the y values ​​obtained from the first 1000 sampling points I The waveform of (n) is as follows Figure 5 As shown in (b) of the diagram.

[0197] Step 3: Based on the known fixed training mode sequence, perform fine synchronization of the training mode signal of the master device to obtain the synchronization signal.

[0198] Figure 6 This is a schematic diagram of the fixed training pattern sequence provided by the present invention. The fixed training pattern sequence obtained by sampling 100,000 sampling points is as follows: Figure 6 As shown in (a), the fixed training pattern sequence obtained from the first 400 sampling points is as follows: Figure 6 As shown in (b) of the diagram. By fixing the training pattern sequence y C (n) and training mode signal y I (n) Perform cross-correlation calculations for precise synchronization; the cross-correlation value R CM The point with the largest (m) is the starting point of the synchronization signal.

[0199]

[0200] Where N represents y I The length of (n) is approximately 4.8 × 10⁻⁶. 6 . Figure 7 This is a schematic diagram illustrating the cross-correlation results between the training mode signal and the fixed training mode sequence provided by the present invention, as shown below. Figure 7 As shown, R CM (m) reaches its maximum value at 11254, indicating that y I The 11254th sampling point of (n) is the starting point of the synchronization signal. The synchronization signal y is generated by extracting a signal of length 100,000 sampling points from this position. IC (n) is used for subsequent fingerprint extraction.

[0201] y IC (n)=y I (n), n∈[11254, 111254]

[0202] Figure 8 This is a schematic diagram of the synchronization signal after fine synchronization provided by the present invention. The synchronization signal obtained by sampling 100,000 sampling points is as follows: Figure 8 As shown in (a) above, the synchronization signal obtained from the first 400 sampling points is shown in (b). It can be seen that the synchronization signal y... IC (n) and the fixed training pattern sequence yC (n) has been synchronized.

[0203] Step 4: Normalize the synchronization signal and divide it into several sub-signals of equal length. Divide the fixed training mode sequence into several sub-sequences of equal length in the same way.

[0204] In this embodiment, the synchronization signal y IC (n) is normalized to obtain the normalized synchronization signal. The normalization method is as follows:

[0205]

[0206] Then, Divide into 10,000 sub-signals, each with a length of 10, and then... C (n) is divided into 10000 subsequences, each of length 10, which are respectively

[0207] and {y C-q (n), q=1, 2,...,10000}

[0208] Since the dimension of the Wigner-Ville distribution matrix depends on the length of the signal, and y C Dividing (n) into multiple sub-signals and sub-sequences can reduce the computational complexity of the Wigner-Ville distribution matrix in step 5.

[0209] Figure 9 This is a schematic diagram of the normalized synchronization signal provided by the present invention. The normalized synchronization signal obtained by sampling 100,000 sampling points is as follows. Figure 9 As shown in (a), the normalized synchronization signal obtained from the first 400 sampling points is as follows: Figure 9 As shown in (b) of the diagram.

[0210] Figure 10 This is a schematic diagram of the four sub-signals of the normalized synchronization signal provided by the present invention. The first four sub-signals of the normalized synchronization signal... Each as Figure 10 As shown in (a), (b), (c), and (d) in the figure.

[0211] Step 5: Calculate the Wigner-Ville distribution of each sub-signal of the synchronization signal and the fixed training mode sequence to obtain the Wigner-Ville distribution matrix. Perform SVD singular value decomposition on each distribution matrix to obtain the corresponding eigenvalue vector. Take the average of the eigenvalue vectors of the sub-signals contained in the synchronization signal and the fixed training mode sequence to obtain the eigenvalue vector of the synchronization signal and the eigenvalue vector of the fixed training mode sequence.

[0212] In this embodiment, the Wigner-Ville distribution is defined as the Fourier transform of the instantaneous autocorrelation function of the signal, reflecting the instantaneous time-frequency relationship of the signal and possessing good time-frequency resolution. Compared to time-frequency methods such as short-time Fourier transform and wavelet transform, the Wigner-Ville distribution can better describe the time-varying characteristics of the signal. Each sub-signal of the normalized synchronization signal... Wigner-Ville distribution matrix (WVD) IC-q (n, k) and each subsequence y of the fixed training mode sequence C-q WVD of the Wigner-Ville distribution matrix of (n) C-q The calculation methods for (n, k) are as follows:

[0213]

[0214] Then, through singular value decomposition, the signal features are reduced from two dimensions to one dimension. According to singular value decomposition theory, for a matrix WVD... IC-q There exist two orthogonal matrices U (n, k). IC-q and V IC-q and a diagonal array Λ IC-q ,satisfy:

[0215]

[0216] Among them, Λ IC-q eigenvalues ​​λ1, λ2, ..., λ 10 This is the Wigner-Ville distribution matrix (WVD) of the q-th sub-signal of the synchronization signal that we are trying to solve. IC-q Eigenvalues ​​of (n, k).

[0217] The feature value of the q-th subsequence of the fixed training pattern sequence is solved in the same way as above, denoted as Λ. C-q .

[0218] Then respectively for Λ IC-q and Λ C-q By averaging, we obtain the feature vector Λ of the synchronization signal and the feature vector Λ of the fixed training mode sequence. C This improves the stability of fingerprints.

[0219]

[0220]

[0221] Figure 11 This is a schematic diagram of the feature value vectors of the synchronization signal and the fixed training mode sequence provided by the present invention. The calculated Λ curve is shown below. Figure 11 As shown in (a), the calculated Λ C Curves Figure 11 As shown in (b) of the diagram.

[0222] Step 6: Divide the eigenvector of the synchronization signal by the eigenvector of the fixed ideal symbol sequence to obtain the device fingerprint DF.

[0223] DF = Λ. / Λ C

[0224] The ". / " symbol represents the division of corresponding elements of the vector. When the denominator is 0, the result of the division is defined as 0. Figure 12 The fingerprint curve diagrams provided by this invention are specifically fingerprint curve diagrams extracted from three devices: device 1, device 2, and device 3. Each device extracted fingerprints five times. Figure 12 As can be seen, the fingerprint curves of device 1, device 2 and device 3 are basically overlapping, indicating stability, and the fingerprint curves of the three devices can be clearly distinguished from each other.

[0225] The fingerprint extraction device for full-duplex Ethernet devices provided by the present invention is described below. The fingerprint extraction device for full-duplex Ethernet devices described below can be referred to in correspondence with the fingerprint extraction method for full-duplex Ethernet devices described above.

[0226] Figure 13 This is a schematic diagram of the full-duplex Ethernet device fingerprint extraction device provided by the present invention, as shown below. Figure 13 As shown, the device includes:

[0227] The signal acquisition module 1300 is used to acquire the training mode signal sent by the main device;

[0228] The synchronization signal determination module 1310 is used to perform cross-correlation operation between the training mode signal and the fixed training mode sequence, and to extract the training mode signal to obtain the synchronization signal based on the result of the cross-correlation operation and the length of the fixed training mode sequence.

[0229] The device fingerprint extraction module 1320 is used to extract the device fingerprint of the master device based on the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training mode sequence.

[0230] Optionally, based on the Wigner-Ville distribution characteristics of the synchronization signal and the Wigner-Ville distribution characteristics of the fixed training pattern sequence, the device fingerprint of the master device is extracted, including:

[0231] The synchronization signal is normalized to obtain a normalized synchronization signal;

[0232] The normalized synchronization signal is divided into Q sub-signals of equal length, and the eigenvalue vector of the synchronization signal is determined according to the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal; where Q is an integer greater than or equal to 1.

[0233] The fixed training pattern sequence is divided into Q subsequences of equal length, and the feature vector of the fixed training pattern sequence is determined based on the feature values ​​of the Wigner-Ville distribution matrix of each subsequence.

[0234] The device fingerprint of the master device is extracted based on the feature vector of the synchronization signal and the feature vector of the fixed training mode sequence.

[0235] Optionally, the eigenvalue vector of the synchronization signal is determined based on the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal, including:

[0236] The average value of the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal is calculated to obtain the eigenvalue vector of the synchronization signal;

[0237] Based on the eigenvalues ​​of the Wigner-Ville distribution matrix of each subsequence, the eigenvalue vector of the fixed training pattern sequence is determined, including:

[0238] The eigenvalues ​​of the Wigner-Ville distribution matrix of each subsequence are averaged to obtain the eigenvalue vector of the fixed training pattern sequence.

[0239] Optionally, based on the feature vector of the synchronization signal and the feature vector of the fixed training pattern sequence, the device fingerprint of the master device is extracted, including:

[0240] The device fingerprint of the master device is obtained by dividing the corresponding element of the feature value vector of the synchronization signal by the corresponding element of the feature value vector of the fixed training mode sequence.

[0241] Optionally, the training mode signal emitted by the master device is acquired, including:

[0242] Acquire handshake process signals between the master and slave devices;

[0243] The training mode signal is obtained by extracting the handshake process signal based on the signal envelope threshold.

[0244] Optionally, the training mode signal is obtained by extracting the handshake process signal based on the signal envelope threshold, including:

[0245] According to the first signal envelope threshold y th1 Second signal envelope threshold y th2 Determine the start position n1 and end position n2 of the training mode signal;

[0246] Based on the start position n1 and the end position n2, the handshake process signal is extracted to obtain the training mode signal;

[0247] Wherein, the starting position n1 and the ending position n2 satisfy:

[0248]

[0249] In the formula, These represent the envelope amplitudes of the handshake process signal at points n1-1, n1, n2, and n2+1, respectively.

[0250] Optionally, based on the result of the cross-correlation operation and the length of the fixed training pattern sequence, a synchronization signal is obtained by truncating the training pattern signal, including:

[0251] The starting position of the synchronization signal is determined based on the maximum cross-correlation value between the training mode signal and the fixed training mode sequence.

[0252] Based on the starting position of the synchronization signal, the training mode signal is extracted to obtain the synchronization signal. The length of the synchronization signal is equal to the length of the fixed training mode sequence.

[0253] It should be noted that the device provided by the present invention can implement all the method steps implemented in the above method embodiments and can achieve the same technical effect. Therefore, the parts and beneficial effects that are the same as those in the method embodiments will not be described in detail here.

[0254] Figure 14 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 14 As shown, the electronic device may include a processor 1410, a communications interface 1420, a memory 1430, and a communication bus 1440, wherein the processor 1410, the communications interface 1420, and the memory 1430 communicate with each other via the communication bus 1440. The processor 1410 can call logical instructions in the memory 1430 to execute any of the full-duplex Ethernet device fingerprint extraction methods provided in the above embodiments.

[0255] Furthermore, the logical instructions in the aforementioned memory 1430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a 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, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0256] It should be noted that the electronic device provided by the present invention can implement all the method steps implemented in the above method embodiments and can achieve the same technical effect. Therefore, the parts and beneficial effects that are the same as those in the method embodiments will not be described in detail here.

[0257] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform any of the full-duplex Ethernet device fingerprint extraction methods provided in the above embodiments.

[0258] It should be noted that the non-transitory computer-readable storage medium provided by the present invention can implement all the method steps implemented in the above method embodiments and can achieve the same technical effect. Here, the parts that are the same as those in the method embodiments and the beneficial effects will not be described in detail.

[0259] The device embodiments described above are merely illustrative. 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 network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0260] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0261] 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 of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for fingerprint extraction from a full-duplex Ethernet device, characterized in that, include: Acquire the training mode signal sent by the master device; A cross-correlation operation is performed between the training mode signal and the fixed training mode sequence, and a synchronization signal is obtained by truncating the training mode signal based on the result of the cross-correlation operation and the length of the fixed training mode sequence. The synchronization signal is normalized to obtain a normalized synchronization signal; The normalized synchronization signal is divided into Q sub-signals of equal length, and the eigenvalue vector of the synchronization signal is determined according to the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal; where Q is an integer greater than or equal to 1. The fixed training pattern sequence is divided into Q subsequences of equal length, and the feature value vector of the fixed training pattern sequence is determined according to the feature values ​​of the Wigner-Ville distribution matrix of each subsequence. The device fingerprint of the master device is extracted based on the feature value vector of the synchronization signal and the feature value vector of the fixed training mode sequence.

2. The fingerprint extraction method for full-duplex Ethernet devices according to claim 1, characterized in that, Determining the eigenvalue vector of the synchronization signal based on the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal includes: The average value of the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal is calculated to obtain the eigenvalue vector of the synchronization signal; Determining the feature vector of the fixed training mode sequence based on the feature values ​​of the Wigner-Ville distribution matrix of each sub-sequence includes: The eigenvalues ​​of the Wigner-Ville distribution matrix of each subsequence are averaged to obtain the eigenvalue vector of the fixed training mode sequence.

3. The fingerprint extraction method for full-duplex Ethernet devices according to claim 1 or 2, characterized in that, The step of extracting the device fingerprint of the master device based on the feature value vector of the synchronization signal and the feature value vector of the fixed training mode sequence includes: The device fingerprint of the master device is obtained by dividing the feature value vector of the synchronization signal by the corresponding element of the feature value vector of the fixed training mode sequence.

4. The fingerprint extraction method for full-duplex Ethernet devices according to claim 3, characterized in that, When there is a feature value of 0 in the feature value vector of the fixed training mode sequence, the result of dividing the corresponding element at the position of the feature value of 0 is defined as 0.

5. The fingerprint extraction method for full-duplex Ethernet devices according to claim 1, characterized in that, The acquisition of the training mode signal emitted by the master device includes: Acquire handshake process signals between the master and slave devices; The training mode signal is obtained by extracting the handshake process signal based on the signal envelope threshold.

6. The fingerprint extraction method for full-duplex Ethernet devices according to claim 5, characterized in that, The step of extracting the handshake process signal to obtain the training mode signal based on the signal envelope threshold includes: According to the first signal envelope threshold y th1 Second signal envelope threshold y th2 Determine the start position n1 and end position n2 of the training mode signal; Based on the start position n1 and the end position n2, the handshake process signal is extracted to obtain the training mode signal; Wherein, the starting position n1 and the ending position n2 satisfy: In the formula, These represent the envelope amplitudes of the signal envelopes at points n1-1, n1, n2, and n2+1, respectively, during the handshake process.

7. The fingerprint extraction method for full-duplex Ethernet devices according to claim 1, characterized in that, The step of extracting a synchronization signal from the training pattern signal based on the result of the cross-correlation operation and the length of the fixed training pattern sequence includes: The starting position of the synchronization signal is determined based on the maximum cross-correlation value between the training mode signal and the fixed training mode sequence. Based on the starting position of the synchronization signal, the training mode signal is extracted to obtain a synchronization signal, the length of which is equal to the length of the fixed training mode sequence.

8. A fingerprint extraction device for a full-duplex Ethernet device, characterized in that, include: The signal acquisition module is used to acquire the training mode signal emitted by the main device; The synchronization signal determination module is used to perform cross-correlation operation between the training mode signal and the fixed training mode sequence, and to extract the training mode signal to obtain the synchronization signal based on the result of the cross-correlation operation and the length of the fixed training mode sequence. The device fingerprint extraction module is used to normalize the synchronization signal to obtain a normalized synchronization signal. The normalized synchronization signal is divided into Q sub-signals of equal length, and the eigenvalue vector of the synchronization signal is determined based on the eigenvalues ​​of the Wigner-Ville distribution matrix of each sub-signal; where Q is an integer greater than or equal to 1. The fixed training pattern sequence is divided into Q sub-sequences of equal length, and the eigenvalue vector of the fixed training pattern sequence is determined based on the eigenvalue vectors of the Wigner-Ville distribution matrix of each sub-sequence. The device fingerprint of the master device is extracted based on the eigenvalue vector of the synchronization signal and the eigenvalue vector of the fixed training pattern sequence.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the device fingerprint extraction method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the device fingerprint extraction method as described in any one of claims 1 to 7.

Images