Device access authentication method and device, computer device and storage medium

Abstract

This application relates to a device access authentication method, apparatus, computer device, computer-readable storage medium, and computer program product. The method includes: acquiring the network interface card (NIC) signal of the device to be accessed, obtaining an original input signal sequence and a reconstructed reference signal sequence; inputting the original input signal sequence in parallel to a first filter and a second filter, wherein the step size of the first filter is larger than the step size of the second filter; calculating the current output error of the first filter based on the reconstructed reference signal sequence and the first filtered signal; if the current output error exceeds a preset error threshold, updating the tap coefficients of the first filter based on the tap coefficients of the second filter; and inputting the fingerprint feature information obtained based on the tap coefficients of the first filter into a classification and recognition model to obtain an authentication result. Using this method, a brief fingerprint of the device can be obtained with a small number of training samples, improving the real-time performance of device access authentication.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a device access authentication method, apparatus, computer equipment, storage medium, and computer program product. Background Technology

[0002] Current fingerprint extraction technologies are only suitable for low-speed network environments. In the gigabit Ethernet field, related technologies based on statistical feature extraction such as spectrum and time domain share a common challenge: a large number of signal samples are required for effective fingerprint extraction. In addition, the high feature dimensionality of fingerprints leads to high fingerprint extraction time and computational costs. Higher fingerprint extraction latency usually means longer time to determine the identity of devices accessing the network, and greater risk of loss when attacked by spoofed devices.

[0003] Related technologies apply adaptive filtering methods to Gigabit Ethernet scenarios, enabling fingerprint extraction from fewer signal samples and improving the real-time performance of device fingerprint extraction, thereby effectively reducing the cost of device access authentication. However, current access authentication technologies based on adaptive filtering for fingerprint extraction suffer from difficulty in converging adaptive filtering errors, resulting in authentication real-time performance that fails to meet usage requirements. Summary of the Invention

[0004] Therefore, it is necessary to provide a device access authentication method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the real-time performance of device access authentication in response to the above-mentioned technical problems.

[0005] Firstly, this application provides a device access authentication method, including:

[0006] The network card signal of the device to be connected is acquired, and the network card signal is preprocessed to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1.

[0007] For each original input signal sequence, the original input signal sequence is input in parallel to a first filter and a second filter to obtain a first filtered signal output by the first filter and a second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter.

[0008] The current output error of the first filter is calculated based on the reconstructed reference signal sequence and the first filtered signal.

[0009] If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter remain unchanged. This process continues until all N original input signal sequences have been processed by the first filter and the second filter, at which point the tap coefficients of the first filter are obtained.

[0010] The fingerprint feature information obtained based on the tap coefficients of the first filter is input into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0011] In one embodiment, if the current output error exceeds a preset error threshold, the method further includes:

[0012] The error data of the second filter is calculated based on the reconstructed reference signal sequence and the second filtered signal, and the current output error of the first filter is replaced by the error data of the second filter.

[0013] In one embodiment, the step of inputting the fingerprint feature information obtained based on the tap coefficients of the first filter into the classification and recognition model to obtain the authentication result of the device to be accessed includes:

[0014] The tap coefficient of the first filter for each is used as an initial fingerprint feature;

[0015] The target fingerprint feature is obtained by performing a weighted average calculation on the N initial fingerprint features;

[0016] The target fingerprint features are input into a classification and recognition model to obtain the authentication result of the device to be accessed.

[0017] In one embodiment, the training method of the classification and recognition model includes:

[0018] Obtain a historical authentication dataset; wherein the historical authentication dataset includes a historical fingerprint feature set and an identity label, and the historical fingerprint feature set includes multiple fingerprint feature samples;

[0019] The initial classification model is trained based on the identity label and the fingerprint feature sample to obtain a classification and recognition model.

[0020] In one embodiment, the method for obtaining the historical fingerprint feature set includes:

[0021] Multiple sets of historical signals are acquired; each set of historical signals includes multiple historical input signal sequences and corresponding historical reconstructed signal sequences.

[0022] For each set of historical signals, perform the following steps:

[0023] Each historical input signal sequence in the historical signal set is input in parallel to the first filter and the second filter to obtain the first historical signal output by the first filter and the second historical signal corresponding to the second filter, respectively.

[0024] The first output error of the first filter is calculated based on the historical reconstructed signal sequence and the first historical signal.

[0025] If the first output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the first output error does not exceed the preset error threshold, the tap coefficients of the first filter remain unchanged. This process continues until all N historical input signal sequences have been processed by the first filter and the second filter, at which point the tap coefficients of the first filter are obtained.

[0026] Fingerprint feature samples are obtained based on the tap coefficients of the first filter;

[0027] By combining the fingerprint feature samples corresponding to multiple sets of historical signal sets, a historical fingerprint feature set is obtained.

[0028] In one embodiment, the preprocessing of the network interface card (NIC) signal to obtain N original input signal sequences and corresponding reconstructed reference signal sequences includes:

[0029] The signal segment of the network card signal is selected and divided into N level intervals, which are used as N original input signal sequences;

[0030] For each of the stated level intervals, the level values ​​at the first and second endpoints of the level intervals are extracted respectively.

[0031] Based on the first endpoint level value and the second endpoint level value, the level within the level interval is reconstructed to obtain N reconstructed reference signal sequences.

[0032] In one embodiment, the first endpoint level value is higher than the second endpoint level value; the step of reconstructing the level within the level interval based on the first endpoint level value and the second endpoint level value includes:

[0033] Calculate the average value of the first endpoint level and the second endpoint level;

[0034] The level signals in the level range that are higher than the average value are reconstructed into the level signals corresponding to the first endpoint level value;

[0035] The level signals in the level range that are lower than the average value are reconstructed into level signals corresponding to the second endpoint level value.

[0036] Secondly, this application also provides a device access authentication apparatus, comprising:

[0037] The signal acquisition module is used to acquire the network card signal of the device to be connected, and to preprocess the network card signal to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1.

[0038] The signal processing module is used to input the original input signal sequence in parallel to a first filter and a second filter for each original input signal sequence, so as to obtain a first filtered signal output by the first filter and a second filtered signal corresponding to the second filter, respectively, wherein the step size of the first filter is greater than the step size of the second filter;

[0039] The data calculation module is used to calculate the current output error of the first filter based on the reconstructed reference signal sequence and the first filtered signal.

[0040] The data update module is used to update the tap coefficients of the first filter according to the tap coefficients of the second filter when the current output error exceeds a preset error threshold, and to keep the tap coefficients of the first filter unchanged when the current output error does not exceed the preset error threshold; until all N original input signal sequences have been processed by the first filter and the second filter, the tap coefficients of the first filter are obtained.

[0041] The device authentication module is used to input the fingerprint feature information obtained based on the tap coefficients of the first filter into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0042] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0043] The network card signal of the device to be connected is acquired, and the network card signal is preprocessed to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1.

[0044] For each original input signal sequence, the original input signal sequence is input in parallel to a first filter and a second filter to obtain a first filtered signal output by the first filter and a second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter.

[0045] Based on the reconstructed reference signal sequence and the first filtered signal, the current output error of the first filter is calculated. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter are kept unchanged.

[0046] The process continues until all N original input signal sequences have been processed by the first and second filters, at which point the tap coefficients of the first filter are obtained.

[0047] The fingerprint feature information obtained based on the tap coefficients of the first filter is input into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0048] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0049] The network card signal of the device to be connected is acquired, and the network card signal is preprocessed to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1.

[0050] For each original input signal sequence, the original input signal sequence is input in parallel to a first filter and a second filter to obtain a first filtered signal output by the first filter and a second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter.

[0051] Based on the reconstructed reference signal sequence and the first filtered signal, the current output error of the first filter is calculated. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter are kept unchanged.

[0052] The process continues until all N original input signal sequences have been processed by the first and second filters, at which point the tap coefficients of the first filter are obtained.

[0053] The fingerprint feature information obtained based on the tap coefficients of the first filter is input into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0054] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0055] The network card signal of the device to be connected is acquired, and the network card signal is preprocessed to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1.

[0056] For each original input signal sequence, the original input signal sequence is input in parallel to a first filter and a second filter to obtain a first filtered signal output by the first filter and a second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter.

[0057] Based on the reconstructed reference signal sequence and the first filtered signal, the current output error of the first filter is calculated. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter are kept unchanged.

[0058] The process continues until all N original input signal sequences have been processed by the first and second filters, at which point the tap coefficients of the first filter are obtained.

[0059] The fingerprint feature information obtained based on the tap coefficients of the first filter is input into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0060] The aforementioned device access authentication method, apparatus, computer equipment, storage medium, and computer program product preprocess the network card signal of the device to be accessed to obtain N original input signal sequences and corresponding reconstructed reference signal sequences. The method sets a first filter with a larger step size for more effective signal convergence and a second filter with a smaller step size to avoid error divergence. Each original input signal sequence is input in parallel to the first and second filters, resulting in a first filtered signal output by the first filter and a corresponding second filtered signal output by the second filter. During signal output, the current output error of the first filter is calculated based on the reconstructed reference signal sequence and the first filtered signal. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated based on the tap coefficients of the second filter. If the output error does not exceed the preset error threshold, the tap coefficients of the first filter remain unchanged. Since the step size of the first filter is larger than that of the second filter, when the tap coefficients of the first filter show signs of pre-divergence due to the large step size during the convergence process of certain signals, the tap coefficients of the first filter can be corrected based on the more accurate tap coefficients of the second filter based on the convergence result. This process continues until all N original input signal sequences have been processed by the first and second filters, and the tap coefficients of the first filter are obtained. Fingerprint feature information is obtained based on the tap coefficients of the first filter, and then the fingerprint feature information is input into the classification and recognition model to obtain the authentication result of the device to be accessed. With the stability of the weight coefficients and low-order filtering characteristics of the adaptive filtering algorithm, a more concise fingerprint extraction can be effectively performed with a smaller number of signal samples, thereby improving the real-time performance of device access authentication. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 This is an application environment diagram of a device access authentication method in one embodiment;

[0063] Figure 2 This is a flowchart illustrating a device access authentication method in one embodiment;

[0064] Figure 3 This is a schematic diagram of the training process of a classification and recognition model in one embodiment;

[0065] Figure 4 This is a schematic diagram illustrating the working principle of a linear classifier using a support vector machine in one embodiment.

[0066] Figure 5 This is a flowchart illustrating step S202 of a device access authentication method in one embodiment;

[0067] Figure 6 This is a schematic diagram comparing the curves before and after the reshaping decision in one embodiment;

[0068] Figure 7 This is a structural block diagram of a device access authentication device in one embodiment;

[0069] Figure 8 This is an internal structural diagram of a computer device in one embodiment;

[0070] Figure 9 This is a diagram of the internal structure of a computer device in another embodiment. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0072] The device access authentication method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. Terminal 102 can be a signal acquisition device and a display device, and it can be used to acquire or receive network card signals from devices to be connected. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated on server 104 or placed on a cloud or other network server. The data storage system can be used to store network card signals, preprocessed raw input signal sequences, and reconstructed reference signal sequences, etc. Terminal 102 can be, but is not limited to, signal acquisition devices such as oscilloscopes, and also includes, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. For example, terminal 102 can be a signal acquisition device such as an oscilloscope, acquiring network card signals from devices on a gigabit Ethernet bus at a sampling frequency of 500MHz. The acquired signals are then processed using techniques such as decoding, demodulation, synchronization, and channel estimation to obtain clear device signals. The collected signals originate from the built-in gigabit network interface cards (NICs) of Ethernet computer devices. Different NIC manufacturers and manufacturing processes result in variations, and these signals carry unique characteristics of the NICs, including but not limited to minor differences in their hardware and circuitry. Server 104 can analyze and process these signals to extract unique fingerprint information associated with each device. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0073] In one exemplary embodiment, such as Figure 2 As shown, a device access authentication method is provided, which is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps S202 to S210. Wherein:

[0074] Step S202: Obtain the network card signal of the device to be connected, and preprocess the network card signal to obtain N original input signal sequences and corresponding reconstructed reference signal sequences.

[0075] Where N is a natural number greater than or equal to 1.

[0076] For example, server 104 can perform preprocessing steps such as frame capture, clock synchronization, and level reconstruction on the network card signal, and then truncate the processed signal to divide it into multiple signal segments. Each signal segment includes a set of signal sequences, resulting in N original input signal sequences and corresponding reconstructed reference signal sequences. The original input signal sequences are the original signals without level reconstruction, and the reconstructed reference signal sequences are the reference signals after level reconstruction.

[0077] Step S204: For each original input signal sequence, the original input signal sequence is input in parallel to the first filter and the second filter to obtain the first filtered signal output by the first filter and the second filtered signal corresponding to the second filter, respectively.

[0078] In this design, the step size of the first filter is larger than that of the second filter. The first filter, with its larger step size, can adapt to the signal characteristics of the original input signal, thereby quickly adjusting parameters to accelerate convergence. The second filter, with its smaller step size, provides backup parameters for the first filter.

[0079] For example, server 104 can first synchronize multiple raw input signals by clock, and under the premise of ensuring that the first filter and the second filter work in parallel, input the raw input signal sequence to the first filter and the second filter simultaneously and in parallel for each raw input signal sequence, so that the two filters can perform adaptive filtering on the same raw input signal in parallel, and obtain the first filtered signal output by the first filter and the second filtered signal corresponding to the second filter, respectively.

[0080] For example, since the adaptive filter tap coefficients extracted by different wired network cards in Gigabit Ethernet using the same method and parameters are on the same order of magnitude and have only minor differences, the server 104 can determine the initial common weight vector w(0), i.e., the tap coefficients, based on a priori methods, so that this vector is the same as the final extracted vector within one significant digit range, i.e., the initial value and the final value are on the same order of magnitude, thereby reducing the number of iterations to reach final convergence. Typically, the initialized weight vector is a non-zero vector. Based on the parallel filter method, the original input signal sequence x(n) is input into the first filter L1 and the second filter L2. The step size (i.e., the learning factor) of L1 is μ1, and the step size of L2 is μ2, where n represents the current time. Both the first filter and the second filter can be adaptive filters of the same order based on the LMS algorithm.

[0081] Subsequently, by combining the current original input signal sequence x(n) with the tap coefficients w1(n) of the first filter and w2(n) of the second filter, the two adaptive filters output the first filtered signal y1(n) and the second filtered signal y2(n) in parallel.

[0082]

[0083] Where k is the order of the adaptive filters L1 and L2, w1(n) and w2(n) are one-dimensional vectors of length k, and w1 T (n) and w2 T (n) represents the transpose of the vector, and x(n) refers to the input signal at the first k time steps when the sampling time in the discrete time domain is n.

[0084] Step S206: Calculate the current output error of the first filter based on the reconstructed reference signal sequence and the first filtered signal.

[0085] Here, the current output error refers to the error data between the output signal of the first filter and the reconstructed reference signal. The server 104 can input the original input signal sequence into the first filter and the second filter in parallel, so that the two filters can work in parallel. By calculating the current output error of the first filter, it can be determined whether the first filter shows signs of pre-divergence due to the tap coefficients caused by the large step size during the convergence process.

[0086] For example, server 104 can use the reconstructed reference signal sequence d(n) as the reference signal to calculate the errors e1(n) and e2(n) between the L1 and L2 output signals y1(n) and y2(n) and the reference signal d(n), respectively.

[0087] e1(n)=y1(n0-d(n)

[0088] e²(n) = y²(n) - d(n)

[0089] Where y1(n), y2(n) and d(n) are all signal level values.

[0090] Step S208: If the current output error exceeds the preset error threshold, update the tap coefficients of the first filter according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, keep the tap coefficients of the first filter unchanged. This process continues until all N original input signal sequences have been processed by the first filter and the second filter, and the tap coefficients of the first filter are obtained.

[0091] When the tap coefficients of the first filter show signs of pre-divergence due to the large step size during convergence, the current tap coefficients and error of the second filter can be dynamically transferred to the first filter. The first filter inherits the stable state of the second filter and continues to converge with a large step size. In adaptive filtering extraction methods, fingerprints obtained at different step sizes have slight differences. Because the step size of this method is relatively fixed, it can better guarantee the recognition accuracy of fingerprint extraction compared to variable step size algorithms such as NLMS that adjust the step size in real time.

[0092] For example, server 104 can calculate the error data of the second filter based on the reconstructed reference signal sequence and the second filtered signal, replace the current output error of the first filter with the error data of the second filter, and replace the tap coefficients of the first filter with the tap coefficients of the second filter.

[0093] For example, server 104 can check whether the error e1(n) of the master adaptive filter L1 exceeds the preset error threshold. If it does, it indicates that pre-divergence has occurred. At this time, the current e2(n) and w2(n) of L2 filter are passed to L1, and L1 continues to work according to the original step size in the following steps.

[0094] e1(n)=e2(n)w1(n)=w2(n)

[0095] If it does not exceed the limit, then continue execution.

[0096] Next, server 104 can use the LMS algorithm to update the tap coefficients w1(n+1) and w2(n+1), which includes normalizing the tap coefficients (weight vector) and combining the errors e1(n), e2(n) and the ideal signal segments z1(n), z2(n) to obtain the updated tap coefficients (i.e., weight vector):

[0097] w1(n+1)=w1(n)+μ1·e1(n)·x(n)

[0098] w2(n+1)=w2(n)+μ2·e2(n)·x(n)

[0099] Repeat the above steps to continuously update the tap coefficients until the preset iteration conditions are met, such as iterating to a preset number of times, and output the tap coefficient w1(n) obtained in L1 at this time.

[0100] Step S210: Input the fingerprint feature information obtained based on the tap coefficients of the first filter into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0101] For example, server 104 can use the tap coefficient of the first filter of each as an initial fingerprint feature; perform a weighted average calculation on N initial fingerprint features to obtain the target fingerprint feature, input the target fingerprint feature into the classification and recognition model, and obtain the authentication result of the device to be accessed.

[0102] For example, server 104 can use w1(n) obtained in step S208 as the initial fingerprint feature γ, and perform weighted averaging of multiple initial fingerprints in each initial sample set to obtain an optimized fingerprint feature information set.

[0103]

[0104] For M initial sample sets, processing them according to the above steps will yield the final device fingerprint F = {f1(k), f2(k), ... f...} extracted based on parallel adaptive filtering. M In the adaptive filtering algorithm, k is the filter order, i.e., the length of a single fingerprint. Typically, due to the characteristics of a 1000BASE-T steady-state signal, the filter order k is not high, meaning the adaptive filter possesses low-order filtering characteristics. This implies that the extracted fingerprint is usually short. Due to the dynamic adjustment of weight coefficients in the adaptive filtering algorithm, it exhibits better stability in noise suppression and signal extraction compared to other information processing methods based on statistical features; that is, the filter coefficient values ​​w1(n) are relatively stable. Therefore, in practical applications, a large number of training samples are usually unnecessary, meaning the M value (the number of training sample groups) can be set lower. This means that, compared to other fingerprint recognition methods, the device access authentication method based on adaptive filtering requires fewer training samples.

[0105] In the above device access authentication method, the network card signal of the device to be accessed is preprocessed to obtain N original input signal sequences and corresponding reconstructed reference signal sequences. The method sets a first filter with a larger step size to more effectively converge the signal, and a second filter with a smaller step size to avoid error divergence. Each original input signal sequence is input in parallel to the first and second filters, resulting in a first filtered signal output by the first filter and a second filtered signal corresponding to the second filter. During signal output, the current output error of the first filter is calculated based on the reconstructed reference signal sequence and the first filtered signal. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter remain unchanged. Since the step size of the first filter is larger than that of the second filter, when the tap coefficients of the first filter show signs of pre-divergence due to the large step size during the convergence of certain signals, the tap coefficients of the first filter can be corrected based on the more accurate tap coefficients of the second filter based on the convergence result. This process continues until all N original input signal sequences have been processed by the first and second filters, and the tap coefficients of the first filter are obtained. Fingerprint feature information is obtained based on the tap coefficients of the first filter, and then the fingerprint feature information is input into the classification and recognition model to obtain the authentication result of the device to be accessed. This allows for effective fingerprint extraction with a smaller number of signal samples, thereby improving the real-time performance of device access authentication.

[0106] In one exemplary embodiment, such as Figure 3 As shown, the training method for the classification and recognition model may include: acquiring a dataset containing fingerprint feature samples and corresponding identity labels, and then training an initial classification model based on the identity labels and the fingerprint feature samples to obtain the classification and recognition model. Specifically, the training method for the classification and recognition model may include steps S302 to S306. Wherein:

[0107] Step S302: Obtain the historical authentication dataset.

[0108] The historical authentication dataset includes historical fingerprint feature sets and identity tags. The historical fingerprint feature sets include multiple fingerprint feature samples.

[0109] For example, server 104 can generate identity tags based on specific devices and concatenate them with historical fingerprint feature sets to form a labeled dataset, which serves as a historical authentication dataset. Server 104 can train a classification and recognition model using a Support Vector Machine (SVM). SVM is a supervised learning algorithm commonly used for classification problems. The basic idea of ​​SVM is to find a hyperplane that can separate samples of different classes while maximizing the minimum distance (margin) from the samples to the hyperplane. The main objective is to find a maximum margin hyperplane that maximizes the distance from data points to the hyperplane.

[0110] Furthermore, the method for obtaining the aforementioned historical fingerprint feature set may include: acquiring multiple sets of historical signal sets; wherein each set of historical signal sets includes multiple historical input signal sequences and corresponding historical reconstructed signal sequences; for each set of historical signal sets, performing the following steps: inputting each historical input signal sequence in the historical signal set into a first filter and a second filter in parallel, respectively obtaining a first historical signal output by the first filter and a second historical signal corresponding to the second filter; calculating a first output error of the first filter based on the historical reconstructed signal sequence and the first historical signal; updating the tap coefficients of the first filter based on the tap coefficients of the second filter when the first output error exceeds a preset error threshold, and keeping the tap coefficients of the first filter unchanged when the first output error does not exceed the preset error threshold; until all N historical input signal sequences have been processed by the first filter and the second filter, obtaining the tap coefficients of the first filter; obtaining fingerprint feature samples based on the tap coefficients of the first filter; and combining the fingerprint feature samples corresponding to multiple sets of historical signal sets to obtain a historical fingerprint feature set.

[0111] Step S304: Train a binary classifier for each identity label.

[0112] In this binary classifier, for each identity label, fingerprint feature samples identical to that label are classified into one class, and fingerprint feature samples different from that label are classified into another. Support Vector Machine (SVM) models are initially designed for binary classification problems, but device fingerprint recognition is a multi-class problem. Therefore, a one-versus-rest strategy is used, treating each category as one class in a binary classification task, and treating all other categories as the other. During the training phase, a binary classifier is trained for each category to distinguish that category from all other categories. During the prediction phase, for a new sample, its category is determined by the prediction results of all binary classifiers.

[0113] For example, server 104 can group samples labeled with the same identity tag into one class, train a binary classifier for each class, and group the remaining samples into another class. In this way, samples from k classes can construct k classifiers and obtain labeled datasets for multiple devices.

[0114] Step S306: Use a binary classifier to train the initial classification model to obtain a classification and recognition model.

[0115] For example, server 104 can divide the above-mentioned labeled dataset into a training set and a test set according to a certain ratio, wherein the training set is used to train the SVM model and the test set is used to evaluate the performance and accuracy of the model.

[0116] When a computer device attempts to connect to a gigabit Ethernet network, the device fingerprint is extracted using the steps described above and classified according to the SVM model to determine its identity. If the fingerprint cannot be classified into an existing identity within the model, the device is considered an unknown device and poses a risk of spoofing attacks.

[0117] Furthermore, server 104 can choose linear kernels, polynomial kernels, and Gaussian kernels as kernel functions; this embodiment uses a linear kernel for training. The training set data is used to train the model using an SVM to obtain an SVM model for device authentication and recognition. For example... Figure 4 As shown, the formula for a linear classifier using a Support Vector Machine (SVM) is as follows:

[0118] Given an input feature vector x, the predicted output of the SVM model is:

[0119] f(x) = sign(w T x+b)

[0120] Where w is the normal vector (weight vector), which determines the direction of the hyperplane, and b is the offset (or intercept), which determines the positional relationship between the hyperplane and the origin.

[0121] Subsequently, server 104 can use the test set data to test the trained SVM model, calculate the model's classification accuracy and performance metrics. If the classification accuracy reaches the target level, the model testing is complete, and the model can then be applied to the classification and recognition of actual device fingerprints. The trained SVM model serves as the system's device classification and recognition model.

[0122] In one exemplary embodiment, such as Figure 5 As shown, step S202 includes steps S402 to S406. Wherein:

[0123] Step S402: Select the signal segment of the network card signal and divide it into N level intervals as N original input signal sequences.

[0124] For example, server 104 can divide the 1000BAST-T level value range [-2,2] into 49 intervals to obtain the level values ​​of 50 endpoints, and then perform continuous segmentation processing on these level intervals to divide them into N segments, thereby obtaining N original input signal sequences.

[0125] Step S404: For each level interval, extract the level value of the first endpoint and the level value of the second endpoint of the level interval respectively.

[0126] The voltage level at the first endpoint is higher than that at the second endpoint. V_H represents the voltage level at the first endpoint of each interval, and V_L represents the voltage level at the second endpoint of each interval.

[0127] For example, server 104 performs a shaping decision on the network card signal for each interval, calculates the average value (V_H+V_L) / 2 for each interval, and obtains the signal after the decision. The signal comparison before and after the decision is as follows: Figure 6 As shown.

[0128] Step S406: Reconstruct the level within the level interval based on the first endpoint level value and the second endpoint level value to obtain N reconstructed reference signal sequences.

[0129] For example, server 104 can calculate the average of the first endpoint level value and the second endpoint level value, reconstruct the level signal in the level range that is higher than the average value as the level signal corresponding to the first endpoint level value, and reconstruct the level signal in the level range that is lower than the average value as the level signal corresponding to the second endpoint level value, thereby obtaining more decision level values, making the reconstructed reference signal fit the original input signal better, and enabling more effective tracking in the process of the original input signal adapting to the ideal signal, improving the performance and adaptability of adaptive filtering, reducing the possibility of error divergence, and enabling the input signal to match the target ideal signal more accurately.

[0130] For example, the server 104 can segment the acquired original signal and the reconstructed ideal signal to obtain N signal segments. Each signal segment is a signal sequence, and each signal segment will be used to generate a fingerprint feature set. The acquired signal segment is used as the original input signal segment of the adaptive filter, and the corresponding reconstructed signal segment is used as the target ideal signal segment, which facilitates subsequent input adaptive filtering to extract fingerprints.

[0131] In another exemplary embodiment, server 104 receives the network card signal of the device to be accessed collected by terminal 102, selects a portion of the network card signal and divides it into multiple level intervals. For each level interval, it extracts the level value of the first endpoint and the level value of the second endpoint, calculates the average value of the first endpoint and the second endpoint, and then reconstructs the level signal in the level interval that is higher than the average value into the level signal corresponding to the first endpoint level value. Finally, it reconstructs the level signal in the level interval that is lower than the average value into the level signal corresponding to the second endpoint level value. Thus, the signal segment after level reconstruction is divided into N reconstruction reference signal sequences, and the signal segment without level reconstruction is divided into N original input signal sequences. For each original input signal sequence, the original input signal sequence is input in parallel to the first filter and the second filter to obtain the first filtered signal output by the first filter and the second filtered signal corresponding to the second filter, respectively.

[0132] Next, server 104 calculates the current output error of the first filter based on the reconstructed reference signal sequence and the first filtered signal. If the current output error exceeds a preset error threshold, server 104 calculates the error data of the second filter based on the reconstructed reference signal sequence and the second filtered signal, and replaces the current output error of the first filter with the error data of the second filter; server 104 also replaces the tap coefficients of the first filter with the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter remain unchanged until all N original input signal sequences have been processed by the first and second filters, thus obtaining the tap coefficients of the first filter.

[0133] Finally, server 104 uses the tap coefficient of the first filter of each as an initial fingerprint feature, performs a weighted average calculation on N initial fingerprint features to obtain the target fingerprint feature, inputs the target fingerprint feature into the classification and recognition model, and obtains the authentication result of the device to be accessed.

[0134] In the above embodiments, by improving the shaping decision process of the original signal, eliminating divergence based on the parameter transfer of the parallel adaptive filter, and setting the initial tap coefficients a priori, the adaptive filtering method achieves a significant improvement in the convergence effect of the error between the original signal and the ideal signal in the gigabit Ethernet scenario. This enables effective fingerprint extraction with a small number of signal samples. Compared with related technologies, the method proposed in the above embodiments can improve the device access authentication speed in the gigabit Ethernet scenario and shorten the length of the device fingerprint required for device authentication, thus reducing the fingerprint extraction time and computational cost.

[0135] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0136] Based on the same inventive concept, this application also provides a device access authentication apparatus for implementing the device access authentication method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more device access authentication apparatus embodiments provided below can be found in the limitations of the device access authentication method described above, and will not be repeated here.

[0137] In one exemplary embodiment, such as Figure 7 As shown, a device access authentication device is provided, including: a signal acquisition module 702, a signal processing module 704, a data calculation module 706, a data update module 708, and a device authentication module 710, wherein:

[0138] The signal acquisition module 702 is used to acquire the network card signal of the device to be connected, and to preprocess the network card signal to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1.

[0139] The signal processing module 704 is used to input the original input signal sequence in parallel to the first filter and the second filter for each original input signal sequence, so as to obtain the first filtered signal output by the first filter and the second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter.

[0140] The data calculation module 706 is used to calculate the current output error of the first filter based on the reconstructed reference signal sequence and the first filtered signal.

[0141] The data update module 708 updates the tap coefficients of the first filter according to the tap coefficients of the second filter when the current output error exceeds the preset error threshold, and keeps the tap coefficients of the first filter unchanged when the current output error does not exceed the preset error threshold; until all N original input signal sequences have been processed by the first filter and the second filter, the tap coefficients of the first filter are obtained.

[0142] The device authentication module 710 is used to input the fingerprint feature information obtained according to the tap coefficients of the first filter into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0143] In one embodiment, the coefficient update module 708 is specifically used to: calculate the error data of the second filter based on the reconstructed reference signal sequence and the second filtered signal, and replace the current output error of the first filter with the error data of the second filter.

[0144] In one embodiment, the device authentication module 710 is specifically used to: take the tap coefficient of each of the first filters as an initial fingerprint feature; perform a weighted average calculation on the N initial fingerprint features to obtain the target fingerprint feature; and input the target fingerprint feature into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0145] In one embodiment, the apparatus further includes a model training module, specifically used for: acquiring a historical authentication dataset; wherein the historical authentication dataset includes a historical fingerprint feature set and an identity label, the historical fingerprint feature set including multiple fingerprint feature samples; and training an initial classification model based on the identity label and fingerprint feature samples to obtain a classification and recognition model.

[0146] In one embodiment, the model training module is further configured to: acquire multiple sets of historical signal sets; wherein each set of historical signal sets includes multiple historical input signal sequences and corresponding historical reconstructed signal sequences; for each set of historical signal sets, perform the following steps: input each historical input signal sequence in the historical signal set into a first filter and a second filter in parallel to obtain a first historical signal output by the first filter and a second historical signal corresponding to the second filter, respectively; calculate a first output error of the first filter based on the historical reconstructed signal sequence and the first historical signal; update the tap coefficients of the first filter based on the tap coefficients of the second filter if the first output error exceeds a preset error threshold, and keep the tap coefficients of the first filter unchanged if the first output error does not exceed the preset error threshold; until all N historical input signal sequences have been processed by the first filter and the second filter, the tap coefficients of the first filter are obtained; obtain fingerprint feature samples based on the tap coefficients of the first filter; and combine the fingerprint feature samples corresponding to multiple sets of historical signal sets to obtain a historical fingerprint feature set.

[0147] In one embodiment, the signal acquisition module 702 includes:

[0148] The interval division unit is used to select the signal segment of the network card signal and divide it into N level intervals, which are used as N original input signal sequences;

[0149] The level extraction unit is used to extract the level values ​​of the first endpoint and the second endpoint of each level interval.

[0150] The level reconstruction unit is used to reconstruct the level within the level interval based on the level values ​​of the first endpoint and the second endpoint, and obtain N reconstructed reference signal sequences.

[0151] In one embodiment, the level value of the first endpoint is higher than the level value of the second endpoint; the level reconstruction unit is specifically used to: calculate the average value of the first endpoint level value and the level value of the second endpoint; reconstruct the level signal in the level range that is higher than the average value into the level signal corresponding to the level value of the first endpoint; and reconstruct the level signal in the level range that is lower than the average value into the level signal corresponding to the level value of the second endpoint.

[0152] The modules in the aforementioned device access authentication device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0153] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores network interface card (NIC) signals, preprocessed raw input signal sequences, and reconstructed reference signal sequences. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a device access authentication method.

[0154] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a device access authentication method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0155] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0156] In one exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the following steps: acquiring the network card signal of the device to be accessed, and preprocessing the network card signal to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1.

[0157] For each original input signal sequence, the original input signal sequence is input into the first filter and the second filter in parallel to obtain the first filtered signal output by the first filter and the second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter.

[0158] Based on the reconstructed reference signal sequence and the first filtered signal, the current output error of the first filter is calculated. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter are kept unchanged.

[0159] The process continues until all N original input signal sequences have been processed by the first and second filters, at which point the tap coefficients of the first filter are obtained.

[0160] The fingerprint feature information obtained from the tap coefficients of the first filter is input into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0161] In one embodiment, when the processor executes the computer program, it further performs the following steps: calculating error data of the second filter based on the reconstructed reference signal sequence and the second filtered signal, and replacing the current output error of the first filter with the error data of the second filter.

[0162] In one embodiment, when the processor executes the computer program, it further performs the following steps: taking the tap coefficient of the first filter of each as an initial fingerprint feature; performing a weighted average calculation on the N initial fingerprint features to obtain the target fingerprint feature; and inputting the target fingerprint feature into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0163] In one embodiment, when the processor executes the computer program, it further performs the following steps: acquiring a historical authentication dataset; wherein the historical authentication dataset includes a historical fingerprint feature set and an identity label, and the historical fingerprint feature set includes multiple fingerprint feature samples; training an initial classification model based on the identity label and fingerprint feature samples to obtain a classification and recognition model.

[0164] In one embodiment, when the processor executes the computer program, it further performs the following steps: acquiring multiple sets of historical signal sets; wherein each set of historical signal sets includes multiple historical input signal sequences and corresponding historical reconstructed signal sequences; for each set of historical signal sets, performing the following steps: inputting each historical input signal sequence in the historical signal set into a first filter and a second filter in parallel to obtain a first historical signal output by the first filter and a second historical signal corresponding to the second filter, respectively; calculating a first output error of the first filter based on the historical reconstructed signal sequence and the first historical signal; updating the tap coefficients of the first filter based on the tap coefficients of the second filter when the first output error exceeds a preset error threshold, and keeping the tap coefficients of the first filter unchanged when the first output error does not exceed the preset error threshold; until all N historical input signal sequences have been processed by the first filter and the second filter, the tap coefficients of the first filter are obtained; obtaining fingerprint feature samples based on the tap coefficients of the first filter; and combining the fingerprint feature samples corresponding to multiple sets of historical signal sets to obtain a historical fingerprint feature set.

[0165] In one embodiment, when the processor executes the computer program, it further performs the following steps: selecting the signal segment of the network card signal to divide it into N level intervals as N original input signal sequences; for each level interval, extracting the level value of the first endpoint and the level value of the second endpoint of the level interval respectively; and reconstructing the level within the level interval based on the level value of the first endpoint and the level value of the second endpoint to obtain N reconstructed reference signal sequences.

[0166] In one embodiment, when the processor executes the computer program, it further performs the following steps: calculating the average of the first endpoint level value and the second endpoint level value; reconstructing the level signal in the level range that is higher than the average value into the level signal corresponding to the first endpoint level value; and reconstructing the level signal in the level range that is lower than the average value into the level signal corresponding to the second endpoint level value.

[0167] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon. When the computer program is executed by a processor, it performs the following steps: acquiring the network card signal of the device to be accessed, and preprocessing the network card signal to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1.

[0168] For each original input signal sequence, the original input signal sequence is input into the first filter and the second filter in parallel to obtain the first filtered signal output by the first filter and the second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter.

[0169] Based on the reconstructed reference signal sequence and the first filtered signal, the current output error of the first filter is calculated. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter are kept unchanged.

[0170] The process continues until all N original input signal sequences have been processed by the first and second filters, at which point the tap coefficients of the first filter are obtained.

[0171] The fingerprint feature information obtained from the tap coefficients of the first filter is input into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0172] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: calculating error data of the second filter based on the reconstructed reference signal sequence and the second filtered signal, and replacing the current output error of the first filter with the error data of the second filter.

[0173] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: taking the tap coefficient of the first filter of each as an initial fingerprint feature; performing a weighted average calculation on the N initial fingerprint features to obtain the target fingerprint feature; and inputting the target fingerprint feature into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0174] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: acquiring a historical authentication dataset; wherein the historical authentication dataset includes a historical fingerprint feature set and an identity label, and the historical fingerprint feature set includes multiple fingerprint feature samples; training an initial classification model based on the identity label and fingerprint feature samples to obtain a classification and recognition model.

[0175] In one embodiment, when the computer program is executed by the processor, it further implements the following steps: acquiring multiple sets of historical signal sets; wherein each set of historical signal sets includes multiple historical input signal sequences and corresponding historical reconstructed signal sequences; for each set of historical signal sets, performing the following steps: inputting each historical input signal sequence in the historical signal set into a first filter and a second filter in parallel to obtain a first historical signal output by the first filter and a second historical signal corresponding to the second filter, respectively; calculating a first output error of the first filter based on the historical reconstructed signal sequence and the first historical signal; updating the tap coefficients of the first filter based on the tap coefficients of the second filter when the first output error exceeds a preset error threshold, and keeping the tap coefficients of the first filter unchanged when the first output error does not exceed the preset error threshold; until all N historical input signal sequences have been processed by the first filter and the second filter, the tap coefficients of the first filter are obtained; obtaining fingerprint feature samples based on the tap coefficients of the first filter; and combining the fingerprint feature samples corresponding to multiple sets of historical signal sets to obtain a historical fingerprint feature set.

[0176] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: selecting the signal segment of the network card signal to divide it into N level intervals as N original input signal sequences; for each level interval, extracting the level value of the first endpoint and the level value of the second endpoint of the level interval respectively; and reconstructing the level within the level interval based on the level value of the first endpoint and the level value of the second endpoint to obtain N reconstructed reference signal sequences.

[0177] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: calculating the average of the first endpoint level value and the second endpoint level value; reconstructing the level signal in the level interval that is higher than the average value into the level signal corresponding to the first endpoint level value; and reconstructing the level signal in the level interval that is lower than the average value into the level signal corresponding to the second endpoint level value.

[0178] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps: acquiring the network card signal of the device to be accessed, and preprocessing the network card signal to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1;

[0179] For each original input signal sequence, the original input signal sequence is input into the first filter and the second filter in parallel to obtain the first filtered signal output by the first filter and the second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter.

[0180] Based on the reconstructed reference signal sequence and the first filtered signal, the current output error of the first filter is calculated. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter are kept unchanged.

[0181] The process continues until all N original input signal sequences have been processed by the first and second filters, at which point the tap coefficients of the first filter are obtained.

[0182] The fingerprint feature information obtained from the tap coefficients of the first filter is input into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0183] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: calculating error data of the second filter based on the reconstructed reference signal sequence and the second filtered signal, and replacing the current output error of the first filter with the error data of the second filter.

[0184] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: taking the tap coefficient of the first filter of each as an initial fingerprint feature; performing a weighted average calculation on the N initial fingerprint features to obtain the target fingerprint feature; and inputting the target fingerprint feature into the classification and recognition model to obtain the authentication result of the device to be accessed.

[0185] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: acquiring a historical authentication dataset; wherein the historical authentication dataset includes a historical fingerprint feature set and an identity label, and the historical fingerprint feature set includes multiple fingerprint feature samples; training an initial classification model based on the identity label and fingerprint feature samples to obtain a classification and recognition model.

[0186] In one embodiment, when the computer program is executed by the processor, it further implements the following steps: acquiring multiple sets of historical signal sets; wherein each set of historical signal sets includes multiple historical input signal sequences and corresponding historical reconstructed signal sequences; for each set of historical signal sets, performing the following steps: inputting each historical input signal sequence in the historical signal set into a first filter and a second filter in parallel to obtain a first historical signal output by the first filter and a second historical signal corresponding to the second filter, respectively; calculating a first output error of the first filter based on the historical reconstructed signal sequence and the first historical signal; updating the tap coefficients of the first filter based on the tap coefficients of the second filter when the first output error exceeds a preset error threshold, and keeping the tap coefficients of the first filter unchanged when the first output error does not exceed the preset error threshold; until all N historical input signal sequences have been processed by the first filter and the second filter, the tap coefficients of the first filter are obtained; obtaining fingerprint feature samples based on the tap coefficients of the first filter; and combining the fingerprint feature samples corresponding to multiple sets of historical signal sets to obtain a historical fingerprint feature set.

[0187] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: selecting the signal segment of the network card signal to divide it into N level intervals as N original input signal sequences; for each level interval, extracting the level value of the first endpoint and the level value of the second endpoint of the level interval respectively; and reconstructing the level within the level interval based on the level value of the first endpoint and the level value of the second endpoint to obtain N reconstructed reference signal sequences.

[0188] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: calculating the average of the first endpoint level value and the second endpoint level value; reconstructing the level signal in the level interval that is higher than the average value into the level signal corresponding to the first endpoint level value; and reconstructing the level signal in the level interval that is lower than the average value into the level signal corresponding to the second endpoint level value.

[0189] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0190] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0191] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A device access authentication method, characterized in that, The method includes: The network card signal of the device to be connected is acquired, and the network card signal is preprocessed to obtain N original input signal sequences and corresponding reconstructed reference signal sequences; N is a natural number greater than or equal to 1. For each original input signal sequence, the original input signal sequence is input in parallel to a first filter and a second filter to obtain a first filtered signal output by the first filter and a second filtered signal corresponding to the second filter, respectively. The step size of the first filter is larger than the step size of the second filter. The current output error of the first filter is calculated based on the reconstructed reference signal sequence and the first filtered signal. If the current output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the current output error does not exceed the preset error threshold, the tap coefficients of the first filter remain unchanged. This process continues until all N original input signal sequences have been processed by the first filter and the second filter, at which point the tap coefficients of the first filter are obtained. The fingerprint feature information obtained based on the tap coefficients of the first filter is input into the classification and recognition model to obtain the authentication result of the device to be accessed.

2. The method according to claim 1, characterized in that, When the current output error exceeds a preset error threshold, the method further includes: The error data of the second filter is calculated based on the reconstructed reference signal sequence and the second filtered signal, and the current output error of the first filter is replaced by the error data of the second filter.

3. The method according to claim 1, characterized in that, The step of inputting the fingerprint feature information obtained based on the tap coefficients of the first filter into the classification and recognition model to obtain the authentication result of the device to be accessed includes: The tap coefficient of the first filter for each is used as an initial fingerprint feature; The target fingerprint feature is obtained by performing a weighted average calculation on the N initial fingerprint features; The target fingerprint features are input into a classification and recognition model to obtain the authentication result of the device to be accessed.

4. The method according to claim 3, characterized in that, The training method for the classification and recognition model includes: Obtain a historical authentication dataset; wherein the historical authentication dataset includes a historical fingerprint feature set and an identity label, and the historical fingerprint feature set includes multiple fingerprint feature samples; The initial classification model is trained based on the identity label and the fingerprint feature sample to obtain a classification and recognition model.

5. The method according to claim 4, characterized in that, The method for obtaining the historical fingerprint feature set includes: Multiple sets of historical signals are acquired; each set of historical signals includes multiple historical input signal sequences and corresponding historical reconstructed signal sequences. For each set of historical signals, perform the following steps: Each historical input signal sequence in the historical signal set is input in parallel to the first filter and the second filter to obtain the first historical signal output by the first filter and the second historical signal corresponding to the second filter, respectively. The first output error of the first filter is calculated based on the historical reconstructed signal sequence and the first historical signal. If the first output error exceeds a preset error threshold, the tap coefficients of the first filter are updated according to the tap coefficients of the second filter. If the first output error does not exceed the preset error threshold, the tap coefficients of the first filter remain unchanged. This process continues until all N historical input signal sequences have been processed by the first filter and the second filter, at which point the tap coefficients of the first filter are obtained. Fingerprint feature samples are obtained based on the tap coefficients of the first filter; By combining the fingerprint feature samples corresponding to multiple sets of historical signal sets, a historical fingerprint feature set is obtained.

6. The method according to claim 1, characterized in that, The preprocessing of the network card signal to obtain N original input signal sequences and corresponding reconstructed reference signal sequences includes: The signal segment of the network card signal is selected and divided into N level intervals, which are used as N original input signal sequences; For each of the stated level intervals, the level values ​​at the first and second endpoints of the level intervals are extracted respectively. Based on the first endpoint level value and the second endpoint level value, the level within the level interval is reconstructed to obtain N reconstructed reference signal sequences.

7. The method according to claim 6, characterized in that, The first endpoint level value is higher than the second endpoint level value; the step of reconstructing the level within the level interval based on the first endpoint level value and the second endpoint level value includes: Calculate the average value of the first endpoint level and the second endpoint level; The level signals in the level range that are higher than the average value are reconstructed into the level signals corresponding to the first endpoint level value; The level signals in the level range that are lower than the average value are reconstructed into level signals corresponding to the second endpoint level value.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images