Multi-feature satellite navigation signal anti-deception jamming detection method based on auto-encoder

Abstract

The invention discloses a multi-feature satellite navigation signal anti-deception jamming detection method based on an auto-encoder, and belongs to the technical field of satellite navigation and signal processing. According to the method, Z-score standardization processing is carried out on a received satellite navigation signal by extracting four key characteristic values including a carrier-to-noise ratio, an in-phase component, an orthogonal component and Doppler frequency shift, and dimensional difference is eliminated; and training and predicting a normal signal mode in the auto-encoder model, and comparing a reconstruction error with a set threshold value to judge whether the signal is abnormal or not. The method does not need extra hardware support, is suitable for resource-constrained equipment, has efficient and accurate detection capability in scenes of unmanned aerial vehicle navigation, vehicle navigation, military communication and the like, has the advantages of high flexibility, low cost and wide applicability, and particularly has remarkable advantages in the aspect of detecting complex deception signals.

Description

Technical Field

[0001] The present invention relates to the field of satellite navigation and signal processing technology, and in particular to a multi-feature satellite navigation signal anti-spoofing interference detection method based on an autoencoder. Background Art

[0002] As the core technology of modern communications, transportation, emergency management and military strategy, the Global Navigation Satellite System (GNSS) has been deeply integrated into people's daily lives and key industries. The positioning, navigation and timing services provided by the system have been widely used in transportation, finance, military and other fields. However, with the increasing reliance on GNSS, its signal security issues have gradually become prominent, especially when encountering deceptive interference, GNSS signals are extremely susceptible to deceptive interference. This interference confuses the receiver by generating false signals or forwarding real signals, causing it to make incorrect positioning, speed and time judgments, resulting in user or system behavior deviating from normal, which may cause traffic accidents, financial system chaos, and even military decision-making errors. Therefore, improving the GNSS system's ability to resist deception and interference has become an issue that the industry urgently needs to solve.

[0003] Existing GNSS anti-spoofing detection methods mainly rely on the matching check of Doppler positioning and pseudorange positioning or digital signature encryption technology. For example, by calculating the Euclidean distance between the Doppler positioning and pseudorange positioning results and combining it with the generalized likelihood ratio test, it can be determined whether there is a spoofing signal. This type of method is suitable for detecting typical pseudorange spoofing attacks, but has limited effect when dealing with other complex types of attacks, and has high requirements for the setting of precise parameters, so it cannot be widely used in cost-sensitive civilian equipment. In addition, some methods use public key encryption to sign navigation data and combine statistical hypothesis testing to verify the rationality of the signal arrival time. Although this method can enhance signal security, it has too high requirements for key management, hardware performance and computing resources, and is difficult to promote, especially for low-power and lightweight devices.

[0004] Current anti-interference technologies mostly use signal processing or encryption technology. Although they can alleviate signal interference to a certain extent, their flexibility and applicability are relatively limited, especially when facing diverse and increasingly complex interference environments. Therefore, it is particularly necessary to develop an efficient and flexible GNSS anti-spoofing interference detection method for application on low-cost, resource-constrained devices to improve the anti-interference capability of the navigation system in various environments and ensure the safe operation of key systems. Summary of the invention

[0005] In view of the limitations of the existing spoofing signal detection methods in application, such as high complexity, strong parameter dependence and high resource requirements, the present invention provides an efficient anti-spoofing interference detection method for satellite navigation signals. The present invention adopts a multi-feature fusion detection method and uses an autoencoder model to jointly analyze the main features of satellite navigation signals to improve the flexibility and applicability of spoofing detection.

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

[0007] A multi-feature satellite navigation signal anti-spoofing interference detection method based on an autoencoder comprises the following steps:

[0008] Step S1: extracting main characteristic values ​​from the received satellite navigation signal: carrier-to-noise ratio, in-phase component, orthogonal component, Doppler frequency shift;

[0009] Extract the main characteristic values ​​such as carrier-to-noise ratio, in-phase component, orthogonal component, Doppler frequency shift, etc. from the received satellite navigation signal. These characteristic values ​​contain the phase, frequency and intensity information of the navigation signal, and can accurately reflect the changing characteristics of the signal during the reception process. Compared with single feature analysis, the present invention improves the comprehensiveness and accuracy of detection through multi-feature fusion.

[0010] Step S2: Standardize the extracted main eigenvalues ​​to eliminate dimensional differences;

[0011] The extracted eigenvalues ​​are Z-score standardized to eliminate dimensional differences, so that the mean of the eigenvalues ​​is 0 and the standard deviation is 1. Standardization reduces the deviation of each feature caused by different data scales and ensures data consistency during model training.

[0012] Step S3: training and predicting the main eigenvalues ​​after the standardization based on the autoencoder model;

[0013] The standardized feature values ​​are input into the autoencoder model for training, and the encoder and decoder structures of the autoencoder are optimized. The encoder part includes two layers of neurons, 3 and 2 neurons respectively, and both use the hyperbolic tangent (tanh) activation function to enable it to compress the input signal into a low-dimensional space. The decoder restores the low-dimensional code to reconstruct an output close to the original features.

[0014] Step S4: Calculate the reconstruction error of the autoencoder model, set a threshold of the reconstruction error, and determine whether the signal is abnormal based on a comparison result between the reconstruction error and the threshold, wherein when the reconstruction error exceeds the threshold, it is determined that the received signal may be subject to deception interference.

[0015] The reconstruction error of the input signal is calculated and compared with a preset threshold. When the reconstruction error is higher than the threshold, it is determined that the signal may be abnormal. The reconstruction error calculation of the present invention adopts the mean square error formula, which ensures the sensitivity of the model in detecting signal deviations. In addition, the present invention sets error thresholds in different scenarios to adapt to a variety of application scenarios (such as drones, vehicle navigation systems, etc.).

[0016] Preferably, the satellite navigation signal data in step S1 is feature extracted based on a TEXBAT dataset, wherein the TEXBAT dataset contains a plurality of real and simulated interference scenarios for training an autoencoder model and optimizing threshold settings.

[0017] Preferably, the standardization of the feature data in step S2 adopts a Z-score standardization method, which includes:

[0018] Calculate the average value μ of the characteristic data, the formula is as follows:

[0019]

[0020] Among them, f i represents the i-th original eigenvalue, n is the total number of feature data sets, μ is the mean of the feature data, which is used to reflect the central trend of the feature data and reduce the deviation;

[0021] Calculate the standard deviation σ of the characteristic data using the following formula:

[0022]

[0023] Among them, σ is the standard deviation of the characteristic data, which is used to measure the degree of dispersion of the characteristic data relative to the mean μ and ensure the scale consistency between different characteristic values;

[0024] Each eigenvalue is standardized and the original eigenvalue f i Convert to the standardized value f i ′, the formula is as follows:

[0025]

[0026] Among them, f i ′ is the i-th eigenvalue after standardization;

[0027] After the Z-score standardization, all feature data will conform to the standard normal distribution with a mean of 0 and a standard deviation of 1.

[0028] Preferably, the autoencoder model in step S3 compresses and reconstructs the input data through an encoder and a decoder, specifically including:

[0029] The encoder part compresses the four input eigenvalues ​​into a low-dimensional space, and reduces the data dimension layer by layer through a double-layer encoder;

[0030] The first layer encoder is set with 3 neurons, the second layer encoder is set with 2 neurons, and both layers of encoders use the hyperbolic tangent tanh activation function;

[0031] The decoder part restores the encoder's low-dimensional data to the original high-dimensional space and reconstructs output data that is as consistent as possible with the input data.

[0032] Preferably, the encoder comprises two layers:

[0033] The first encoder layer receives an input vector x = [x1, x2, x3, x4], where x∈R 4 represents the input vector containing four eigenvalues;

[0034] The first layer encoder is passed through the weight matrix W1∈R 3×4 and the bias vector b1∈R 3 Calculate the hidden layer output h1∈R3;

[0035] The first layer encoder uses the hyperbolic tangent activation function tanh, and the calculation formula is:

[0036] h1=tanh(W1x+b1)

[0037] For each hidden neuron i=1,2,3, the calculation process is as follows:

[0038]

[0039] in, is the weight matrix W1 from the input feature x j To the hidden neuron h 1i The weight of b 1i is the bias term that allows the model to produce non-zero output even when the input is zero.

[0040] Preferably, the encoder comprises two layers:

[0041] The first encoder layer receives an input vector x = [x1, x2, x3, x4], where x∈R 4 represents the input vector containing four eigenvalues;

[0042] The first layer encoder is passed through the weight matrix W1∈R 3×4 and the bias vector b1∈R 3 Calculate the hidden layer output h1∈R3;

[0043] The first layer encoder uses the hyperbolic tangent activation function tanh, and the calculation formula is:

[0044] h1=tanh(W1x+b1)

[0045] For each hidden neuron i=1,2,3, the calculation process is as follows:

[0046]

[0047] in, is the weight matrix W1 from the input feature x j To the hidden neuron h 1i The weight of b 1i is the bias term that allows the model to produce non-zero output even when the input is zero.

[0048] Preferably, the decoder part comprises two layers:

[0049] The first decoder layer receives the 2D encoded representation h2 = [h 21 ,h 22 ] T ;

[0050] Through the weight matrix W3∈R 3×2 and the bias vector b3∈R 3 Calculate the output vector h3∈R 3 ;

[0051] The activation function uses the sigmoid function, and the calculation formula is as follows:

[0052] h3=sigmoid(W3h2+b3)

[0053] For each hidden neuron i=1,2,3, the specific calculation process is:

[0054]

[0055] in, is an element in the weight matrix W3, representing the weight from the kth input node to the ith hidden neuron, b 3i is the corresponding bias term.

[0056] Preferably, the second layer of the decoder receives the output h3 of the first layer of the decoder and restores it to a 4-dimensional space:

[0057] The second layer decoder is passed through the weight matrix W4∈R 4×3 and the bias vector b4∈R 4 Calculate the final reconstructed output vector Make it close to the original input vector x;

[0058] The activation function uses the sigmoid function, and the calculation formula is as follows:

[0059]

[0060] For each output node j=1,2,3,4, the specific calculation process is:

[0061]

[0062] in, is the reconstructed output value, is an element in the weight matrix W4, representing the 3i To the output node The weight of b 4j is the corresponding bias term.

[0063] Preferably, the step of calculating the reconstruction error and setting the threshold in step S4 includes:

[0064] S41: The input multidimensional feature signal is a feature vector y = [y1, y2, ..., y n ], where: y represents the original eigenvector; n represents the dimension of the eigenvector, that is, the number of signal components; y n represents the nth component of the original signal;

[0065] S42: Use the trained autoencoder model to reconstruct the input signal and obtain the reconstructed signal vector in: Represents the feature vector reconstructed by the autoencoder model; represents the i-th component in the reconstructed signal vector, that is, the i-th signal component reconstructed by the autoencoder;

[0066] S43: The reconstruction error is expressed as mean square error, and the calculation formula is as follows:

[0067]

[0068] Where: MSE stands for mean square error, which is used to measure the reconstructed signal The difference between the original signal y; n represents the dimension of the feature vector; and i Represent the i-th component of the reconstructed signal and the original signal respectively;

[0069] S44: Based on the statistical characteristics of the data, the threshold T is set to 37. In a non-spoofing scenario, the MSE value is low, usually below 20, and the maximum value does not exceed 37; in a spoofing scenario, the MSE value increases significantly, and the maximum value can reach more than 1200, where: T represents the threshold of the mean square error, which is used to determine whether the signal is interfered;

[0070] S45: Compare the calculated reconstruction error MSE with the set threshold T:

[0071] If MSE≤T, the signal is judged to be normal;

[0072] If MSE>T, an abnormality is detected and it is determined that the signal may be interfered by deception.

[0073] Preferably, the method is applicable to interference detection in drone navigation, vehicle navigation systems and military communications, and specifically includes:

[0074] In drone navigation, the multi-dimensional data of the drone during normal operation is used as a training data set to input the autoencoder for training, and the abnormality of the input data is judged based on the reconstruction error. If the calculated reconstruction error exceeds the set threshold, it is judged that the drone may have suffered a GPS spoofing attack;

[0075] In vehicle navigation systems and military communications, since vehicle navigation systems and military communications rely on the positioning, navigation and timing services provided by the Global Navigation Satellite System GNSS, and the main eigenvalues ​​used by both are common signal parameters.

[0076] Compared with the prior art, the beneficial effects of the present invention are as follows: the multi-parameter detection method based on the autoencoder of the present invention does not require additional hardware support and is suitable for wide promotion and application. By extracting multiple key features such as carrier-to-noise ratio, in-phase component, orthogonal component and Doppler frequency shift from the received satellite navigation signal, and performing Z-score standardization on these features, so that they have the same scale in the model, the detection stability and applicability of the model are guaranteed. Compared with the traditional detection method that relies on a single eigenvalue, the present invention performs a joint analysis of multiple eigenvalues ​​through an autoencoder model, which makes up for the limitations of single parameter detection, thereby greatly improving the accuracy of detection. The autoencoder model adopts a two-layer encoder and a two-layer decoder structure, which can effectively reconstruct the characteristic mode of the input data under the condition of no deceptive signal, and judge the signal abnormality by reconstruction error. The reconstruction error threshold adopted is simple and has low calculation cost. By setting the threshold to 37, when the reconstruction error exceeds this value, it is judged that the signal may be interfered by deception. The detection method of the present invention is applicable to various scenarios such as unmanned aerial vehicle navigation, vehicle navigation system and military communication, which can significantly improve the security of GNSS signals in these fields and reduce potential safety hazards caused by deception interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0078] in:

[0079] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention;

[0080] Figure 2 It is a schematic diagram of the implementation process of the embodiment of the present invention in the actual detection process;

[0081] Figure 3 A schematic diagram of a data flow for feature value normalization processing in an embodiment of the present invention;

[0082] Figure 4 This is a schematic diagram of the structure of an autoencoder model in an embodiment of the present invention;

[0083] Figure 5 A schematic diagram of a curve showing a change in training loss during the training of an autoencoder in an embodiment of the present invention;

[0084] Figure 6 Schematic diagram of a curve showing a change in verification loss during the training of an autoencoder in an embodiment of the present invention. DETAILED DESCRIPTION

[0085] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0086] Example 1

[0087] like Figure 1 , Figure 2 As shown in FIG. 1 , an embodiment of the present invention is provided, which provides a multi-feature satellite navigation signal anti-spoofing interference detection method based on an autoencoder, comprising:

[0088] Step S1: extracting main characteristic values ​​from the received satellite navigation signal: carrier-to-noise ratio, in-phase component, orthogonal component, Doppler frequency shift;

[0089] Specifically, the satellite navigation signal data in step S1 is feature extracted based on the TEXBAT dataset, wherein the TEXBAT dataset contains multiple real and simulated interference scenarios for training the autoencoder model and optimizing the threshold setting.

[0090] In the present invention, four key characteristic values ​​are firstly extracted from the received satellite navigation signal: carrier-to-noise ratio (C / N0), in-phase component (Prompt_i), quadrature component (Prompt_q) and Doppler frequency shift (Carrier_Doppler_Hz).

[0091] Carrier-to-noise ratio (C / N0): This parameter indicates the signal-to-noise ratio of the signal. It is a key indicator used by GNSS receivers to evaluate signal quality. It can measure signal strength and noise interference. A low C / N0 may indicate that the signal is interfered.

[0092] In-phase component (Prompt_i) and quadrature component (Prompt_q): These two parameters represent the amplitude information and phase information of the signal respectively, and are used to judge the integrity of the signal. Their changes can reflect the possible phase or amplitude abnormality of the signal, which is suitable for detecting deceptive interference.

[0093] Doppler shift (Carrier_Doppler_Hz): reflects the frequency change caused by relative motion and is widely used in positioning and speed measurement. Abnormal frequency shift may indicate that the signal is interfered.

[0094] By monitoring the changes in these characteristic values, the method of the present invention provides basic data for detecting complex deceptive interference.

[0095] Further, such as Figure 3 As shown, step S2: standardize the extracted main eigenvalues ​​to eliminate dimensional differences;

[0096] The standardization of the feature data in step S2 adopts the Z-score standardization method, and the steps include:

[0097] Calculate the average value μ of the characteristic data, the formula is as follows:

[0098]

[0099] Among them, f i represents the i-th original eigenvalue, n is the total number of feature data sets, μ is the mean of the feature data, which is used to reflect the central trend of the feature data and reduce the deviation;

[0100] Calculate the standard deviation σ of the characteristic data using the following formula:

[0101]

[0102] Among them, σ is the standard deviation of the characteristic data, which is used to measure the degree of dispersion of the characteristic data relative to the mean μ and ensure the scale consistency between different characteristic values;

[0103] Each eigenvalue is standardized and the original eigenvalue f i Convert to the standardized value f i ′, the formula is as follows:

[0104]

[0105] Among them, f i ′ is the i-th eigenvalue after standardization;

[0106] After the Z-score standardization, all feature data will conform to the standard normal distribution with a mean of 0 and a standard deviation of 1.

[0107] In order to eliminate the dimensional differences of the eigenvalues, the eigenvalues ​​extracted in step S1 are subjected to Z-score normalization. Through standardization, all eigenvalues ​​conform to the standard normal distribution with a mean of 0 and a standard deviation of 1. This process ensures the scale consistency of different eigenvalues, helps the autoencoder model to accurately learn the normal mode of the signal, and improves the adaptability of detection in complex environments.

[0108] Further, such as Figure 4 As shown, step S3: training and predicting the main feature values ​​after the standardization based on the autoencoder model;

[0109] Specifically, the autoencoder model in step S3 compresses and reconstructs the input data through an encoder and a decoder, specifically including:

[0110] The encoder part compresses the four input eigenvalues ​​into a low-dimensional space, and reduces the data dimension layer by layer through a double-layer encoder;

[0111] The first layer encoder is set with 3 neurons, the second layer encoder is set with 2 neurons, and both layers of encoders use the hyperbolic tangent tanh activation function;

[0112] The decoder part restores the encoder's low-dimensional data to the original high-dimensional space and reconstructs output data that is as consistent as possible with the input data.

[0113] Among them, the encoder consists of two layers:

[0114] The first encoder layer receives an input vector x = [x1, x2, x3, x4], where x∈R 4 represents the input vector containing four eigenvalues;

[0115] The first layer encoder is passed through the weight matrix W1∈R 3×4 and the bias vector b1∈R 3 Calculate the hidden layer output h1∈R3;

[0116] The first layer encoder uses the hyperbolic tangent activation function tanh, and the calculation formula is:

[0117] h1=tanh(W1x+b1)

[0118] For each hidden neuron i=1,2,3, the calculation process is as follows:

[0119]

[0120] in, is the weight matrix W1 from the input feature x j To the hidden neuron h 1i The weight of b 1i is the bias term that allows the model to produce non-zero output even when the input is zero.

[0121] Specifically, the second layer of the encoder receives the output h1 of the first layer and further compresses the data into a 2D space:

[0122] The second layer encoder is passed through the weight matrix W2∈R 2×3 and the bias vector b2∈R 2 Calculate the output vector h2∈R 2 ;

[0123] The activation function also uses hyperbolic tangent tanh, and the calculation formula is:

[0124] h2=tanh(W2h1+b2)

[0125] Specifically, for each hidden neuron k=1,2, the calculation process is:

[0126]

[0127] in, is an element in the weight matrix W2, representing the weights from the first layer of hidden neurons h 1i To the second layer of hidden neurons h 2k The weight of b 2k is the corresponding bias term.

[0128] Among them, the decoder part contains two layers:

[0129] The first decoder layer receives the 2D encoded representation h2 = [h 21 ,h 22 ] T ;

[0130] Through the weight matrix W3∈R 3×2 and the bias vector b3∈R 3 Calculate the output vector h3∈R 3 ;

[0131] The activation function uses the sigmoid function, and the calculation formula is as follows:

[0132] h3=sigmoid(W3h2+b3)

[0133] For each hidden neuron i=1,2,3, the specific calculation process is:

[0134]

[0135] in, is an element in the weight matrix W3, representing the weight from the kth input node to the ith hidden neuron, b 3i is the corresponding bias term.

[0136] Specifically, the second layer of the decoder receives the output h3 of the first layer of the decoder and restores it to a 4-dimensional space:

[0137] The second layer decoder is passed through the weight matrix W4∈R 4×3 and the bias vector b4∈R 4 Calculate the final reconstructed output vector Make it close to the original input vector x;

[0138] The activation function uses the sigmoid function, and the calculation formula is as follows:

[0139]

[0140] For each output node j=1,2,3,4, the specific calculation process is:

[0141]

[0142] in, is the reconstructed output value, is an element in the weight matrix W4, representing the 3i To the output node The weight of b 4j is the corresponding bias term.

[0143] The present invention adopts an autoencoder model to train the eigenvalues ​​after standardization processing to establish a normal mode of the navigation signal.

[0144] Encoder: The input layer consists of four neurons, corresponding to four standardized feature values. The encoder consists of two layers. The first layer compresses the features into an intermediate representation of three neurons, and the second layer further compresses it into two neurons. The activation function is hyperbolic tangent (tanh). In this way, the model can effectively reduce the data dimension and extract the main features of the signal.

[0145] Decoder: gradually restores the encoded data to the original 4-dimensional space. By trying to reconstruct the input data, the decoder outputs the reconstructed data as close to the input as possible.

[0146] Training and prediction: The autoencoder model is trained on normal data to enable the model to learn the reconstruction pattern of normal signals. When detecting deceptive signals, the data to be tested is input into the model and the reconstructed signal data is output.

[0147] like Figure 5 As shown in the figure, the changing curve of training loss during the training process of the autoencoder model is shown. The training loss decreases rapidly within the first 20 training cycles epochs, indicating that the model quickly learns the basic characteristics of the data in the initial stage; then, the training loss gradually stabilizes, and finally stabilizes around 50 training cycles epochs, indicating that the learning process of the model has achieved a good reconstruction effect.

[0148] Further, step S4: calculate the reconstruction error of the autoencoder model, set a threshold of the reconstruction error, and judge whether the signal is abnormal based on the comparison result between the reconstruction error and the threshold, wherein when the reconstruction error exceeds the threshold, it is judged that the received signal may be deceptively interfered.

[0149] Specifically, the step of calculating the reconstruction error and setting the threshold in step S4 includes:

[0150] S41: The input multidimensional feature signal is a feature vector y = [y1, y2, ..., y n ], where: y represents the original eigenvector; n represents the dimension of the eigenvector, that is, the number of signal components; y n represents the nth component of the original signal;

[0151] S42: Use the trained autoencoder model to reconstruct the input signal and obtain the reconstructed signal vector in: Represents the feature vector reconstructed by the autoencoder model; represents the i-th component in the reconstructed signal vector, that is, the i-th signal component reconstructed by the autoencoder;

[0152] S43: The reconstruction error is expressed as mean square error, and the calculation formula is as follows:

[0153]

[0154] Where: MSE stands for mean square error, which is used to measure the reconstructed signal The difference between the original signal y; n represents the feature

[0155] The dimension of the vector; and i Represent the i-th component of the reconstructed signal and the original signal respectively;

[0156] S44: Based on the statistical characteristics of the data, the threshold T is set to 37. In a non-spoofing scenario, the MSE value is low, usually below 20, and the maximum value does not exceed 37; in a spoofing scenario, the MSE value increases significantly, and the maximum value can reach more than 1200, where: T represents the threshold of the mean square error, which is used to determine whether the signal is interfered;

[0157] S45: Compare the calculated reconstruction error MSE with the set threshold T:

[0158] If MSE≤T, the signal is judged to be normal;

[0159] If MSE>T, an abnormality is detected and it is determined that the signal may be interfered by deception.

[0160] like Figure 6 As shown in the figure, the change curve of the validation loss during the autoencoder training process is shown. The validation loss curve also shows a similar change trend to the training loss. It drops rapidly in the early stage, then tends to be stable, and remains close to the training loss. This shows that the autoencoder model performs well on the validation set without obvious overfitting.

[0161] Furthermore, the method is applicable to interference detection in drone navigation, vehicle navigation systems and military communications, including:

[0162] In drone navigation, the multi-dimensional data of the drone during normal operation is used as a training data set to input the autoencoder for training, and the abnormality of the input data is judged based on the reconstruction error. If the calculated reconstruction error exceeds the set threshold, it is judged that the drone may have suffered a GPS spoofing attack;

[0163] In vehicle navigation systems and military communications, since vehicle navigation systems and military communications rely on the positioning, navigation and timing services provided by the Global Navigation Satellite System GNSS, and the main eigenvalues ​​used by both are common signal parameters.

[0164] The advantages of this embodiment are:

[0165] The present invention is applicable to multiple scenarios such as drone navigation, vehicle navigation systems, military communications, etc., and can particularly work stably on devices with limited resources, thereby improving the flexibility of GNSS spoofing detection.

[0166] The low-dimensional encoding and high-dimensional reconstruction characteristics of the autoencoder model enable the model to have a higher matching degree with normal signals. Therefore, when a deceptive signal enters, the model will produce obvious reconstruction errors, greatly improving the detection accuracy of the deceptive signal.

[0167] Compared with encryption technology, the present invention does not rely on key management and high-performance hardware devices, can be implemented on low-power devices, and saves costs and resource consumption in application scenarios.

[0168] The present invention can combat various types of interference signals by means of multi-feature fusion, and can especially stably detect deception signals in complex environments, thus significantly improving the anti-interference capability of GNSS signals.

[0169] By adjusting the parameters of the feature extraction module and the threshold setting of the autoencoder model, this method can adapt to the interference detection needs in different scenarios and has broad application potential in military communications and low-power navigation equipment.

[0170] Example 2

[0171] In this embodiment, the satellite navigation signal anti-spoofing jamming detection method of the present invention is tested in multiple scenarios based on the TEXBAT dataset. The TEXBAT dataset contains multiple simulated and real jamming scenarios, including scenarios 1, 2, 4, and 8, which represent different types of spoofing jamming and are suitable for evaluating the applicability and accuracy of the model in complex environments.

[0172] Experimental steps Data preparation: Satellite navigation signal data of scenes 1, 2, 4 and 8 are selected from the TEXBAT dataset. The deception signals of these scenes contain different eigenvalue changes to simulate the deception interference in the real environment.

[0173] Feature extraction: For the signal in each scene, four main feature values ​​are extracted: carrier-to-noise ratio (C / N0), in-phase component (Prompt_i), quadrature component (Prompt_q), and Doppler frequency shift (Carrier_Doppler_Hz). These features reflect the changes in signal strength, phase, and frequency, and are used to detect whether the signal is interfered.

[0174] Data standardization: The extracted eigenvalues ​​are Z-score standardized to eliminate dimensional differences and ensure that each eigenvalue has the same scale, thereby improving the detection accuracy of the model.

[0175] Autoencoder model training: The standardized data is input into the autoencoder model for training. The model contains two layers of encoders and two layers of decoders. The encoder compresses the input data into a low-dimensional space, while the decoder is responsible for restoring it. By calculating the reconstruction error between the input and output, the abnormality of the signal is judged.

[0176] Reconstruction error calculation and threshold setting: After the model reconstructs the input signal, it calculates the reconstruction error (mean square error, MSE) and sets the threshold T = 37. If the MSE exceeds the threshold, it is determined that the signal may be deceptively interfered.

[0177] Experimental Results

[0178] In scenarios 1, 2, 4, and 8, the model's detection results are shown in the following table:

[0179] Scenario Total number of channels Number of alarm triggers Alarm rate Remark 1 8 August 8 100％ Accurate detection 2 8 August 8 100％ Accurate detection 3 8 June 8 75％ Some missed detection 4 8 August 8 100％ Accurate detection 8 8 August 8 100％ Accurate detection

[0180] In scenarios 1, 2, 4, and 8, the method of the present invention can detect 100% of the spoofing signals in 8 channels, showing the high sensitivity and accuracy of the detection model. In scenario 3, the model detected spoofing signals in 6 channels, and some channels failed to trigger alarms. The detection rate was 75%, slightly lower than other scenarios. Overall, the model showed high detection accuracy in most scenarios.

[0181] Experimental results analysis of high accuracy scenarios: The 100% detection rate of scenarios 1, 2, 4, and 8 proves the effectiveness of the autoencoder model under multi-feature joint analysis, and can make accurate judgments on various types of deception signals.

[0182] Adaptability and robustness: The model can maintain a high detection accuracy in multiple scenarios and is suitable for complex navigation signal interference environments, especially for drones, vehicle navigation systems, and military communications.

[0183] This example verifies the applicability and efficiency of the method of the present invention in different scenarios, and provides strong data support for practical applications.

[0184] In summary, the present invention conducts joint analysis of multi-parameter eigenvalues ​​through an autoencoder model, significantly improves the detection capability of complex interference, and has high creativity. This method combines four key features of carrier-to-noise ratio, in-phase component, orthogonal component, and Doppler frequency shift for the first time in deception detection, overcomes the defect that single parameter detection of traditional methods is easily limited, and establishes the normal characteristic pattern of the signal through the low-dimensional encoding and high-dimensional reconstruction process of the autoencoder model. In particular, the present invention adopts the Z-score standardization method to make the dimensions of different features consistent, effectively improving the adaptability and detection accuracy of the model in various environments. In addition, the reconstruction error threshold of the present invention is set to 37, which not only makes the detection judgment faster and easier, but also ensures the reliability of the detection result by statistically setting the signal reconstruction error in different scenarios. Without the need for additional hardware support, the detection process is suitable for deployment in low-cost equipment with limited resources, and can effectively adapt to the actual application needs in the fields of drone navigation, vehicle navigation systems, military communications, etc. Compared with existing encryption or complex hardware methods, the present invention has significant advantages of low cost, easy deployment and sensitive detection. It has extremely high application and promotion potential, fully meets the high security requirements of various navigation systems, and demonstrates unique technical innovation and good authorization prospects.

[0185] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0186] Any process or method description in the flow chart or otherwise described herein can be understood to represent a module, fragment or portion of a code including one or more executable instructions for implementing the steps of a specific logical function or process. And the scope of the preferred embodiment of the present application includes other implementations, in which the functions may not be performed in the order shown or discussed, including in a substantially simultaneous manner or in a reverse order according to the functions involved.

[0187] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder, characterized in that: The following steps are involved: Step S1: extracting main characteristic values ​​from the received satellite navigation signal: carrier-to-noise ratio, in-phase component, orthogonal component, Doppler frequency shift; Step S2: Standardize the extracted main eigenvalues ​​to eliminate dimensional differences; Step S3: training and predicting the main eigenvalues ​​after the standardization based on the autoencoder model; Step S4: Calculate the reconstruction error of the autoencoder model, set a threshold of the reconstruction error, and determine whether the signal is abnormal based on a comparison result between the reconstruction error and the threshold, wherein when the reconstruction error exceeds the threshold, it is determined that the received signal may be subject to deception interference.

2. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 1 is characterized in that: The satellite navigation signal data in step S1 is feature extracted based on the TEXBAT dataset, wherein the TEXBAT dataset contains multiple real and simulated interference scenarios for training the autoencoder model and optimizing the threshold setting.

3. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 1 is characterized in that: The standardization of the characteristic data in step S2 adopts the Z-score standardization method, and the steps include: Calculate the average value μ of the characteristic data, the formula is as follows: Among them, f i represents the i-th original eigenvalue, n is the total number of feature data sets, μ is the mean of the feature data, which is used to reflect the central trend of the feature data and reduce the deviation; Calculate the standard deviation σ of the characteristic data using the following formula: Among them, σ is the standard deviation of the characteristic data, which is used to measure the degree of dispersion of the characteristic data relative to the mean μ and ensure the scale consistency between different characteristic values; Each eigenvalue is standardized and the original eigenvalue f i Convert to the standardized value f i ′, the formula is as follows: Among them, f i ′ is the i-th eigenvalue after standardization; After the Z-score standardization, all feature data will conform to the standard normal distribution with a mean of 0 and a standard deviation of 1.

4. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 1 is characterized in that: The autoencoder model in step S3 compresses and reconstructs the input data through an encoder and a decoder, specifically including: The encoder part compresses the four input eigenvalues ​​into a low-dimensional space, and reduces the data dimension layer by layer through a double-layer encoder; The first layer encoder is set with 3 neurons, the second layer encoder is set with 2 neurons, and both layers of encoders use the hyperbolic tangent tanh activation function; The decoder part restores the encoder's low-dimensional data to the original high-dimensional space and reconstructs output data that is as consistent as possible with the input data.

5. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 4 is characterized in that: The encoder consists of two layers: The first encoder layer receives an input vector x = [x1, x2, x3, x4], where x∈R 4 Represents an input vector containing four eigenvalues; the first layer encoder is passed through the weight matrix W1∈R 3×4 and the bias vector b1∈R 3 Calculate the hidden layer output h1∈R3; The first layer encoder uses the hyperbolic tangent activation function tanh, and the calculation formula is: h1=tanh(W1x+b1) For each hidden neuron i=1,2,3, the calculation process is as follows: in, is the weight matrix W1 from the input feature x j To the hidden neuron h 1i The weight of b 1i is the bias term that allows the model to produce non-zero output even when the input is zero.

6. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 5 is characterized in that: The second layer of the encoder receives the output h1 of the first layer and further compresses the data into a 2D space: The second layer encoder is passed through the weight matrix W2∈R 2×3 and the bias vector b2∈R 2 Calculate the output vector h2∈R 2 ; The activation function also uses hyperbolic tangent tanh, and the calculation formula is: h2=tanh(W2h1+b2) Specifically, for each hidden neuron k=1,2, the calculation process is: in, is an element in the weight matrix W2, representing the weights from the first layer of hidden neurons h 1i To the second layer of hidden neurons h 2k The weight of b 2k is the corresponding bias term.

7. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 4 is characterized in that: The decoder part contains two layers: The first decoder layer receives the 2D encoded representation h2 = [h 21 ,h 22 ] T ; Through the weight matrix W3∈R 3×2 and the bias vector b3∈R 3 Calculate the output vector h3∈R 3 ; The activation function uses the sigmoid function, and the calculation formula is as follows: h3=sigmoid(W3h2+b3) For each hidden neuron i=1,2,3, the specific calculation process is: in, is an element in the weight matrix W3, representing the weight from the kth input node to the ith hidden neuron, b 3i is the corresponding bias term.

8. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 7 is characterized in that: The second layer of the decoder receives the output h3 of the first layer of the decoder and restores it to a 4-dimensional space: The second layer decoder is passed through the weight matrix W4∈R 4×3 and the bias vector b4∈R 4 Calculate the final reconstructed output vector Make it close to the original input vector x; The activation function uses the sigmoid function, and the calculation formula is as follows: For each output node j=1,2,3,4, the specific calculation process is: in, is the reconstructed output value, is an element in the weight matrix W4, representing the 3i To the output node The weight of b 4j is the corresponding bias term.

9. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 1 is characterized in that: The step of calculating the reconstruction error and setting the threshold in step S4 includes: S41: The input multidimensional feature signal is a feature vector y = [y1, y2, ..., y n ], where: y represents the original eigenvector; n represents the dimension of the eigenvector, that is, the number of signal components; y n represents the nth component of the original signal; S42: Use the trained autoencoder model to reconstruct the input signal and obtain the reconstructed signal vector in: Represents the feature vector reconstructed by the autoencoder model; represents the i-th component in the reconstructed signal vector, that is, the i-th signal component reconstructed by the autoencoder; S43: The reconstruction error is expressed as mean square error, and the calculation formula is as follows: Where: MSE stands for mean square error, which is used to measure the reconstructed signal The difference between the original signal y; n represents the dimension of the feature vector; and i Represent the i-th component of the reconstructed signal and the original signal respectively; S44: Based on the statistical characteristics of the data, the threshold T is set to 37. In the non-spoofing scenario, the MSE value is low, below 20, and the maximum value does not exceed 37; in the spoofing scenario, the MSE value increases significantly, and the maximum value can reach more than 1200, where: T represents the threshold of the mean square error, which is used to determine whether the signal is interfered; S45: Compare the calculated reconstruction error MSE with the set threshold T: If MSE≤T, the signal is judged to be normal; If MSE>T, an abnormality is detected and it is determined that the signal may be interfered by fraud.

10. The multi-feature satellite navigation signal anti-spoofing interference detection method based on autoencoder according to claim 1, characterized in that: The method is applicable to interference detection in UAV navigation, vehicle navigation systems and military communications, and specifically includes: in UAV navigation, multi-dimensional data of the normal operation of the UAV is input into the autoencoder as a training data set for training, and abnormal conditions of the input data are judged based on the reconstruction error. If the calculated reconstruction error exceeds a set threshold, it is judged that the UAV may have suffered a GPS spoofing attack; In vehicle navigation systems and military communications, since vehicle navigation systems and military communications rely on the positioning, navigation and timing services provided by the Global Navigation Satellite System GNSS, and the main eigenvalues ​​used by both are common signal parameters.

Images