A GNSS induced spoofing detection method based on RMS sliding envelope and SVM

Abstract

The application discloses a GNSS induced spoofing detection method based on RMS sliding envelope and SVM, first, using a software receiver to capture and track visible satellites from GNSS intermediate frequency signals, taking IQ data, carrier-to-noise ratio C / N0 and Doppler frequency shift of an ELP tracking loop as original features; then setting a window length and calculating extreme points of the original features, and then using a cubic spline interpolation method to interpolate the maximum and minimum points to obtain upper and lower envelope curves of the original features; calculating a sliding root mean square of the envelope curves as new features after processing; after normalizing the new features, using principal component analysis for feature dimension reduction; finally, dividing the features into a training set and a test set, inputting test samples after training a support vector machine to obtain the final classification result. The application reduces data redundancy and operation time, is easy to implement and has low complexity, has low requirements on a receiver, greatly reduces equipment cost of the receiver, and has a wider application scenario.

Description

Technical Field

[0001] This invention relates to the field of satellite navigation interference detection technology, and in particular to a GNSS induced deception detection method based on RMS sliding envelope and SVM. Background Technology

[0002] Since 1994, the Global Positioning System (GNSS) has provided location, velocity, and time (PVT) information to civilian, commercial, and military users. Major powers are vigorously developing their own GNSS systems. China's BeiDou BDS, the United States' GPS, Europe's Galileo, and Russia's GLONASS are known as the world's four major GNSS systems. After the GNSS entered the commercial and civilian market, many components of critical infrastructure and widely used applications began to rely on the continuous availability of PVT information. If the GNSS were to suddenly shut down or be interfered with, resulting in the provision of incorrect time and location information, it would have a significant impact or even a devastating effect on various industries around the world.

[0003] The increasing complexity of the electromagnetic environment and the dramatic changes in the international situation pose serious challenges to the safe application of satellite navigation. In civilian GNSS systems, navigation signals become extremely weak after long-distance transmission, and because their structure is publicly available, they are susceptible to deception interference. With the development of science and technology, including integrated electronics, sensor technology, and radio technology, deception interference has become increasingly easier, cheaper, and more flexible to implement. Therefore, it is essential to detect, weaken, or eliminate navigation deception. Deception detection is crucial; only by correctly detecting satellite navigation deception can it be further weakened or eliminated, or even further, the location of the deception source can be determined for its removal.

[0004] In recent years, numerous solutions for GNSS spoofing detection have been proposed both domestically and internationally, including absolute power monitoring, relative power monitoring, multi-correlation peak detection, signal angle of arrival (AoA) detection, time consistency detection, autonomous integrity detection (RAIM), and navigation and time information fusion detection. Absolute power monitoring for spoofing interference detection involves setting a threshold for the power of the received signal. When the received signal power exceeds the threshold, the signal is considered spoofed. However, this method is susceptible to false alarms due to factors such as antenna type, antenna attitude, and multipath effects. AoA monitoring is based on the assumption that the spoofing interference source is a single-antenna source, and the transmitted spoofing signal has the same azimuth angle at the receiver antenna. By monitoring the AoA, multiple antennas of the receiver can be used to detect the AoA of the received satellite navigation signal, and the spoofing interference can be identified by determining whether the AoA and AoA are the same. The basic principle of time consistency monitoring is that the information carried by most spoofing signals is not synchronized with the time of the real signal. Spoofing devices introduce a Doppler shift to the deceived receiver. This shift is monitored using a local, highly stable, and accurate clock, while simultaneously comparing the time difference between the observed time and the satellite signal ephemeris time. This method is primarily designed to address generated spoofing interference. The basic principle of receiver autonomous integrity monitoring (RAIM) is to use a sampling detection method on the received satellite signals to detect spoofing. A receiver needs at least four satellite signals to determine its position. Therefore, four satellite signals are randomly selected multiple times from all received signals to calculate the receiver's position. By comparing these multiple position calculations, the presence of spoofing interference in the signal is determined. This method requires at least five satellite signals to be received.

[0005] The methods and technologies mentioned above achieve good detection performance by deploying additional hardware equipment, but this increases economic costs or implementation difficulty, making them impractical for small and medium-sized enterprises and thus difficult to popularize. In contrast, deceptive interference detection technology based on information computation uses algorithms to analyze and process the data acquired by the receiver for judgment, without changing the satellite navigation signal system or installing additional equipment. This is currently the focus of research in deceptive interference detection technology. Summary of the Invention

[0006] Based on the technical problems existing in the background technology, this invention proposes a GNSS induced deception detection method based on RMS sliding envelope and SVM.

[0007] This invention proposes a GNSS induced spoofing detection method based on RMS sliding envelope and SVM, comprising the following steps:

[0008] Step 1: Use a GNSS software receiver to capture visible satellites from the GNSS intermediate frequency signal received by the radio frequency front end, track the captured visible satellites, calculate the SQM characteristics and carrier-to-noise ratio C / N0 based on the IQ branch outputs of the lead, instant, and lag loops during the tracking phase, and estimate the Doppler frequency shift. Use the five calculated characteristics as the original characteristics.

[0009] Step 2: After calculating the RMS sliding envelope of the 5 features in Step 1, use mean-variance normalized features and principal component analysis for dimensionality reduction, and use the processed data as training data for the new features.

[0010] Step 3: Train the Support Vector Machine (SVM) classifier using the training data obtained in Step 2. After training, a trained SVM model is obtained. The trained SVM model can be used to determine whether there is a deceptive signal. The predicted label is compared with the real label, various evaluation indicators are calculated, and the parameters are adjusted to optimize the model performance.

[0011] Step 4: Use the SVM model trained in Step 3 to detect the signals received by the GNSS receiver, and complete the GNSS spoofing interference detection based on RMS sliding envelope and SVM in signal acquisition and tracking.

[0012] The detection steps of this invention are as follows:

[0013] (1) Process and save the GNSS intermediate frequency signal received by the radio frequency front end;

[0014] (2) Use a GNSS software receiver to capture visible satellites;

[0015] (3) Track and calculate the SQM characteristics of the j-th visible satellite, including Ratio, Delta, and ELP:

[0016] (3.1) Calculate the in-phase and quadrature branch outputs of the leading, instantaneous, and lagging loops of the tracking loop, denoted as I. E I P I L and Q E Q P Q L :

[0017] (3.2) Calculate the SQM characteristics of the j-th satellite:

[0018]

[0019]

[0020]

[0021] Among them, IE I P I L The in-phase components of the outputs of the correlated branches of the lead, immediate, and lag codes, respectively, Q E Q P Q L These are the orthogonal components of the outputs of the related branches of the lead, instant, and lag codes, respectively.

[0022] (4) Estimate the carrier-to-noise ratio C / N0 of the j-th satellite:

[0023] (4.1) Calculate the average normalized values ​​of broadband power and narrowband power based on the outputs of the in-phase and quadrature branches calculated in step (3.1):

[0024]

[0025] Where K is the number of correlators, T is the bit duration of the navigation data, and M is the number of outputs of the correlation integrator.

[0026] (4.2) C / N0 can be estimated based on the average normalized value of broadband power and narrowband power:

[0027]

[0028] (5) Estimate the Doppler shift of the j-th satellite:

[0029]

[0030] Among them, f s V is the carrier frequency for satellite transmission. s Let C be the tangential velocity of the satellite, C be the signal propagation speed, and A be the angle between the user's radius vector from the satellite and its tangential velocity vector.

[0031] (6) Calculate the RMS sliding envelope of the original features of the j-th satellite obtained in steps (3)-(5), including Ratio, Delta, ELP, C / N0 and f. d :

[0032] (6.1) Find the maximum points of all original features, denoted as I. m ;

[0033] (6.2) to I m The upper envelope curve is obtained by interpolating the set of maxima in the region.

[0034] (6.3) Find the minimum points of all original features, denoted as I. n ;

[0035] (6.4) to I n The lower envelope curve is obtained by interpolating the set of local minimum points in the region.

[0036] (6.5) The length w and sliding step μ of the sliding window;

[0037] (6.6) Calculate the sliding RMS of the envelope curve in steps (6.2) and (6.4). The formula for calculating the sliding RMS of the input discrete sequence X(k) is:

[0038]

[0039] (7) Repeat steps (3)-(6) until the RMS sliding envelopes of all the visible satellites captured in step (2) are calculated;

[0040] (8) Take all the processed data from step (7) as new features and perform preprocessing:

[0041] (8.1) Normalize the data using mean and variance preprocessing:

[0042]

[0043] (8.2) Principal component analysis is used to select features from normalized data to reduce data dimensionality while ensuring no loss of data information, and to check the quality of combined features. Reprojection can be represented as follows:

[0044]

[0045] Among them, a ij Let n be a constant, and n be the projection space. The dimension, b ij PC is a set of constants in the iteration process, and iteration continues until the error between the reprojected features and the original features is minimized. i A linear combination of the original features:

[0046]

[0047] (9) Set the parameters of SVM and divide the dataset.

[0048] (9.1) Support Vector Machines (SVMs) construct hyperplanes for classification by mapping the original training data to a multidimensional space. The constructed hyperplanes will separate different categories. In SVM parameter settings, C is the penalty coefficient, which can be understood as adjusting the weight of the two indicators (margin size and classification accuracy) in the optimization direction, i.e., the tolerance for error. The higher the C, the less tolerant it is of errors, and the easier it is to overfit. The lower the C, the easier it is to underfit. If C is too large or too small, the generalization ability will be poor. Set C to 0.98;

[0049] (9.2) Select the RBF function as the kernel;

[0050] (9.3) gamma is a parameter that comes with the RBF function after selecting it as the kernel. It implicitly determines the distribution of the data after it is mapped to the new feature space. The larger the gamma, the fewer the support vectors, and the smaller the gamma value, the more support vectors. The number of support vectors affects the training and prediction speed. Set gamma to 0.2.

[0051] (9.4) When splitting the dataset, 70% is used as training data and the remaining 30% is used as test data.

[0052] (10) Training and evaluation model.

[0053] (10.1) After setting the parameters and dividing the dataset according to step (9), train the SVM on the training set;

[0054] (10.2) The model is then evaluated on the test set. After inputting the test sample, the SVM outputs a prediction label of 1 or -1, where 1 represents hypothesis H1, that is, the detected sample is a spoofed GSNN signal; and -1 represents hypothesis H0, that is, the detected sample is a normal signal and has not been subjected to GNSS spoofing attack.

[0055] This invention discloses a GNSS induced spoofing detection method based on RMS sliding envelope and SVM. This method applies RMS of signal envelope and machine learning algorithms to GNSS spoofing detection. It is a spoofing detection method based on SQM technology and signal waveform observations. Compared with the traditional SQM detection method, this invention reduces data redundancy by extracting the upper and lower envelopes, making the data distribution of spoofing signals and real signals more discrete. Therefore, the detection rate is greatly improved, and better detection performance is achieved by using fewer features and less computation time. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating an embodiment of the present invention;

[0057] Figure 2 A schematic diagram of a satellite navigation system's induced deception interference and detection model;

[0058] Figure 3 A schematic diagram of the correlation peak search results for the FFT algorithm-based acquisition process of a navigation satellite software receiver;

[0059] Figure 4 A schematic diagram of the tracking loop framework and IQ data acquisition;

[0060] Figure 5 This shows the distribution of real and spoofed ELP signals from the visible satellite GPS-03.

[0061] Figure 6The distribution of real and spoofed ELP signals and data from GPS-03 under different window lengths;

[0062] Figure 7 ROC curves for GNSS spoofing detection under different window lengths;

[0063] Figure 8 Statistics on F1 scores and time consumption for GNSS deception detection under different window lengths;

[0064] Figure 9 The diagram illustrates the confusion matrix detected under different deception scenarios. Detailed Implementation

[0065] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0066] Reference Figure 1-9 A GNSS induced spoofing detection method based on RMS sliding envelope and SVM includes the following steps:

[0067] 1. Process and save the GNSS intermediate frequency signals received by the radio frequency front-end, and use the GNSS software receiver to capture visible satellites from them, and save the result data;

[0068] Step 1 specifically includes:

[0069] 1.1 GNSS signals propagate through space and are incident on the user's GNSS receiving antenna. After attenuation and thermal noise interference, the amplitude is greatly weakened, and the frequency is too high. The radio frequency front-end uses a combination of amplifiers, downconverters, bandpass filters, and its own oscillator to give the signal a higher amplitude, lower frequency, and narrower bandwidth. The front-end output of satellite k is:

[0070]

[0071] Where P, τ, and φ represent power, time delay, and carrier phase, respectively, and T s The sampling interval is C, and C and D are time intervals nT and T, respectively. s The corresponding extended code sequence and navigation data sequence, ζ(nT) s ) is zero mean and variance σ 2 Additive white Gaussian noise, k is the PRN number of the satellite, and f is the intermediate frequency to which the carrier frequency is down-converted by the front end;

[0072] Since the signal structure of the deception jamming is generated by mimicking the real GPS signal, the deception signal received by the receiver is:

[0073]

[0074] In this context, the superscripts a and s represent the real signal and the spoofed signal, respectively. When the receiver is spoofed, the received signal is a combination of signals from multiple satellites.

[0075]

[0076] Among them, ϒ a and ϒ s These are the PRN sets representing the real signals and the spoofed signals, respectively.

[0077] The cross-correlation function between the local code and the received hybrid satellite signal can be obtained by carrier stripping and coherent integration over 1 ms, expressed as:

[0078]

[0079] Among them, R a (·) represents the cross-correlation function between the real satellite signal and the local code; R s (·) represents the cross-correlation function between the deception signal and the local code; ζ'(t,τ) represents the cross-correlation result between the filtered Gaussian white noise and other satellite signals and the local code;

[0080] 1.2 The receiver must know which satellites are visible to the user, therefore it needs to acquire satellite signals. The acquisition phase mainly focuses on frequency and code phase, searching for the maximum value of the correlation peak of the signal power. When the maximum value exceeds a set threshold, the satellite is acquired and tracked at the corresponding frequency and phase. The block diagram of the tracking loop is as follows: Figure 4 As shown, the in-phase and quadrature components of the outputs of the lead, instantaneous, and lag code related branches in the tracking process can be expressed as:

[0081]

[0082]

[0083] Where d is the correlator interval, τ is the time delay, θ0 is the initial phase, and θ1 is the phase delay;

[0084] 2. Calculate the original features. The original features include CN0, Doppler frequency shift, and SQM features calculated based on the IQ branch output of the ELP tracking loop;

[0085] Step 2 specifically includes:

[0086] 2.1 Calculation of Carrier-to-Noise Ratio (C / N0): The carrier-to-noise ratio (C / N0) is estimated based on the ratio of the signal's broadband power to its narrowband power, and is used to measure the quality of the acquired GNSS signal.

[0087]

[0088] Where T is the bit duration of the navigation data, M is the number of outputs of the relevant integrator, and μ NP The average normalized value of the ratio of broadband power to narrowband power is calculated using the following formula:

[0089]

[0090] In the formula: K is the number of correlators;

[0091] 2.2 SQM Feature Calculation: Commonly used statistical indicators for SQM monitoring include Ratio, Delta, and ELP. Specific formulas are as follows:

[0092]

[0093]

[0094]

[0095] Among them, I E I P I L The in-phase components of the outputs of the correlated branches of the lead, immediate, and lag codes, respectively, Q E Q P Q L These are the orthogonal components of the outputs of the related branches of the lead, instant, and lag codes, respectively.

[0096] 3. Calculate the extreme points of the original feature. Let X(t) be a discrete sequence, and satisfy:

[0097] ,

[0098]

[0099] X(t) has M maxima and N minima, corresponding to sequence indices (I0, I ... m , I n ), space (T) m , T n The function values ​​(U, V) are denoted as:

[0100]

[0101]

[0102]

[0103]

[0104] 4. Use cubic spline interpolation to interpolate the extreme points and calculate the sliding RMS of the characteristic envelope curve;

[0105] Step 4 specifically includes:

[0106] 4.1 Using the cubic spline interpolation function y i = f (x i For each segmented interval [t] i , t i+1 Perform interpolation to find all minimum points I excluding endpoints. n The lower envelope curve of the interpolation interval is formed by these piecewise interpolation curves. The lower envelope curve formed by these piecewise interpolation curves is the lower envelope curve excluding the left and right endpoints. Similarly, the upper envelope curve is obtained for the difference of all maximum points.

[0107] 4.2 Setting the window length: The window length determines the number of sample points in the sliding window, which has a significant impact on the smoothness of the entire envelope and the data distribution. The window length is set based on the sample length.

[0108] 4.3 Calculate the sliding RMS of the envelope. The calculation formula is as follows:

[0109]

[0110] Where w is the window length and μ is the sliding step size.

[0111] Figure 5 The original data and RMS sliding envelope data of the SQM feature ELP of the GPS-03 navigation satellite are shown.

[0112] 5. Data preprocessing and dataset creation.

[0113] Step 5 mainly includes:

[0114] 5.1 Data Standardization: After obtaining the RMS sliding envelope, the features are standardized using the following formula:

[0115]

[0116] in, This represents the expectation of the i-th eigenvalue of the k-th captured satellite. Let represent the standard deviation of the i-th eigenvalue of the k-th captured satellite.

[0117] 5.2 Feature Reduction: Principal Component Analysis (PCA) is a technique for reducing the dimensionality of a dataset. The basic method is to reduce the dimensionality of the dataset from the original space... To another space The linear projection of can be expressed by the following equation:

[0118]

[0119] Among them, a ij PC is a constant. i The original features are linear combinations, and n is the projection space. Dimensions.

[0120] Principal component analysis uses reprojection to examine the quality of combined features. Reprojection can be represented as follows:

[0121]

[0122] Among them, b ij These are a set of constants in the iteration process, which are iterated until the error between the reprojected features and the original features is minimized;

[0123] 5.3 Dataset partitioning: To train and evaluate the SVM classifier, 70% of the dataset is used for training, and 30% is used for evaluating the trained classifier.

[0124] 6. Train and test the SVM classifier.

[0125] Step 6 mainly includes:

[0126] 6.1 Training the SVM classifier involves constructing a hyperplane for classification by mapping the original training data to a multidimensional space. The constructed hyperplane will separate different classes. The hyperplane with the largest distance between support vectors is set as the final hyperplane used for classification. The data points closest to the hyperplane are called "support vectors." Therefore, to classify with minimal classification error, our goal is to maximize the distance between edge data points and the hyperplane. To find the optimal hyperplane, we can define it as follows:

[0127]

[0128] Among them, w T is the normal vector of the hyperplane, containing the weights of different data points, X is the sample point that defines the hyperplane, and b is the bias constant.

[0129] The dataset consists of vectors z = z1, z2, ..., z m and corresponding label y i Composition, where y i =1 indicates a positive sample, y i =-1 represents a negative sample. Further, a decision boundary is defined, which should correctly divide all points into:

[0130]

[0131] Let all z = z1, z2, ..., z mIn the hyperplane separating the data, there exists only one hyperplane that maximizes the separation margin between the two classes of samples. This optimal solution can be found through optimization.

[0132]

[0133] 6.2 In SVM parameter settings, C is the penalty coefficient, which can be understood as adjusting the weight of the two indicators (margin size and classification accuracy) in the optimization direction, i.e., the tolerance for error. The higher the C, the less tolerant it is of errors, and the easier it is to overfit; the lower the C, the easier it is to underfit; if C is too large or too small, the generalization ability deteriorates. Set C to 0.98;

[0134] 6.3 Select the RBF function as the kernel;

[0135] 6.4 Gamma is a parameter inherent to the RBF function when choosing it as the kernel. It implicitly determines the distribution of data after mapping to the new feature space. A larger gamma value results in fewer support vectors, and a smaller gamma value results in more support vectors. The number of support vectors affects the speed of training and prediction. We set gamma to 0.2.

[0136] 7. Evaluate the SVM classifier.

[0137] 7.1 After training the SVM classifier, evaluate it on the test sample set, first determining the evaluation metrics. The evaluation metrics include the Area Under the ROC (Receiver Operating Characteristic) curve (AUC), accuracy, precision, recall, and F1 score. AUC is defined as the area under the ROC curve and the coordinate axis; this area will not exceed 1. The closer the AUC is to 1.0, the higher the realism of the detection method; below 0.5, the realism is the lowest, and it has no application value. Accuracy is the proportion of correctly classified samples to the total number of samples, representing the overall accuracy; it becomes ineffective when the samples are imbalanced. Precision is the ratio of correctly classified positive samples to all positive samples, representing only the accuracy of positive sample detection. Recall is the ratio of correctly predicted positive samples to all positive samples, representing the probability of deception not being missed. The F1 score measures both recall and precision. The calculation methods for the above metrics are as follows:

[0138] ,

[0139] ,

[0140] 7.2 Input test samples, model output predicted labels, SVM output predicted labels 1 or -1, where 1 represents hypothesis H1, that is, the detected sample is a spoofed GSNN signal; -1 represents hypothesis H0, that is, the detected sample is a normal signal and has not been subjected to GNSS spoofing attack.

[0141] 7.3 Compare the predicted label y_pred with the true label y_true in the above steps to obtain the number of correctly classified and incorrectly classified samples, namely TP, TN, FP, and FN. Substitute these values ​​into the formula in step (7.1) to solve for each index.

[0142] 7.4 Evaluate each indicator in the above steps, adjust the parameter settings in steps (6.2)-(6.4), and re-implement steps (7.1)-(7.3) until the model is trained to the optimal state.

[0143] 8. Test on real-time samples. After confirming that the trained model can achieve good performance on the test set, test it on real-time received samples to detect GNSS spoofing interference in real time.

[0144] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A GNSS induced spoofing detection method based on RMS sliding envelope and SVM, characterized in that, Includes the following steps: Step 1: Use a GNSS software receiver to acquire visible satellites from the GNSS intermediate frequency signal received by the RF front-end. Track the acquired visible satellites and calculate the SQM characteristics and carrier-to-noise ratio C / N0 based on the IQ branch outputs of the lead, instant, and lag loops during the tracking phase. Track and calculate the SQM characteristics of the j-th visible satellite: Including Ratio, Delta, and ELP, and simultaneously estimating Doppler frequency shift, the five calculated features are used as the original features; Step 2: After calculating the RMS sliding envelope of the 5 features in Step 1, use mean-variance normalized features and principal component analysis for dimensionality reduction, and use the processed data as training data for the new features. Step 3: Train the Support Vector Machine (SVM) classifier using the training data obtained in Step 2. After training, a trained SVM model is obtained. The trained SVM model can be used to determine whether there is a deceptive signal. The predicted label is compared with the real label, various evaluation indicators are calculated, and the parameters are adjusted to optimize the model performance. Step 4: Use the SVM model trained in Step 3 to detect the signals received by the GNSS receiver, and complete the GNSS spoofing interference detection based on RMS sliding envelope and SVM in signal acquisition and tracking.

2. The GNSS induced spoofing detection method based on RMS sliding envelope and SVM according to claim 1, characterized in that, In step 2, the RMS sliding envelope is calculated as follows: find the extreme points of all original features, where the set of maximum points is denoted as I. m The set of minimum points is denoted as I. n The maximum value represents the maximum boundary of the change of the original feature within a certain interval, and the minimum value represents the minimum boundary of the change of the original feature within a certain period.

3. The GNSS induced spoofing detection method based on RMS sliding envelope and SVM according to claim 1, characterized in that, In step 2, the RMS sliding envelope is calculated by interpolating the intervals of the extreme point set and traversing the entire set to obtain the upper and lower envelope curves of the original feature.

4. The GNSS induced spoofing detection method based on RMS sliding envelope and SVM according to claim 1, characterized in that, In step 2, the sliding RMS value of the RMS sliding envelope X is: Where w is the length of the sliding window, μ is the sliding step size, and ϒ(i) is the root mean square of the i-th window.

5. The GNSS induced spoofing detection method based on RMS sliding envelope and SVM according to claim 4, characterized in that, The length w of the sliding window determines the flatness of the envelope and the redundancy of the data. The length of the sliding window does not exceed 20% of the original feature sample length.

6. The GNSS induced spoofing detection method based on RMS sliding envelope and SVM according to claim 1, characterized in that, In step 1, the extracted feature parameters F=[Ratio, Delta, ELP, C / N0, f d The calculation method is as follows: ， Among them, I E I P I L The in-phase components of the outputs of the correlated branches of the lead, immediate, and lag codes, respectively, Q E Q P Q L These are the orthogonal components of the outputs of the lead, instant, and lag code correlation branches, respectively; K is the number of correlators, T is the bit duration of the navigation data, and M is the number of outputs of the correlation integrator; f s V is the carrier frequency for satellite transmission. s Let C be the tangential velocity of the satellite, C be the signal propagation speed, and A be the angle between the user's radius vector from the satellite and its tangential velocity vector.

7. The GNSS induced spoofing detection method based on RMS sliding envelope and SVM according to claim 1, characterized in that, In step 3, if a spoofing signal exists, the SVM outputs a prediction label of 1; otherwise, it outputs a prediction label of -1, corresponding to binary hypotheses H0 and H1, respectively.

8. The GNSS induced spoofing detection method based on RMS sliding envelope and SVM according to claim 7, characterized in that, The binary assumption refers to the assumption regarding the existence or non-existence of GNSS satellite spoofing signals, expressed by the following formula: Where T is the detection statistic, ζ is the channel noise, the superscripts a and s represent the real signal and the spoof signal, respectively, and the subscript k represents the k-th satellite.

Images