Self-adaptive navigation interference optimization method based on time sequence convolutional neural network

Abstract

The invention provides a self-adaptive navigation interference optimization method based on a time sequence convolutional neural network, and relates to the field of communication, and the method comprises the steps: collecting a navigation signal transmitted by a target satellite to a target receiver in real time, carrying out the feature analysis, obtaining the feature information of the navigation signal, and when the navigation signal is locked by the target receiver, carrying out the self-adaptive navigation interference optimization. Generating a first interference signal until the target receiver loses the locking of the navigation signal, generating a second interference signal when the navigation signal is not locked by the target receiver, inputting the feature information into a preset signal parameter prediction model to predict the signal parameter of the navigation signal, and adjusting the signal parameter of the interference signal based on the prediction result. And then sending to a target receiver. By means of the method, the whole process from communication cut-off to misleading of the target is achieved, the flexibility and adaptability of interference are enhanced, the success rate and comprehensive efficiency of interference are improved, and redundant consumption of interference resources is remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the field of communications, and in particular to an adaptive navigation interference optimization method based on a temporal convolutional neural network. Background Art

[0002] With the advancement of navigation technology, navigation jamming technology has also gained widespread application. Jamming systems are deployed near key civilian sites such as airports and nuclear power plants to prevent unauthorized drone intrusion or malicious attacks. At large-scale events such as concerts and sporting events, navigation jamming can prevent drones from operating normally, preventing illegal filming or disruption of the event. With the continuous development of navigation positioning, diversified anti-jamming measures, and enhanced mobile connectivity, the demand for navigation jamming optimization technology is growing.

[0003] However, existing interference optimization methods primarily target specific signals for interference and suffer from issues such as a single interference dimension, poor real-time parameter setting, and slow convergence of the interference optimization algorithm. As low-orbit satellites become the primary providers of positioning, navigation, and timing services, their high signal power, narrow beams, and wide coverage make them highly resistant to interference. This has significantly reduced the effectiveness of traditional interference optimization methods. Summary of the Invention

[0004] In order to solve the problems of single interference mode and poor real-time parameter setting in the existing technology and improve the interference effect, the present invention proposes an adaptive navigation interference optimization method based on a temporal convolutional neural network, which includes:

[0005] Step 1: collecting the navigation signal transmitted by the target satellite to the target receiver in real time, and performing feature analysis to obtain feature information of the navigation signal;

[0006] Step 2, determining whether the navigation signal is locked by the target receiver;

[0007] Step 3: When the navigation signal is locked by the target receiver, generating a first interference signal until the target receiver loses the lock on the navigation signal; wherein the first interference signal is used to establish a navigation signal denial environment;

[0008] Step 4: When the navigation signal is not locked by the target receiver, generating a second interference signal; wherein the second interference signal is used to cause the target receiver to generate an erroneous positioning;

[0009] Step 5: Inputting the characteristic information into a preset signal parameter prediction model to predict the signal parameters of the navigation signal to obtain a parameter prediction result of the navigation signal; wherein the signal parameter prediction model is built based on a time domain convolutional neural network;

[0010] Step 6: Adjust the signal parameters of the first interference signal or the second interference signal based on the parameter prediction result, and then send the adjusted first interference signal or the second interference signal to the target receiver.

[0011] Optionally, the second interference signal includes a forwarding interference signal, and step 4 includes:

[0012] Step 41: When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal is in an unknown state, performing signal conversion processing on the navigation signal to obtain a forwarding interference signal.

[0013] Optionally, the forwarded interference signal includes a directly forwarded signal; and step 41 includes:

[0014] When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal and the modulation mode of the navigation signal are both in an unknown state, delay processing is performed on the navigation signal to obtain a directly forwarded signal.

[0015] Optionally, the forwarded interference signal includes a partially reconstructed forwarded signal; and step 41 includes:

[0016] When the navigation signal is not locked by the target receiver, the structure of the pseudo-random code included in the navigation signal is in an unknown state, and the modulation method of the navigation signal is in a known state, the navigation signal is processed based on the symbol domain to obtain a partially reconstructed forwarding signal.

[0017] Optionally, the symbol domain-based processing includes at least one of superposition processing, replacement processing and phase flip processing.

[0018] Optionally, the second interference signal includes a pseudo low-orbit satellite signal;

[0019] The step 4 comprises:

[0020] When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal is in a known state, parsing the navigation signal to obtain a parsing result;

[0021] A pseudo low-orbit satellite signal is generated according to the analysis result and the structure of the pseudo-random code; the pseudo low-orbit satellite signal is used to be locked by the target receiver when the target receiver has not locked the navigation signal, and then send error position information to the target receiver.

[0022] Optionally, a type of the first interference signal includes at least one of narrowband interference, broadband noise interference and swept frequency interference.

[0023] Optionally, the number of jammers used to generate the interference signal is at least two, and the interference signal received by the target receiver is a superimposed signal of the interference signals respectively sent by the at least two jammers.

[0024] Optionally, the parameter prediction result includes predicted power and predicted frequency;

[0025] The step 6 comprises:

[0026] When the interference signal is a first interference signal, adjusting the power of the first interference signal so that the adjusted power of the first interference signal is greater than the predicted power, and then sending the adjusted first interference signal to the target receiver;

[0027] When the interference signal is a second interference signal, the frequency of the second interference signal is adjusted to the predicted frequency, and then the adjusted second interference signal is sent to the target receiver.

[0028] Optionally, the signal parameter prediction model is obtained by training based on signal parameter residual data; wherein the signal parameter residual data is represented as the difference between the true signal parameter and the predicted signal parameter obtained based on polynomial regression.

[0029] The present invention has the following advantages:

[0030] The present invention provides an adaptive navigation interference optimization method based on a time-domain convolutional neural network. The method collects navigation signals in real time and performs feature analysis to obtain characteristic information of the navigation signals. It then determines whether the navigation signals are locked onto by a target receiver. If the navigation signals are locked onto by the target receiver, a first interference signal is generated to create a navigation signal denial environment until the target receiver loses lock on the navigation signals. If the navigation signals are not locked onto by the target receiver, a second interference signal is generated to cause the target receiver to misposition. The interference signal is then adjusted by inputting the characteristic information into a preset signal parameter prediction model based on a time-domain convolutional neural network to predict the signal parameters of the navigation signals. The interference signal parameters are then adjusted based on the parameter prediction results, and the adjusted interference signal is then transmitted.

[0031] Therefore, at the interference method level, the present invention completes the whole process from cutting off communication to deceiving the target through flexible switching of suppressive interference to build a navigation signal denial environment and deceptive interference to mislead the target receiver. This breaks through the limitations of the traditional single mode, enhances the flexibility and adaptability of interference, and synergistically improves the success rate and comprehensive effectiveness of interference through different interference methods; at the parameter control level, the parameter prediction model built by the time series convolutional neural network is used to realize the refined dynamic control and real-time control of the interference signal parameters, which significantly reduces the redundant consumption of interference resources while improving the interference success rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments with reference to the accompanying drawings, in which:

[0033] Figure 1 A flowchart of the steps of an adaptive navigation interference optimization method based on a temporal convolutional neural network provided by an embodiment of the present invention;

[0034] Figure 2 A flowchart of another method for adaptive navigation interference optimization based on a temporal convolutional neural network provided by an embodiment of the present invention;

[0035] Figure 3 A schematic diagram of the structure of a residual block of a temporal convolutional neural network provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0037] Reference Figure 1 , a flowchart of a method for adaptive navigation interference optimization based on a temporal convolutional neural network provided by an embodiment of the present invention is provided, the method comprising:

[0038] Step 1: collect the navigation signal transmitted by the target satellite to the target receiver in real time, and perform feature analysis to obtain feature information of the navigation signal.

[0039] Among them, the characteristic information of the navigation signal includes but is not limited to environmental characteristics such as spectrum characteristic information, Doppler frequency shift information and power baseline. Specifically, the satellite downlink signal, that is, the navigation signal transmitted by the target satellite to the target receiver, can be collected through environmental perception equipment, and characteristic analysis can be performed to realize environmental characteristic monitoring of the navigation signal, so as to determine different interference methods according to different navigation signals and environmental characteristics, generate different interference signals, and thus realize multi-dimensional intelligent interference of the navigation signal.

[0040] Step 2: Determine whether the navigation signal is locked by the target receiver.

[0041] Receiver lock-on to the navigation signal means the receiver has successfully established a stable connection with the target navigation satellite's signal, enabling it to continuously decode its navigation information and calculate position, velocity, and time. By determining whether the navigation signal is locked onto the target receiver, the system can determine targeted jamming methods for different situations. This not only increases the jamming success rate, but also significantly reduces redundant jamming resource consumption and improves jamming effectiveness.

[0042] In an embodiment of the present invention, it is possible to determine whether the navigation signal is locked onto the target receiver by electronic reconnaissance, for example, by collecting data output by the target receiver and signals transmitted by the RF front end. It is also possible to determine by the behavior of the device to which the target receiver belongs. For example, if the receiver in a drone loses lock on the navigation signal, the drone may stop moving or make an emergency landing.

[0043] Step 3: When the navigation signal is locked by the target receiver, generate a first interference signal until the target receiver loses the lock on the navigation signal; wherein the first interference signal is used to construct a navigation signal denial environment.

[0044] When the target receiver locks onto the navigation signal, the jamming mode is determined to be a suppression jamming mode, generating a suppression jamming signal, i.e., a first jamming signal. The first jamming signal can be at least one of narrowband jamming, broadband noise jamming, and swept-frequency jamming. The suppression jamming mode creates a navigation signal denial environment by jamming multiple different signals, causing the target receiver to gradually lose lock on the navigation signal.

[0045] As an example, the first interference signal can be expressed as follows:

[0046] , (1)

[0047] in, is the number of jammers; Indicates the The noise envelope of the first interference signal of the nth type generated by the jammer obeys the Rayleigh distribution; Indicates the The interference carrier frequency of each jammer can be adjusted in real time according to environmental observation data, and can cover multiple frequency bands at the same time; Indicates the The random phase of the jammer; Re ( ) represents a function for extracting the real part of a complex number; is the imaginary number symbol;

[0048] Step 4: When the navigation signal is not locked by the target receiver, generate a second interference signal; wherein the second interference signal is used to cause the target receiver to generate an erroneous positioning.

[0049] When the target receiver loses lock on to the navigation signal, a second jamming signal containing erroneous position information is generated, deceiving the target receiver into mispositioning. For example, the second jamming signal can be generated by delaying, symbol flipping, or overlaying the navigation signal. Alternatively, the second jamming signal can be generated by analyzing the navigation signal, forcing the target receiver to lock onto the second jamming signal even when it is not locked on to the navigation signal. The jamming signal then continues to send erroneous position information to the target receiver, achieving navigation interference control.

[0050] By flexibly switching between suppressive jamming to create a navigation signal denial environment and deceptive jamming to mislead the target receiver, the entire process from communication interruption to target deception is achieved. This breaks through the limitations of traditional single-mode jamming, avoiding the blind spots of single-scale effects, enhancing the flexibility and adaptability of jamming, and synergizing different jamming methods to improve the success rate and overall effectiveness of jamming.

[0051] Step 5: Input the characteristic information into a preset signal parameter prediction model to predict the signal parameters of the navigation signal to obtain a parameter prediction result of the navigation signal; wherein the signal parameter prediction model is built based on a time domain convolutional neural network.

[0052] Signal parameters include, but are not limited to, power, frequency, Doppler shift, code phase, and other information. The signal parameter prediction model built on a time-domain convolutional neural network combines causal convolution with dilated convolution, enabling full-sequence parallel computing in modeling. This allows for efficient processing of navigation signal time series data and accurately captures the time-varying patterns of signal parameters.

[0053] As an example, the process of dilated causal convolution can be expressed as follows:

[0054] , (2)

[0055] in, represents the output of the dilated causal convolution, d is the dilation factor, which is used to control the sampling interval of the convolution kernel. is the convolution kernel weight, k is the size of the convolution kernel, ” is the convolution symbol; The input signal for the convolution.

[0056] Step 6: Adjust the signal parameters of the first interference signal or the second interference signal based on the parameter prediction result, and then send the adjusted first interference signal or the second interference signal to the target receiver.

[0057] For different situations, using different interference strategies to adjust the interference signal helps to improve the success rate of interference. Specifically, when the receiver locks the navigation signal and the interference signal generated by the jammer is the first interference signal, the purpose is to suppress the navigation signal through covering interference, thereby causing the target receiver to lose the lock on the navigation signal. Therefore, the power of the adjusted first interference signal is greater than the predicted power. For example, the difference between the power of the adjusted first interference signal and the predicted power is greater than 20dB. The specific value can be adjusted according to the actual situation. When the receiver loses the lock on the navigation signal and the interference signal generated by the jammer is the second interference signal, the purpose is to transmit the wrong position information to the target receiver. To transmit the information, it is first necessary to ensure that the target receiver will receive the second interference signal. Therefore, the frequency of the adjusted second interference signal is equal to the predicted frequency. Of course, in the actual interference process, it is not only the frequency, power and other parameters that need to be adjusted. Other parameters can be determined according to the communication situation. How to adjust them is not described in detail in the present invention.

[0058] A parameter prediction model built using a time-series convolutional neural network is used to predict parameters, and the interference signal is adjusted in real time based on the predicted parameters, realizing the transition of interference energy from static configuration to dynamic optimization. While improving the interference success rate, the redundant consumption of interference resources is significantly reduced.

[0059] In summary, an embodiment of the present invention provides an adaptive navigation interference optimization method based on a time-domain convolutional neural network. Navigation signals are collected in real time and feature analysis is performed to obtain feature information of the navigation signals. Then, it is determined whether the navigation signals are locked by the target receiver. When the navigation signals are locked by the target receiver, a first interference signal is generated to construct a navigation signal denial environment until the target receiver loses the lock on the navigation signals. When the navigation signals are not locked by the target receiver, a second interference signal is generated to cause the target receiver to produce an erroneous positioning. Subsequently, the interference signal is adjusted, and the feature information is input into a preset signal parameter prediction model based on a time-domain convolutional neural network to predict the signal parameters of the navigation signal. The signal parameters of the interference signal are adjusted according to the parameter prediction results, and then the adjusted interference signal is sent.

[0060] Therefore, at the interference method level, the embodiment of the present invention completes the whole process from cutting off communication to deceiving the target through flexible switching of suppressive interference to build a navigation signal denial environment and deceptive interference to mislead the target receiver. This breaks through the limitations of the traditional single mode, enhances the flexibility and adaptability of interference, and synergistically improves the success rate and comprehensive effectiveness of interference through different interference methods; at the parameter control level, the parameter prediction model built by the time series convolutional neural network is used to realize the refined dynamic control and real-time control of the interference signal parameters, which significantly reduces the redundant consumption of interference resources while improving the interference success rate.

[0061] Optionally, the second interference signal includes a forwarding interference signal, and step 4 includes:

[0062] Step 41: When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal is in an unknown state, performing signal conversion processing on the navigation signal to obtain a forwarding interference signal.

[0063] Among them, a pseudo-random code is a binary code sequence that looks like random noise but is actually deterministic and reproducible. It is used for synchronization of navigation signals and receivers, calculation of pseudoranges, etc. Signal transformation includes but is not limited to delay processing, sign flipping, substitution, phase flipping, etc.

[0064] Reference Figure 2 A flowchart of another adaptive navigation jamming optimization method based on a sequential convolutional neural network, provided by an embodiment of the present invention, is provided. This method uses intelligent judgment to determine whether the receiver has locked onto the navigation signal and whether the pseudo-random code structure is in a known state. This method implements full-scale jamming, from suppressive jamming to deceptive jamming, effectively improving jamming effectiveness.

[0065] Determining whether the pseudo-random code structure included in the navigation signal is known primarily depends on whether the pseudo-random code structure is publicly available. If it is publicly available, the pseudo-random code structure included in the navigation signal is known; otherwise, it is unknown. Of course, it is also possible to obtain the pseudo-random code structure through analysis within legal scope, such as for teaching and experimental purposes. This is not discussed in detail in this invention.

[0066] When the structure of the pseudo-random code is unknown, although the target receiver cannot directly lock onto the interference signal, the interference strategy can be adjusted to a forwarding interference mode, that is, the collected navigation signal is directly transformed to obtain a forwarding interference signal. This can change the system time, satellite orbit parameters, and code phase used to calculate the pseudorange recorded in the navigation signal, thereby causing the target receiver to produce an incorrect positioning based on the erroneous position information, thereby achieving the interference effect.

[0067] Optionally, the forwarded interference signal includes a directly forwarded signal; and step 41 includes:

[0068] Step 411: When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal and the modulation mode of the navigation signal are both unknown, delay processing is performed on the navigation signal to obtain a directly forwarded signal.

[0069] like Figure 2 As shown in the figure, depending on whether the modulation mode of the navigation signal is known, the forwarding interference mode can be divided into two interference strategies: direct forwarding mode and partial reconstruction forwarding mode. The method for determining whether the modulation mode is known is the same as that for pseudo-random codes, and will not be elaborated on in this invention.

[0070] Corresponding to step 411, that is, when the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal and the modulation method of the navigation signal are both in an unknown state, only the navigation signal is delayed, and the positioning error is created by changing the pseudo-range to achieve a navigation interference effect.

[0071] Optionally, the forwarded interference signal includes a partially reconstructed forwarded signal; and step 41 includes:

[0072] Step 412: When the navigation signal is not locked by the target receiver, the structure of the pseudo-random code included in the navigation signal is in an unknown state, and the modulation method of the navigation signal is in a known state, the navigation signal is processed based on the symbol domain to obtain a partially reconstructed forwarding signal.

[0073] Corresponding to step 412, when the navigation signal is not locked by the target receiver, the structure of the pseudo-random code included in the navigation signal is in an unknown state, and the modulation method of the navigation signal is in a known state, as shown in FIG. Figure 2As shown in the figure, the interference strategy is adjusted to partial reconstruction forwarding mode. The collected navigation signal is processed based on the symbol domain to generate an interference symbol set and reconstruct the time domain waveform. The resulting partially reconstructed forwarded signal changes the location information carried in the original navigation signal, resulting in a stronger interference effect than directly forwarding the signal. Of course, in partial reconstruction forwarding mode, in addition to the symbol domain processing of the signal, time delay processing can also be performed to increase the error level of the location information received by the target receiver.

[0074] Optionally, the symbol domain-based processing includes at least one of superposition processing, replacement processing and phase flip processing.

[0075] Optionally, the second interference signal includes a pseudo low-orbit satellite signal;

[0076] The step 4 comprises:

[0077] Step 421: When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal is known, the navigation signal is parsed to obtain a parsing result.

[0078] Step 422: Generate a pseudo-LOW-ORB signal based on the analysis result and the structure of the pseudo-random code; the pseudo-LOW-ORB signal is used to be locked by the target receiver when the target receiver has not locked onto the navigation signal, thereby sending error position information to the target receiver.

[0079] As you can understand, satellite signals contain a unique pseudo-random noise code (PRN). The receiver internally generates a local replica of the target satellite's PRN code. The receiver locks onto the signal by cross-correlating the received signal with its locally generated PRN code. Cross-correlation measures the degree of similarity between two signals at varying time offsets. When the local code is perfectly aligned with the pseudo-random code in the navigation signal, i.e., time-synchronized, the correlation reaches its maximum, generating a peak known as the correlation peak. The receiver searches for the correlation peak in the received signal by sliding the local code. Once a correlation peak exceeding the threshold is found, the receiver confirms successful lock onto the satellite signal, i.e., synchronization with the satellite signal is achieved, and can then continuously receive information from that satellite signal.

[0080] Based on the principle of the above receiver locking satellite signals, such as Figure 2As shown in the figure, when the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal is in a known state, the jamming strategy can be adjusted to a generative jamming mode. That is, the known pseudo-random code structure and the parsed navigation signal parameter information are used to generate an jamming signal that can be locked by the target receiver, namely a pseudo-LOW-Earth-orbit satellite signal. After the target receiver locks on to the pseudo-LOW-Earth-orbit satellite signal, the jammer can continuously send error information, causing the target receiver to continuously resolve errors, resulting in incorrect positioning, and effectively enhancing the jamming effect.

[0081] As an example, the pseudo low-orbit satellite signal can be expressed according to the following formula:

[0082] , (3)

[0083] in, Indicates the number of target satellites, The amplitude of the pseudo LEO satellite signal generated by the i-th jammer, represents the code phase of the pseudo LEO satellite signal generated by the i-th jammer.

[0084] represents the bit stream of the pseudo LEO satellite signal generated by the i-th jammer based on the parsing result of the navigation signal sent by the k-th target satellite; represents the pseudo-random code of the pseudo-LOW-Earth-orbit satellite signal generated by the i-th jammer according to the structure of the pseudo-random code in the navigation signal sent by the k-th target satellite; represents the signal frequency of the pseudo LEO satellite signal generated by the i-th jammer based on the analysis result of the navigation signal sent by the k-th target satellite; It represents the initial phase of the pseudo LEO satellite signal generated by the i-th jammer based on the analysis result of the navigation signal sent by the k-th target satellite.

[0085] Before the target receiver locks onto the pseudo-LEO satellite signal, it is necessary to ensure that the pseudo-LEO satellite signal generated by the jammer and the navigation signal have the same signal structure, modulation method, pseudo-random code, and code phase. This will ensure that when the receiver searches for correlation peaks, it will prioritize locking onto the correlation peak generated by the receiver's local code and the pseudo-LEO satellite signal, and then lock onto the pseudo-LEO satellite signal. After the pseudo-LEO satellite signal is locked onto by the target receiver, the pseudo-LEO satellite signal's code phase, carrier phase, part of the navigation message, bit stream, and other information can be gradually changed to cause the target receiver to continuously misposition.

[0086] Optionally, the number of jammers used to generate the interference signal is at least two, and the interference signal received by the target receiver is a superimposed signal of the interference signals respectively sent by the at least two jammers.

[0087] Taking the forwarding interference signal as an example, the interference signal received by the target receiver can be expressed as:

[0088] , (4)

[0089] in, Indicates the The amplitude of the repeater interference signal emitted by each jammer, Indicates navigation signal After delay The forwarded signal obtained after It represents the delay when the i-th jammer performs delay processing on the navigation signal.

[0090] Optionally, the parameter prediction result includes predicted power and predicted frequency;

[0091] The step 6 comprises:

[0092] When the interference signal is a first interference signal, adjusting the power of the first interference signal so that the adjusted power of the first interference signal is greater than the predicted power, and then sending the adjusted first interference signal to the target receiver;

[0093] When the interference signal is a second interference signal, the frequency of the second interference signal is adjusted to the predicted frequency, and then the adjusted second interference signal is sent to the target receiver.

[0094] Specifically, if Figure 2 As shown, when the interference signal is the first interference signal, that is, the interference strategy is the interference suppression mode, after generating different types of interference signals, the power of the first interference signal is adjusted so that the power of the adjusted first interference signal is greater than the predicted power, and the frequency can also be adjusted according to actual conditions such as the communication environment and equipment, and then the adjusted first interference signal is sent to the target receiver until the target receiver loses lock on the navigation signal.

[0095] When the interference signal is the second interference signal, taking the second interference signal as a pseudo-low-orbit satellite signal as an example, while adjusting the frequency of the pseudo-low-orbit satellite signal to the predicted frequency, the power of the pseudo-low-orbit satellite signal can also be adjusted so that the power of the adjusted pseudo-low-orbit satellite signal is greater than the predicted power. In this way, when the receiver slides to search for the correlation peak, it will first lock onto the correlation peak with a larger amplitude generated by the pseudo-low-orbit satellite signal, thereby improving the target receiver's locking rate on the pseudo-low-orbit satellite signal.

[0096] In addition, if Figure 2As shown, the second interference signal, including the repeater interference signal and the pseudo low-orbit satellite signal, can be adjusted based on the parameter prediction results. The present invention will not be described in detail here. In addition, the parameter prediction results include but are not limited to signal parameters such as predicted power, predicted frequency, predicted code phase, and predicted Doppler shift. After adjusting the interference signal based on the parameter prediction results and the type of interference signal, the adjusted interference signal is sent to the target receiver to cause the receiver to resolve the error and generate an incorrect positioning.

[0097] Optionally, the signal parameter prediction model is obtained by training based on signal parameter residual data; wherein the signal parameter residual data is represented as the difference between the true signal parameter and the predicted signal parameter obtained based on polynomial regression.

[0098] Reference Figure 3 , a schematic diagram of the structure of a residual block of a temporal convolutional neural network provided by an embodiment of the present invention is given. Among them, represents the output of the i-th residual block, represents the input of the i-1th residual block, is the input of the model, which represents a sequence consisting of the difference between the true signal parameters and the predicted signal parameters based on polynomial regression. represents a set of discrete sequences from 1 to T, 1 The signal parameter residual data includes but is not limited to power residual data and frequency residual data.

[0099] exist Figure 3 In the residual block structure shown, the first layer uses dilated causal convolution to expand the temporal receptive field and capture long-range dependencies through an interval sampling mechanism. While maintaining a constant convolution kernel size and network depth, increasing the dilation factor expands the input range covered by the receptive field of the top-level neurons, allowing networks with limited layers to capture a wider range of historical information. This eliminates the need to increase the convolution kernel size or network depth, effectively avoiding problems such as gradient vanishing and parameter redundancy in deep networks. The second layer further utilizes dilated causal convolution to deepen feature abstraction. Weight normalization is embedded between both layers to decompose the convolution weight vector, accelerating model convergence.

[0100] After each convolution and weight normalization layer, nonlinear activation functions are sequentially applied to introduce nonlinear representation capabilities. A channel-level regularization strategy is implemented after the activation layer: during training, entire channels of the feature map are randomly set to zero, forcing the network to build robust feature redundancy. To address the mismatch between the input and output tensor dimensions, the module introduces 1×1 convolutions to map the original input to the channel dimension, ensuring that tensor addition operations in the residual connections can be executed.

[0101] During signal parameter prediction model training, backpropagation is used to update the parameters of the time series convolutional neural network to minimize the residual prediction error. The Adam optimizer, with its momentum term and adaptive learning rate, enables rapid training convergence, reducing training time, improving generalization, and avoiding overfitting while mitigating vanishing and exploding gradients.

[0102] Specifically, using power residual data as an example, the power data of real satellite signals in a continuous time series is used as the training set. Polynomial regression is used to extract the long-term trend of the sequence, and the residual is used as the short-term fluctuation feature to input into the time series convolutional neural network.

[0103] Get a small window of data from the training set and define it according to the following formula:

[0104] , (5)

[0105] in, Indicates the starting position of the window. is the window length, Indicates that the time series By constructing features and target values ​​through sliding windows, a large number of training samples are generated from long sequences, increasing data diversity and reducing the risk of overfitting. This also facilitates batch parallel computing and improves computational efficiency.

[0106] Using window data, fitting the n-order function, the polynomial regression function is expressed as:

[0107] , (6)

[0108] , (7)

[0109] in, represents the prediction power of the time series at time t, The coefficient representing the nth power; represents a Vandermonde matrix, where each row corresponds to a time point in the sliding window, and each column corresponds to the power of the polynomial basis function. Trend fitting is a preprocessing of the data that can reduce data complexity, reduce the model burden, and improve the model's robustness to outliers.

[0110] Based on formulas (5)-(7), the power residual is obtained by subtracting the predicted value from the true value using the polynomial regression function:

[0111] , (8)

[0112] in, Represents the next moment true value at the end of the window, Indicates the power prediction value at the next moment at the end of the window. It is based on a set of discrete power residual data from 1 to T.

[0113] Optionally, after step 6, the method further includes:

[0114] The interference bit error rate is calculated based on the following formula to measure the reliability of the method described in the embodiment of the present invention:

[0115] , (9)

[0116] in, represents the interference bit error rate, represents the bit error rate function; Indicates the amplitude of the real signal; represents the noise power spectral density; Indicates the amplitude of the interference signal;

[0117] The root mean square error and mean absolute percentage error were calculated based on the following formulas to quantify the performance of the signal parameter prediction model:

[0118] , (10)

[0119] , (11)

[0120] Among them, MAPE stands for mean absolute percentage error, RMSE stands for root mean square error, Indicates time The true value of the navigation signal, Indicates time The predicted value of the navigation signal.

[0121] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0122] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and purpose of the present invention.

Claims

1. An adaptive navigation interference optimization method based on temporal convolutional neural network, characterized in that: The method comprises: Step 1: collecting the navigation signal transmitted by the target satellite to the target receiver in real time, and performing feature analysis to obtain feature information of the navigation signal; Step 2, determining whether the navigation signal is locked by the target receiver; Step 3: When the navigation signal is locked by the target receiver, generating a first interference signal until the target receiver loses the lock on the navigation signal; wherein the first interference signal is used to establish a navigation signal denial environment; Step 4: When the navigation signal is not locked by the target receiver, generating a second interference signal; wherein the second interference signal is used to cause the target receiver to generate an erroneous positioning; Step 5: Inputting the characteristic information into a preset signal parameter prediction model to predict the signal parameters of the navigation signal to obtain a parameter prediction result of the navigation signal; wherein the signal parameter prediction model is built based on a time domain convolutional neural network; Step 6: Adjust the signal parameters of the first interference signal or the second interference signal based on the parameter prediction result, and then send the adjusted first interference signal or the second interference signal to the target receiver.

2. The method according to claim 1, characterized in that The second interference signal includes a forwarding interference signal, and step 4 includes: Step 41: When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal is in an unknown state, performing signal conversion processing on the navigation signal to obtain a forwarding interference signal.

3. The method according to claim 2, characterized in that The forwarding interference signal includes a directly forwarded signal; and step 41 includes: When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal and the modulation mode of the navigation signal are both in an unknown state, delay processing is performed on the navigation signal to obtain a directly forwarded signal.

4. The method according to claim 2, characterized in that The forwarding interference signal includes a partially reconstructed forwarding signal; the step 41 includes: When the navigation signal is not locked by the target receiver, the structure of the pseudo-random code included in the navigation signal is in an unknown state, and the modulation method of the navigation signal is in a known state, the navigation signal is processed based on the symbol domain to obtain a partially reconstructed forwarding signal.

5. The method according to claim 4, characterized in that The symbol domain-based processing includes at least one of a superposition process, a replacement process, and a phase flip process.

6. The method according to claim 1, characterized in that The second interference signal includes a pseudo low-orbit satellite signal; The step 4 comprises: When the navigation signal is not locked by the target receiver and the structure of the pseudo-random code included in the navigation signal is in a known state, parsing the navigation signal to obtain a parsing result; A pseudo low-orbit satellite signal is generated according to the analysis result and the structure of the pseudo-random code; the pseudo low-orbit satellite signal is used to be locked by the target receiver when the target receiver has not locked the navigation signal, and then send error position information to the target receiver.

7. The method according to claim 1, characterized in that The type of the first interference signal includes at least one of narrowband interference, broadband noise interference and swept frequency interference.

8. The method according to claim 1, characterized in that The number of jammers used to generate the interference signal is at least two, and the interference signal received by the target receiver is a superimposed signal of the interference signals respectively sent by the at least two jammers.

9. The method according to claim 1, characterized in that The parameter prediction results include predicted power and predicted frequency; The step 6 comprises: When the interference signal is a first interference signal, adjusting the power of the first interference signal so that the adjusted power of the first interference signal is greater than the predicted power, and then sending the adjusted first interference signal to the target receiver; When the interference signal is a second interference signal, the frequency of the second interference signal is adjusted to the predicted frequency, and then the adjusted second interference signal is sent to the target receiver.

10. The method according to claim 1, characterized in that The signal parameter prediction model is obtained by training based on signal parameter residual data; wherein, the signal parameter residual data is represented as the difference between the true signal parameter and the predicted signal parameter obtained based on polynomial regression.

Images