GNSS (Global Navigation Satellite System) suppressing interference number identification method, device, equipment and medium

Abstract

The invention relates to a GNSS (Global Navigation Satellite System) suppressing interference number identification method, device and equipment and a medium, and relates to the technical field of satellite navigation interference resistance. The method comprises the following steps: receiving a spatial domain signal through an array antenna, obtaining an array baseband signal through down-conversion and analog-to-digital conversion, constructing a spatial domain correlation matrix in a complex matrix form based on signal vectors at a plurality of snapshot moments and conjugate transpose operation, extracting all elements of an upper triangular part of the matrix, arranging the elements according to a preset sequence to form a complex vector as input data, and outputting the input data. And outputting a classification result by using a pre-trained deep learning model, and determining the number of suppressing disturbances. According to the method, data simplification, complete information reservation and adaptive model input are realized by means of complex matrix characteristics, subjective threshold is not needed through deep learning, the limitation of a traditional method is solved, high robustness and accuracy are still achieved under the environment of low signal to interference plus noise ratio and small snapshot number, offline training and real-time reasoning are achieved, the calculation complexity is reduced, the real-time requirement of a terminal is met, and the method is suitable for popularization and application. And reliable support is provided for an anti-interference algorithm.

Description

Technical Field

[0001] This application relates to the field of satellite navigation anti-interference technology, and in particular to a method, device, equipment and medium for identifying the number of GNSS suppression interferences. Background Technology

[0002] GNSS (such as GPS, BeiDou, GLONASS, Galileo, etc.) plays a crucial role in providing precise positioning, navigation, and timing services to users worldwide. However, because GNSS signals are extremely weak when they reach the ground from space, they are highly susceptible to malicious interference, especially powerful suppression jamming. Suppression jamming can cover or drown out the weak useful GNSS signals, preventing receivers from properly acquiring and tracking satellite signals, thus causing positioning failure.

[0003] To combat interference suppression, spatial anti-jamming techniques based on array antennas have become the mainstream solution. This technique suppresses interference by adjusting the weight vector of the array antenna to create nulls in the direction of incoming interference waves. However, most advanced anti-jamming algorithms (such as those based on eigenspaces) require prior knowledge of the exact number of interference signals. Incorrect estimation of the interference number can severely degrade interference suppression performance and may even lead to the misinterpretation and suppression of useful signals as interference.

[0004] Currently, interference number estimation methods are mainly based on information theory criteria, such as the Akaike Information Criterion (AIC) and the Minimum Description Length (MDL) criterion. These methods determine the number of interferences by analyzing the eigenvalues ​​of the received signal covariance matrix. However, these traditional methods have obvious drawbacks: (1) They are sensitive to the signal-to-interference-plus-noise ratio (SINR), meaning that in low SINR environments, the eigenvalue distribution is ambiguous, and the estimation accuracy of traditional methods drops sharply. (2) They require subjective threshold setting: Some improved algorithms require threshold values ​​to be set based on experience, resulting in poor adaptability and unstable performance in different scenarios. (3) They have high computational complexity: They require complex operations such as eigenvalue decomposition, and for GNSS terminals with limited computing resources, real-time performance is difficult to guarantee. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, device, equipment, and medium for identifying the number of GNSS suppression interferences that can adapt to complex electromagnetic environments, have high estimation accuracy, and good real-time performance, in order to address the above-mentioned technical problems.

[0006] A method for identifying the number of GNSS suppression interferences, the method comprising: The radio frequency (RF) signal is obtained by receiving spatial signals from a GNSS array antenna. The RF signal is then subjected to down-conversion and analog-to-digital conversion processes to obtain the array baseband signal. Based on the array baseband signal, calculate the spatial correlation matrix, which is a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose. The spatial correlation matrix is ​​converted into a format suitable for deep learning model input through preprocessing to generate input data. The preprocessing of the spatial correlation matrix includes extracting all elements of the upper triangular part or the lower triangular part of the spatial correlation matrix, arranging all the extracted elements in a preset order to form a complex vector, and obtaining the input data. The input data is fed into a pre-trained deep learning model to obtain a classification result corresponding to the number of interferences. Based on the classification result, the number of suppression interferences present in the current environment is determined.

[0007] In one embodiment, the GNSS array antenna is an M-element uniform linear array, where M is a positive integer greater than 1; The analog-to-digital converter is used for ADC sampling, and the resulting array baseband signal is a digital baseband signal.

[0008] In one embodiment, the formula for calculating the spatial correlation matrix is ​​expressed as follows:

[0009] In the above formula, This represents the spatial correlation matrix. Indicates the first The array received signal vector at each snapshot moment For the number of snapshots, Indicates to Perform the conjugate transpose operation.

[0010] In one embodiment, the step of extracting all elements from the upper or lower triangular portion of the spatial correlation matrix, arranging the extracted elements in a preset order to form a complex vector, and obtaining the input data includes: Extract all elements from the triangular portion of the main diagonal of the spatial correlation matrix; Arrange all extracted elements into a complex vector in a row-major order according to a preset order.

[0011] In one embodiment, the deep learning model is one of a convolutional neural network, a fully connected neural network, a recurrent neural network, or a Transformer architecture.

[0012] In one embodiment, when the deep learning model is a fully connected neural network, the fully connected neural network includes an input layer, at least one hidden layer, and an output layer. The hidden layer uses the ReLU activation function, the output layer uses the Softmax activation function, and the number of neurons in the output layer is [number missing]. ,in, This indicates the maximum preset number of interferences, and includes scenarios with 0 interferences.

[0013] In one embodiment, training the deep learning model includes: Multiple array received signal samples under different interference scenarios are generated through simulation or actual measurement, and each signal sample is labeled with the actual number of interferences. The interference scenarios include combinations of different interference numbers, interference direction of arrival, interference-to-signal ratio (JSR), and GNSS useful signal direction of arrival. For each of the signal samples, the spatial correlation matrix is ​​calculated and preprocessed to obtain the corresponding training data; The training data is used as the input to the deep learning model, and the corresponding number of disturbances is used as the expected output. The deep learning model is trained in a supervised manner until the model converges to obtain the pre-trained deep learning model.

[0014] This application also provides a GNSS suppression interference number identification device, the device comprising: The radio frequency signal receiving and processing module is used to acquire radio frequency signals, which are obtained by receiving spatial signals from a GNSS array antenna. The radio frequency signals are then subjected to down-conversion and analog-to-digital conversion processes to obtain array baseband signals. The spatial correlation matrix calculation module is used to calculate the spatial correlation matrix based on the array baseband signal. The spatial correlation matrix is ​​a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose vector. The preprocessing module is used to convert the spatial correlation matrix into a format suitable for input to a deep learning model through preprocessing, and generate input data. The preprocessing of the spatial correlation matrix includes extracting all elements of the upper triangular part or the lower triangular part of the spatial correlation matrix, arranging all the extracted elements in a preset order to form a complex vector, and obtaining the input data. The suppression interference number determination module is used to input the input data into a pre-trained deep learning model to obtain a classification result corresponding to the number of interferences, and determine the number of suppression interferences existing in the current environment based on the classification result.

[0015] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in the above-described GNSS suppression interference number identification method.

[0016] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described GNSS suppression interference number identification method.

[0017] The aforementioned GNSS suppression interference number identification method, device, equipment, and medium involve sequentially down-converting and analog-to-digital converting the radio frequency signal received by the GNSS array antenna in the spatial domain to obtain the array baseband signal. Based on the array baseband signal, a spatial correlation matrix is ​​calculated. The spatial correlation matrix is ​​a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose vector. Then, all elements of the upper or lower triangular part of the spatial correlation matrix are extracted and arranged in a preset order to form a complex vector, resulting in input data suitable for the input format of a deep learning model. A pre-trained deep learning model is used to obtain a classification result corresponding to the number of interferences based on the input data, and the number of suppression interferences present in the current environment is determined based on the classification result.

[0018] This method, on the one hand, utilizes the conjugate symmetry of Hermite complex matrices to extract partial elements to form input data. This simplifies the data while fully preserving key signal information, accurately adapting to the input requirements of deep learning models and avoiding the impact of redundant data on model inference efficiency. On the other hand, it analyzes the input data through a deep learning model, eliminating the need to rely on subjective experience to set threshold values. This effectively overcomes the limitations of traditional methods that rely on eigenvalue analysis. Even in complex electromagnetic environments with low signal-to-interference-plus-noise ratios and small snapshot counts, it can still stably and accurately identify the number of interferences, significantly improving the robustness and accuracy of the identification results. Furthermore, model training is completed offline, and after deployment, only forward inference is needed to quickly output classification results, avoiding complex real-time calculations and significantly reducing computational complexity. This meets the stringent real-time processing requirements of GNSS terminals, providing reliable support for the efficient execution of subsequent anti-interference algorithms and ensuring that the GNSS system stably provides positioning, navigation, and timing services in complex interference environments. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a method for identifying the number of GNSS suppression interference instances in one embodiment. Figure 2 This is a schematic diagram of the structure of a deep learning model in one embodiment; Figure 3This is a flowchart illustrating a deep learning model training method in one embodiment; Figure 4 This is a structural block diagram of a GNSS suppression interference number identification device in one embodiment; Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0021] To address the shortcomings of existing GNSS interference number estimation methods based on information theory criteria, such as sensitivity to signal-to-interference-plus-noise ratio (SINR), sharp decline in accuracy under low SINR or small snapshot count scenarios, poor adaptability due to reliance on subjective experience in setting thresholds, high computational complexity due to complex operations like eigenvalue decomposition, difficulty in meeting the real-time requirements of GNSS terminals, and the failure of some deep learning-based solutions to fully utilize the inherent characteristics of the spatial correlation matrix of GNSS array signals, resulting in data redundancy and insufficient adaptability in preprocessing methods, thus affecting model recognition accuracy and terminal deployment efficiency, this application, as shown in Figure 1, provides a GNSS suppression interference number identification method, including the following steps: Step 100: Acquire the radio frequency signal. The radio frequency signal is obtained by receiving spatial signals from the GNSS array antenna. The radio frequency signal is then subjected to down-conversion and analog-to-digital conversion processes to obtain the array baseband signal.

[0022] Step 110: Calculate the spatial correlation matrix based on the array baseband signal. The spatial correlation matrix is ​​a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose.

[0023] Step 120: Convert the spatial correlation matrix into a format suitable for deep learning model input through preprocessing to generate input data. The preprocessing of the spatial correlation matrix includes extracting all elements of the upper triangular or lower triangular part of the spatial correlation matrix, arranging all extracted elements in a preset order to form a complex vector, and obtaining the input data.

[0024] Step 130: Input the input data into the pre-trained deep learning model to obtain the classification result corresponding to the number of interferences. Based on the classification result, determine the number of suppression interferences present in the current environment.

[0025] In this method, spatial signals are received via a GNSS array antenna and down-converted and analog-to-digital converted to obtain the array baseband signal. A Hermite complex matrix spatial correlation matrix is ​​constructed based on the signal vectors from multiple snapshots. Then, all elements of the upper or lower triangular portions of the matrix are selectively extracted and arranged into complex vectors. This method not only fully preserves key signal information but also achieves reasonable data simplification, accurately adapting to the input requirements of deep learning models. Combining this with a pre-trained deep learning model for classification inference eliminates the need for subjective threshold settings, effectively overcoming the performance bottlenecks of traditional methods. It maintains high robustness and recognition accuracy even in complex electromagnetic environments. Furthermore, the offline training and real-time inference mode significantly reduces computational complexity, perfectly adapting to the resource constraints and real-time processing requirements of GNSS terminals.

[0026] In step S100, signals in the airspace are received through a GNSS array antenna. These signals include useful GNSS signals, positioning, navigation, and timing signals from satellite systems such as GPS, BeiDou, GLONASS, and Galileo, as well as at least one possible suppression interference signal, such as an interference signal maliciously emitted by an individual with significantly greater power than the useful GNSS signal, forming a radio frequency hybrid signal that covers both the target signal and the interference signal.

[0027] Because the useful GNSS signal is weak after being transmitted from the satellite to the ground, it is easily suppressed and covered or drowned out by interference signals. Therefore, targeted signal conditioning processing is required for the received radio frequency mixed signal: First, the carrier frequency of the radio frequency signal is reduced to the baseband frequency range by downconversion to eliminate the influence of the high-frequency carrier on subsequent signal processing, resulting in an analog baseband signal. Then, an analog-to-digital converter (ADC) module is used to sample and quantize the analog baseband signal, converting the continuous analog signal into a discrete digital baseband signal. That is, the array baseband signal, where, Number the array elements. For the snapshot index.

[0028] In this embodiment, the GNSS array antenna is preferably an M-element uniform linear array, where M is a positive integer greater than 1. Each element synchronously receives spatial signals and completes signal conversion. The final output array baseband signal is in the form of an M-dimensional column vector, the dimension of which is consistent with the number of elements. Each dimension corresponds to the digital baseband signal of one element, providing a complete and synchronous signal foundation for the subsequent calculation of the spatial correlation matrix.

[0029] In step S110, the signals from the nth snapshot are combined into a vector. Estimate the spatial correlation matrix using N snapshots. , is represented as:

[0030] In the above formula, Represents the spatial correlation matrix. Indicates the first The array received signal vector at each snapshot moment For the number of snapshots, Indicates to Perform the conjugate transpose operation.

[0031] Specifically, It is A 3D Hermite complex matrix is ​​represented as:

[0032] In the above formula, It is a Hermite matrix, that is or equivalent ,in This indicates a conjugate operation, in which all information is fully contained within the elements of the upper triangular portion (including the main diagonal).

[0033] In step S120, all elements of the upper or lower triangular part of the spatial correlation matrix are extracted, and the extracted elements are arranged in a preset order to form a complex vector to obtain input data. This includes: extracting all elements of the upper triangular part of the spatial correlation matrix containing the main diagonal, and arranging the extracted elements in a preset row-major order to form a complex vector.

[0034] In this embodiment, due to It is a conjugate symmetric matrix, and all its information is contained in the upper triangular part, including the main diagonal. These elements are extracted and rearranged into a complex vector in row-major order. , is represented as:

[0035] Because, vector The length L is the total number of elements in the upper triangular part (including the main diagonal), i.e. .therefore, ,That The complex field is represented by . Finally, this complex vector is... As input to deep learning models.

[0036] In step S130, the deep learning model is one of the following: convolutional neural network, fully connected neural network, recurrent neural network, or Transformer architecture.

[0037] In this embodiment, the preprocessed vector is input into a pre-trained deep learning model. In this embodiment, the model is a fully connected neural network with three hidden layers, such as... Figure 2 As shown. A fully connected neural network includes an input layer, at least one hidden layer, and an output layer. The hidden layer uses the ReLU activation function, and the output layer uses the Softmax activation function. The number of neurons in the output layer is... ,in, This indicates the maximum preset number of interferences, and includes scenarios with 0 interferences.

[0038] Specifically, the input to the input layer is a preprocessed complex vector. , dimension , where M is the number of array elements. Each node in the input layer corresponds to a complex element in a complex vector.

[0039] Furthermore, each hidden layer contains a number of neurons, the specific number of which can be a hyperparameter, such as 128, 64, or 32. All neurons between layers are fully connected. Each hidden layer neuron performs a weighted sum of all its inputs and then applies ReLU (Rectified Linear Unit) as a non-linear activation function, the mathematical expression of which is: The introduction of the ReLU function gives the model the ability to learn nonlinear mapping relationships, while also helping to alleviate the gradient vanishing problem during training and accelerating convergence.

[0040] Furthermore, the number of neurons in the output layer corresponds to the number of classification categories of the interference, i.e. (From 0 interferences to) (Number of disturbances). The output layer uses the Softmax activation function to convert the output of the hidden layer into a probability distribution vector, where the output value of each node represents the probability that the input data belongs to the corresponding number of disturbance categories.

[0041] In this embodiment, the fully connected neural network operates in a forward propagation manner: it transforms complex feature vectors... The data is input from the input layer, and features are extracted step by step through three hidden layers using nonlinear transformations (ReLU activation). Finally, the output layer (Softmax activation) generates the probability distribution. ,in Indicates the number of interferences. The probability of the given time is used to identify the category with the highest probability.

[0042] In this embodiment, when determining the number of interferences, the category corresponding to the node with the highest probability in the output layer is taken as the number of identified interferences. For example, if the probability of the third node in the output layer is the highest, then it is determined that there are currently two interferences (assuming that the categories are numbered starting from 0).

[0043] like Figure 3 As shown, this method also provides a method for training a deep learning model, including: generating multiple array-received signal samples under different interference scenarios through simulation or actual measurement, with each signal sample labeled with a real interference number label. The interference scenarios include combinations of different interference numbers, interference direction of arrival, interference-to-signal ratio (JSR), and GNSS useful signal direction of arrival. Then, for each signal sample, the spatial correlation matrix is ​​calculated and preprocessed to obtain corresponding training data. Finally, the training data is used as the input to the deep learning model, and the corresponding interference number label is used as the desired output. Supervised training is employed to train the deep learning model until the model converges, resulting in a pre-trained deep learning model.

[0044] In this embodiment, computer simulation is first used to simulate different numbers of suppression interference (0, 1, ...). For each scenario, different parameters such as interference direction of arrival, interference-to-signal ratio (JSR), and signal direction of arrival are set to generate a large number (e.g., 100,000 groups) of array received signal samples, and each sample is labeled with the actual number of interferences. Then, the spatial correlation matrix of each sample signal is calculated, and preprocessed according to the method in step S120 to obtain the input feature vector.

[0045] Furthermore, a DNN (fully connected neural network) model is constructed by inputting the training dataset (feature vectors and labels) into the model. The model is trained using the cross-entropy loss function and the Adam optimizer, and the model parameters are continuously adjusted through backpropagation until the loss function converges.

[0046] The aforementioned GNSS suppression interference number identification method utilizes the conjugate symmetry of Hermite complex matrices by selectively extracting upper or lower triangular elements from the spatial correlation matrix and arranging them into complex vectors. This approach achieves both reasonable simplification of data dimensions and complete preservation of effective information, while also accurately adapting to the input requirements of deep learning models. Combined with the powerful nonlinear feature extraction and pattern recognition capabilities of deep learning models, it eliminates the need for subjective threshold settings and effectively overcomes the technical bottleneck of traditional information theory-based methods where accuracy drops sharply in complex electromagnetic environments such as low signal-to-interference-plus-noise ratio (SINR) and small snapshot counts. This significantly improves the robustness and accuracy of interference number identification. Meanwhile, the model training process is completed offline, and after deployment, classification results can be quickly output through forward inference alone. This avoids complex operations such as feature value decomposition in traditional methods, significantly reducing computational complexity and meeting the stringent real-time processing requirements of GNSS terminals. Furthermore, through sample training covering various interference scenarios, the model has excellent generalization ability and can flexibly cope with unknown interference scenarios with different interference wave directions and interference-to-signal ratios. This provides reliable interference data support for the accurate execution of subsequent array antenna anti-interference algorithms, ensuring stable positioning, navigation, and timing services for GNSS receivers in complex interference environments.

[0047] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0048] In one embodiment, such as Figure 4 As shown, a GNSS jamming number identification device is provided, comprising: a radio frequency signal receiving and processing module 200, a spatial correlation matrix calculation module 210, a preprocessing module 220, and a jamming number determination module 230, wherein: The radio frequency signal receiving and processing module 200 is used to acquire radio frequency signals, which are obtained by receiving spatial signals from a GNSS array antenna. The radio frequency signals are then subjected to down-conversion and analog-to-digital conversion processes to obtain array baseband signals. The spatial correlation matrix calculation module 210 is used to calculate the spatial correlation matrix based on the array baseband signal. The spatial correlation matrix is ​​a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose vector. The preprocessing module 220 is used to convert the spatial correlation matrix into a format suitable for deep learning model input through preprocessing to generate input data. The preprocessing of the spatial correlation matrix includes extracting all elements of the upper triangular part or the lower triangular part of the spatial correlation matrix, arranging all the extracted elements in a preset order to form a complex vector, and obtaining the input data. The suppression interference number determination module 230 is used to input the input data into a pre-trained deep learning model to obtain a classification result corresponding to the number of interferences, and determine the number of suppression interferences existing in the current environment based on the classification result.

[0049] Specific limitations regarding the GNSS suppression and interference number identification device can be found in the limitations of the GNSS suppression and interference number identification method described above, and will not be repeated here. Each module in the aforementioned GNSS suppression and interference number identification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0050] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for identifying the number of GNSS suppression jamming devices. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0051] Those skilled in the art will understand that Figure 5The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0052] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps: The radio frequency (RF) signal is obtained by receiving spatial signals from a GNSS array antenna. The RF signal is then subjected to down-conversion and analog-to-digital conversion processes to obtain the array baseband signal. Based on the array baseband signal, calculate the spatial correlation matrix, which is a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose. The spatial correlation matrix is ​​converted into a format suitable for deep learning model input through preprocessing to generate input data. The preprocessing of the spatial correlation matrix includes extracting all elements of the upper triangular part or the lower triangular part of the spatial correlation matrix, arranging all the extracted elements in a preset order to form a complex vector, and obtaining the input data. The input data is fed into a pre-trained deep learning model to obtain a classification result corresponding to the number of interferences. Based on the classification result, the number of suppression interferences present in the current environment is determined.

[0053] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: The radio frequency (RF) signal is obtained by receiving spatial signals from a GNSS array antenna. The RF signal is then subjected to down-conversion and analog-to-digital conversion processes to obtain the array baseband signal. Based on the array baseband signal, calculate the spatial correlation matrix, which is a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose. The spatial correlation matrix is ​​converted into a format suitable for deep learning model input through preprocessing to generate input data. The preprocessing of the spatial correlation matrix includes extracting all elements of the upper triangular part or the lower triangular part of the spatial correlation matrix, arranging all the extracted elements in a preset order to form a complex vector, and obtaining the input data. The input data is fed into a pre-trained deep learning model to obtain a classification result corresponding to the number of interferences. Based on the classification result, the number of suppression interferences present in the current environment is determined.

[0054] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0055] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0056] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for identifying the number of GNSS suppression interferences, characterized in that, The method includes: The radio frequency (RF) signal is obtained by receiving spatial signals from a GNSS array antenna. The RF signal is then subjected to down-conversion and analog-to-digital conversion processes to obtain the array baseband signal. Based on the array baseband signal, calculate the spatial correlation matrix, which is a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose. The spatial correlation matrix is ​​converted into a format suitable for deep learning model input through preprocessing to generate input data. The preprocessing of the spatial correlation matrix includes extracting all elements of the upper triangular part or the lower triangular part of the spatial correlation matrix, arranging all the extracted elements in a preset order to form a complex vector, and obtaining the input data. The input data is fed into a pre-trained deep learning model to obtain a classification result corresponding to the number of interferences. Based on the classification result, the number of suppression interferences present in the current environment is determined.

2. The GNSS suppression interference number identification method according to claim 1, characterized in that, The GNSS array antenna is an M-element uniform linear array, where M is a positive integer greater than 1; The analog-to-digital converter is used for ADC sampling, and the resulting array baseband signal is a digital baseband signal.

3. The GNSS suppression interference number identification method according to claim 1, characterized in that, The formula for calculating the spatial correlation matrix is ​​as follows: In the above formula, This represents the spatial correlation matrix. Indicates the first The array received signal vector at each snapshot moment For the number of snapshots, Indicates to Perform the conjugate transpose operation.

4. The GNSS suppression interference number identification method according to claim 1, characterized in that, The step of extracting all elements from the upper or lower triangular portion of the spatial correlation matrix, arranging all extracted elements in a preset order to form a complex vector, and obtaining the input data includes: Extract all elements from the triangular portion of the main diagonal of the spatial correlation matrix; Arrange all extracted elements into a complex vector in a row-major order according to a preset order.

5. The GNSS suppression interference number identification method according to claim 1, characterized in that, The deep learning model is one of the following: convolutional neural network, fully connected neural network, recurrent neural network, or Transformer architecture.

6. The GNSS suppression interference number identification method according to claim 5, characterized in that, When the deep learning model is a fully connected neural network, the fully connected neural network includes an input layer, at least one hidden layer, and an output layer. The hidden layer uses the ReLU activation function, the output layer uses the Softmax activation function, and the number of neurons in the output layer is [number missing]. ,in, This indicates the maximum preset number of interferences, and includes scenarios with 0 interferences.

7. The GNSS suppression interference number identification method according to claim 5, characterized in that, Training the deep learning model includes: Multiple array received signal samples under different interference scenarios are generated through simulation or actual measurement, and each signal sample is labeled with the actual number of interferences. The interference scenarios include combinations of different interference numbers, interference direction of arrival, interference-to-signal ratio (JSR), and GNSS useful signal direction of arrival. For each of the signal samples, the spatial correlation matrix is ​​calculated and preprocessed to obtain the corresponding training data; The training data is used as the input to the deep learning model, and the corresponding number of disturbances is used as the expected output. The deep learning model is trained in a supervised manner until the model converges to obtain the pre-trained deep learning model.

8. A GNSS suppression interference number identification device, characterized in that, The device includes: The radio frequency signal receiving and processing module is used to acquire radio frequency signals, which are obtained by receiving spatial signals from a GNSS array antenna. The radio frequency signals are then subjected to down-conversion and analog-to-digital conversion processes to obtain array baseband signals. The spatial correlation matrix calculation module is used to calculate the spatial correlation matrix based on the array baseband signal. The spatial correlation matrix is ​​a Hermite complex matrix constructed by multiplying the array received signal vector at multiple snapshot times with its conjugate transpose vector. The preprocessing module is used to convert the spatial correlation matrix into a format suitable for input to a deep learning model through preprocessing, and generate input data. The preprocessing of the spatial correlation matrix includes extracting all elements of the upper triangular part or the lower triangular part of the spatial correlation matrix, arranging all the extracted elements in a preset order to form a complex vector, and obtaining the input data. The suppression interference number determination module is used to input the input data into a pre-trained deep learning model to obtain a classification result corresponding to the number of interferences, and determine the number of suppression interferences existing in the current environment based on the classification result.

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

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

Images