A method for identifying multiple interference sources of a distributed radar, an electronic device, and a storage medium

Abstract

The present disclosure provides a method for identifying multiple interference sources of a distributed radar, an electronic device, and a storage medium, including: first, constructing the signals received by the distributed radar into a graph structure, where nodes represent signals and edges indicate whether the nodes are affected by the same type of interference. Then, the real and imaginary parts of the signals are processed separately by a dual-channel convolutional neural network, and each channel includes two convolutional layers, a pooling layer, and a fully connected layer to extract signal features. Next, these features are input into a graph self-attention network for interference type classification. Finally, the signal features with classification information enter a decoupled graph convolutional network to decompose the node features into multiple independent subspaces, identify the nodes affected by the same interference source, and construct a predicted adjacency matrix to distinguish different interference sources. The present disclosure can improve the recognition accuracy of interference types in a complex electromagnetic environment, and at the same time enhance the efficiency and accuracy of the model in processing large-scale data.

Description

Technical Field

[0001] The present invention relates to the technical field of radar interference source identification, and particularly relates to a distributed radar multi-interference source identification method, an electronic device, and a storage medium. Background Art

[0002] Traditional radar interference source identification methods rely on manually extracted features, which often fail to comprehensively reflect the essential features of interference signals, resulting in low identification accuracy. Secondly, with the development of technology, radar systems face a more complex interference environment, especially when dealing with multiple jammers working in coordination and new interference methods based on digital technology. The coordinated operation of multiple jammers increases the probability of interference overlap in different domains, making classification and identification more difficult, and existing methods are difficult to handle complex interference scenarios. In addition, the differentiation of radar interference sources usually requires parameter estimation of interference, while existing interference parameter estimation methods are often limited to specific types of interference and are difficult to accurately estimate the interference parameters of each component in complex interference. Summary of the Invention

[0003] In a first aspect of the present disclosure, a distributed radar multi-interference source identification method is provided, including the following steps:

[0004] Construct a graph structure with the signals received by the distributed radar as nodes, where the edges connecting the nodes represent that the corresponding connected nodes are affected by the same type of interference;

[0005] Decompose the real and imaginary parts of the signal, extract the signal features of the real part signal through the first channel, and extract the signal features of the imaginary part signal through the second channel, where both the first channel and the second channel include a convolutional neural network structure, which sequentially includes a first convolutional layer, a second convolutional layer, a pooling layer, and a fully connected layer;

[0006] Take the extracted signal features of the real part and the imaginary part as nodes and input them into a graph self-attention network, classify the interference types in the signal through the signal features, each signal feature corresponds to one or more feature vectors, and the feature vectors represent the attributes of the corresponding nodes;

[0007] Take the signals with interference type classification information as nodes and input them into a decoupled graph convolutional network, identify whether the nodes are affected by the same interference source by decomposing the node features into multiple independent subspaces, and then construct a prediction adjacency matrix to distinguish different interference sources.

[0008] In combination with the first aspect, the graph structure is , where is a set of radar nodes, is the node feature matrix, where each row corresponds to the feature vector of a node.

[0009] Combined with the first aspect, the real part signal includes the in-phase baseband signal of the distributed radar, and the imaginary part signal includes the quadrature baseband signal of the distributed radar.

[0010] Combined with the first aspect, the first convolutional layer and the second convolutional layer output multiple channels, and the number of output channels of the first convolutional layer is less than that of the second convolutional layer. All convolutional layers adopt the SAME boundary mode. The output of the second convolutional layer is sent to the pooling layer to compress the feature map and average the values of each channel.

[0011] Combined with the first aspect, the feature representation output by the fully connected layer is:

[0012] ,

[0013] where, is the feature representation of all nodes, is the th feature vector of the node, represents the total number of nodes, is the dimension of each feature vector, represents the dimension of the original feature.

[0014] Combined with the first aspect, the classification of the interference type in the signal by the signal feature includes:

[0015] For each pair of connected nodes, calculate an attention coefficient;

[0016] Normalize the nodes for which the attention coefficient is calculated;

[0017] Update the feature representation of each node according to the normalized attention coefficient.

[0018] Combined with the first aspect, the calculation of the attention coefficient uses the following formula:

[0019] ,

[0020] where, represents the degree of attention of node to node , is the activation function, is the parameter vector, is the weight matrix used to linearly transform the node feature vector, and respectively represent the original feature vectors of node and node .

[0021] Combined with the first aspect, the normalization processing of the nodes for calculating the attention coefficients uses the following formula:

[0022] ,

[0023] where, is the normalized attention coefficient of node to node . represents the set of all neighbor nodes of the node, is the subspace.

[0024] Combined with the first aspect, the updating of the feature representation of each node uses the following formula:

[0025] ,

[0026] where, is the updated feature vector, is the non-linear activation function.

[0027] Combined with the first aspect, the method further includes splicing the updated output features using multi-head attention after updating the feature representation of each node:

[0028] .

[0029] Combined with the first aspect, the identifying whether nodes are affected by the same interference source by decomposing node features into multiple independent subspaces, and then constructing a prediction adjacency matrix to distinguish different interference sources includes:

[0030] Projecting the feature vector of each node into different subspaces, and initializing the embedding vectors of channels:

[0031] ,

[0032] where and are the parameters of the th subspace, is the low-dimensional projection of the th subspace;

[0033] For each node, updating its decoupled representation using its neighborhood information, and the calculation formula is:

[0034] ,

[0035] ,

[0036] ,

[0037] ,

[0038] wherein, is the number of iterations, represents the connection probability caused by the -th factor between nodes and , and represents the importance of aggregating node relative to node in the -th subspace, satisfying , and for all , there are and , and are trainable trade-off coefficients, which will be adjusted during the training process, and the adjustment range is within ;

[0039] Judge the interference type of each node. If it is judged that a certain node does not have a certain type of interference, the corresponding subspace output will be set to 0, otherwise it remains unchanged;

[0040] Use the formula to construct the predicted adjacency matrix, is a matrix of , where is the number of nodes, is the feature dimension of each node, is transpose matrix, is the activation function.

[0041] In a second aspect of the present disclosure, there is provided an electronic device, including:

[0042] One or more processors;

[0043] A storage unit for storing one or more programs, which when executed by the one or more processors, can enable the one or more processors to implement the distributed radar multi-interference source identification method.

[0044] In a third aspect of the present disclosure, there is provided a computer-readable storage medium, on which a computer program is stored, characterized in that the computer program can implement the distributed radar multi-interference source identification method when executed by a processor.

[0045] Beneficial effects: A method for identifying multiple interference sources of distributed radars provided by the present disclosure constructs a distributed radar interference perception network model, comprehensively analyzes data from multiple radar nodes, realizes accurate identification of different types of interference sources, and is conducive to the effective acquisition of parameter information in subsequent tasks. This method utilizes the collaboration between each radar station, enhances the anti-interference ability of the system in a complex electromagnetic environment and the accuracy of interference type identification, thereby greatly improving the accuracy of multiple interference source identification and the performance of subsequent radar cooperative anti-interference. Description of the Drawings

[0046] Figure 1 It is a schematic flowchart of a method for identifying multiple interference sources of distributed radars according to an embodiment of the present disclosure;

[0047] Figure 2 It is an electronic device according to an embodiment of the present disclosure. Detailed Embodiments

[0048] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present disclosure.

[0049] The terms used in the embodiments of the present disclosure are only for the purpose of describing specific embodiments, and are not intended to limit the embodiments of the present disclosure. The singular forms "a", "the", and "said" used in the embodiments of the present disclosure and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0050] It should be understood that although the terms first, second, third, etc. may be used in the embodiments of the present disclosure to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the embodiments of the present disclosure, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0051] As Figure 1 shown, it is a schematic flowchart of a method for identifying multiple interference sources of distributed radars according to an embodiment of the present disclosure, including:

[0052] S101: Construct a graph structure with the signals received by the distributed radar as nodes, where the edges connecting the nodes represent that the corresponding connected nodes are subject to the same type of interference.

[0053] Specifically, in a distributed radar system, the signals received by each radar node can be regarded as nodes in the graph structure. This graph structure is an abstract data structure used to represent objects (nodes or vertices) and the relationships (edges) between these objects. When considering interference identification, constructing such a graph model can help us better understand the propagation pattern and influence range of interference in the radar network.

[0054] Node: The signal received by each radar node is regarded as a node. This means that each node represents the information received by the radar at a specific location.

[0055] Edge: If two nodes are subject to the same type of interference, then there will be an edge connecting these two nodes. This edge indicates that there is a certain correlation between these two nodes, that is, they may be affected by the same interference source.

[0056] In such a graph structure, the connection between nodes not only indicates physical proximity or direct communication, but more importantly reflects the interaction under the interference environment. For example, if there are multiple radar nodes in a certain area and these nodes are affected by the same interference source due to geographical proximity or other reasons, then these nodes will form a closely connected subgraph in the graph, which helps to identify the interference pattern in that area.

[0057] Beneficial effects: Using the signals received by the distributed radar as nodes and establishing connections based on the common interference type can not only help us intuitively depict the existence form and influence degree of interference in the entire radar network, but also provide strong support for further interference suppression strategies. This method integrates the idea of graph theory and opens up a new way for understanding and solving complex electromagnetic interference problems.

[0058] Furthermore, the graph structure is , where is a set of radar nodes, is the set of edges,

[0059] Specifically, the set represents A set of radar nodes. Each node corresponds to a radar receiving station or sensor, which is responsible for collecting electromagnetic wave information at a specific geographical location. These nodes form the basic framework of the entire graph.

[0060] The set of edges contains all the connection relationships between the nodes. An edge indicates that there is a certain form of relationship between the nodes. Specifically, in the interference recognition scenario, this means that these two nodes are affected by the same type of interference. The presence or absence of an edge and its weight (if any) reflect the intensity of the interference correlation between the nodes.

[0061] Each row in the node feature matrix corresponds to the feature vector of a node. It can contain various information, such as signal strength, spectral characteristics, time series data, etc., which are important parameters used to describe the state of the radar site represented by the node. For example, in the interference recognition task, the feature vector may include information such as the frequency components, power levels, arrival angles of the interference signals.

[0062] S102: Decompose the real and imaginary parts of the signal, extract the signal features of the real part signal through the first channel, and extract the signal features of the imaginary part signal through the second channel, where both the first channel and the second channel include a convolutional neural network structure, which sequentially includes a first convolutional layer, a second convolutional layer, a pooling layer, and a fully connected layer.

[0063] Specifically, decompose the signal received by the distributed radar into a real part signal and an imaginary part signal, where the real part signal includes the in-phase baseband signal of the distributed radar, and the imaginary part signal includes the quadrature baseband signal of the distributed radar.

[0064] Among them, the in-phase baseband signal refers to the signal component that is in phase with the carrier wave, which is used to carry information and can be modulated or demodulated independently of the quadrature signal.

[0065] The quadrature baseband signal refers to the signal component that has a 90-degree (π / 2 radians) phase difference relative to the carrier wave and can carry information independently without interfering with the information transmission of the in-phase component.

[0066] Beneficial effects: It not only helps to more accurately analyze the received information but also provides convenient conditions for subsequent signal processing. In addition, through such a representation method, it is also easier to implement digital signal processing algorithms.

[0067] Furthermore, in order to remove redundant information and improve the efficiency and accuracy of subsequent recognition, it is necessary to effectively extract the key features of the radar signals of each node and achieve dimensionality reduction processing of the signals.

[0068] Specifically, the above-mentioned in-phase baseband signal is input into the first channel, and the quadrature baseband signal is input into the second channel. Both the first channel and the second channel include a convolutional neural network structure, which successively includes a first convolutional layer, a second convolutional layer, a pooling layer, and a fully connected layer. The first convolutional layer outputs 32 channels, and the convolutional kernel size is 5×5. The second layer outputs 128 channels, using dilated convolution with a size of 3 and a dilation rate of 1, which helps to expand the receptive field without increasing the number of parameters and capture a wider range of spatial relationships.

[0069] Optionally, each convolutional layer implements the SAME boundary mode to keep the signal dimension size consistent before and after convolution. After passing through two convolutional blocks, the global average pooling layer averages the values of each channel of the feature map and reduces the feature dimension.

[0070] Subsequently, it passes through a fully connected layer to prepare for subsequent graph structure processing. The feature representation finally output by the fully connected layer is:

[0071] ,

[0072] where, is the feature representation of all nodes, is the feature vector of the -th node, represents the total number of nodes, is the dimension of each feature vector, represents the dimension of the original feature.

[0073] S103: Use the extracted signal features of the real part and the imaginary part as nodes and input them into the graph self-attention network. Classify the interference types in the signal through the signal features. Each signal feature corresponds to one or more feature vectors, and the feature vectors represent the attributes of the corresponding nodes.

[0074] Specifically, the Graph Self-Attention Network is a neural network model specifically designed to process graph-structured data. It enhances the learning ability of node features by introducing the self-attention mechanism. This model can apply the attention mechanism to each node in the graph, enabling each node to dynamically adjust the way of information aggregation according to the importance of its neighbor nodes, thereby capturing more complex patterns and dependencies.

[0075] In this disclosure, the input of the graph self-attention network is a set of different radar signals, and each node has its own feature vector, which are the signal features extracted previously.

[0076] Furthermore, classifying the interference types in the signal through the signal features includes:

[0077] For each pair of connected nodes, calculate an attention coefficient;

[0078] Normalize the nodes for which the attention coefficients are calculated;

[0079] Update the feature representation of each node according to the normalized attention coefficients.

[0080] Specifically, for each pair of connected nodes (i.e., edges) in the graph, calculate an attention coefficient, which reflects the importance of the target node relative to the source node.

[0081] The attention coefficient is calculated using the following formula:

[0082] ,

[0083] where, represents the degree of attention of node to node , is the activation function, is the parameter vector, is the weight matrix for linearly transforming the node feature vector, and represent the original feature vectors of node and node respectively.

[0084] Next, to ensure that these attention coefficients can be used as weights, they are usually normalized using softmax so that the sum of the attention coefficients corresponding to the neighbor nodes of each node is 1.

[0085] The normalization of the nodes for which the attention coefficients are calculated is carried out using the following formula:

[0086] ,

[0087] where, is the normalized attention coefficient of node to node , represents the set of all neighbor nodes of the node, is the subspace.

[0088] Finally, update the feature representation of each node by weighted averaging the features of the neighbor nodes using the normalized attention coefficients. This step allows each node to learn its new feature representation based on the relevance or importance of its neighbor nodes.

[0089] The update of the feature representation of each node is carried out using the following formula:

[0090] ,

[0091] Among them, is the updated feature vector, is a non-linear activation function.

[0092] Optionally, in order to further improve the fitting ability of the network, a multi-head attention mechanism is also adopted in the present invention. A total of independent attention mechanisms are used in the network, and their results are concatenated:

[0093] .

[0094] Since in the defined scenario, the first-order neighboring nodes related to node already contain all the information for the recognition of the same type of interference, only one layer of graph self-attention network is needed. That is:

[0095] ,

[0096] Among them, is the output feature vector obtained after node passes through the graph self-attention layer, is the new feature vector obtained after node applies the graph self-attention mechanism. In a single-layer GAT, this means that the node not only considers its own features, but also considers the information from its direct neighbor nodes, and these information are weighted and summed according to the attention coefficients.

[0097] The output result will be used as a mask for the representation learned by the decoupled graph convolutional network to enhance the network's ability to identify interference sources. In addition, the loss function used for classification is binary cross-entropy:

[0098] .

[0099] Among them, represents the total number of samples or nodes, which is the number of all radar nodes that need to perform interference recognition, For node , is the true binary label (0 or 1), indicating whether there is a specific type of interference. is the predicted probability of whether there is a specific type of interference for node , that is, the probability that the model thinks the node belongs to the positive class. When , the loss mainly depends on , encouraging the model to predict the probability of the positive class as accurately as possible; while when , it is , prompting the model to correctly predict the probability of the negative class.

[0100] S104: Input the signal with interference type classification information as nodes into the decoupled graph convolutional network. By decomposing the node features into multiple independent subspaces, identify whether the nodes are affected by the same interference source, and then construct a prediction adjacency matrix to distinguish different interference sources.

[0101] Specifically, in the original input graph, the defined edge is that two nodes receive the same type of interference, and this relationship includes two nodes being interfered by the same interference source and two nodes being interfered by different interference sources under the same interference type.

[0102] To further distinguish these two potential relationships, the present disclosure uses a decoupled graph convolutional network to predict the new connection relationships between nodes, that is, to determine whether two nodes are affected by the same interference source, so as to achieve the separation of interference sources and lay a solid foundation for subsequent cooperative anti-interference.

[0103] Assume that different types of interference sources are independent of each other and do not affect each other. For the node input of the decoupled graph convolutional network, that is, the key features extracted by the radar nodes consist of at least independent components, which indicates that at least potential factors need to be decoupled. For a node , its latent representation is expressed as , where is used to describe the th aspect of the node .

[0104] The network first projects the feature vector into different subspaces to initialize the embedding vectors of channels:

[0105]

[0106] where and are the parameters of the th subspace, and is the low-dimensional projection of the th subspace. To comprehensively understand the th aspect of the node , it is necessary to extract information from its neighborhood. This involves using and to construct .

[0107] Therefore, three assumptions are proposed for the relationships between nodes:

[0108] Hypothesis 1: The generation of links between nodes stems from the interaction of various hidden factors.

[0109] Hypothesis 2: If a node and its adjacent node are similar in the th subspace, this indicates that the link between these two nodes may be attributed to factor .

[0110] Hypothesis 3: If adjacent nodes show similarity in the th aspect and form a cluster in the th subspace, then factor is very likely the reason for connecting node to this specific group of adjacent nodes.

[0111] Hypothesis 1 and Hypothesis 2 conform to the definition of the radar node graph structure in the present invention. The generation of links in the distributed radar node graph stems from receiving similar signals, that is, node and its adjacent node receive the th type of interference simultaneously.

[0112] When the decoupled subspaces can well correspond to the interference types one by one, then the link between these two nodes can be attributed to factor .

[0113] Hypothesis 3 explains the principle of link prediction using the decoupled graph convolutional network. Adjacent nodes can form clusters by calculating the similarity distance in the th subspace, and these clusters are very likely to correspond to different interference sources under the same interference type.

[0114] According to the above three hypotheses, the decoupled representation of nodes can be learned by using a decoupling layer based on a dynamic allocation mechanism:

[0115] ,

[0116] ,

[0117] ,

[0118] ,

[0119] where is the number of iterations. represents the connection probability between node and caused by the th factor, while Indicates aggregation nodes in the th subspace relative to the node importance. They satisfy , and for all , there is and . and are trainable trade-off coefficients, which will be adjusted during the training process, and their adjustment range is within .

[0120] The dynamic allocation mechanism will iteratively infer and , and construct . At the same time, this mechanism allows the coexistence of multiple relationships between nodes. It should be noted that there are a total of decoupling layers, and in each layer , is finally assigned the value of .

[0121] The decoupled embedding of the node output by the decoupling layer is the connection of the components learned in all channels, that is,

[0122] ,

[0123] Since the main goal of the decoupled graph convolutional network is to enable the graph encoder to generate decoupled representations for each latent factor . The different factors extracted by the dynamic routing mechanism are designed to focus on different reasons for the connections between nodes, and each channel captures mutually exclusive information related to each latent factor.

[0124] To improve the decoupling effect and obtain a more interpretable explanation, when executing the dynamic allocation mechanism, the present invention imposes constraints on these subspaces to ensure that the focused perspectives of the K subspaces focus on different interference types. The specific method is to use the judgment result of the classifier on the interference type as a mask. As shown in the following formula:

[0125] ,

[0126] When the classifier determines that the th interference type does not exist, the output of the corresponding th subspace is set to 0, otherwise no operation is performed. The final representation learned by all nodes after being masked is:

[0127] ,

[0128] The finally predicted adjacency matrix is constructed by the inner product of , that is . Since the upper triangular part of the adjacency matrix contains all edges, therefore, the upper triangular part of the adjacency matrix for link prediction is compared with the label, and the binary cross-entropy between them is calculated:

[0129] ,

[0130] The loss function of the entire network framework is:

[0131] ,

[0132] where is a hyperparameter used to control the balance between the decoupled graph convolutional network and the classifier.

[0133] Beneficial effects: A distributed radar multi-jamming source identification method provided by the present disclosure realizes precise identification of different types of jamming sources by constructing a distributed radar interference perception network model and comprehensively analyzing data from multiple radar nodes, which is beneficial to the effective acquisition of parameter information in subsequent tasks. This method utilizes the cooperation between each radar site, enhances the anti-jamming ability of the system in a complex electromagnetic environment and the accuracy of jamming type identification, thereby greatly improving the multi-jamming source identification accuracy and the performance of subsequent radar cooperative anti-jamming.

[0134] The electronic device 200 may be a desktop computer, a notebook, a palm computer, a cloud server, and other electronic devices. The electronic device 200 may include but is not limited to a processor 201 and a memory 202. Those skilled in the art can understand that Figure 2 merely exemplifies the electronic device 200 and does not limit the electronic device 200. It may include more or fewer components than shown in the figure, or combine some components, or different components. For example, the electronic device may further include input and output devices, network access devices, buses, etc.

[0135] The processor 201 can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0136] The memory 202 can be an internal storage unit of the electronic device 200, for example, the hard disk or memory of the electronic device 200. The memory 202 can also be an external storage device of the electronic device 200, for example, a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. equipped on the electronic device 200. Further, the memory 202 can also include both the internal storage unit and the external storage device of the electronic device 200. The memory 202 is used to store the computer program 203 and other programs and data required by the electronic device. The memory 202 can also be used to temporarily store the data that has been output or will be output.

[0137] In the embodiments provided in the present disclosure, it should be understood that the disclosed device / electronic device and method can be implemented in other ways. For example, the device / electronic device embodiments described above are merely illustrative. For example, the division of modules or units is only a logical function division. In actual implementation, there can be other division methods. Multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical, mechanical or other form.

[0138] When an integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above-described embodiment methods of the present disclosure, it can also be completed by a computer program instructing relevant hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described various method embodiments can be implemented. The computer program can include computer program code, and the computer program code can be in the form of source code, object code, an executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0139] The above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than to limit them; although the present disclosure has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present disclosure, and should all be included in the protection scope of the present disclosure.

Claims

1. A distributed radar multi-interference source identification method, characterized in that: The following steps are involved: The signals received by the distributed radar are used as nodes to construct a graph structure, where the edges connecting the nodes indicate that the corresponding connected nodes are subject to the same type of interference; Decomposing the real part and the imaginary part of the signal, extracting the signal feature of the real part signal through a first channel, and extracting the signal feature of the imaginary part signal through a second channel, wherein the first channel and the second channel both include a convolutional neural network structure, which sequentially includes a first convolutional layer, a second convolutional layer, a pooling layer, and a fully connected layer; The extracted real part signal features and imaginary part signal features are input as nodes into the graph self-attention network, and the interference type in the signal is classified according to the signal features, each signal feature corresponds to one or more feature vectors, and the feature vectors represent the attributes of the corresponding nodes; The signal with interference type classification information is input as a node into the decoupled graph convolutional network. By decomposing the node features into multiple independent subspaces, it is identified whether the nodes are affected by the same interference source, and then a prediction adjacency matrix is ​​constructed to distinguish different interference sources.

2. The method according to claim 1, characterized in that: The graph structure is ,in yes A collection of radar nodes, is the set of edges, is a node feature matrix, where each row corresponds to a node's feature vector.

3. The method according to claim 1, characterized in that The real signal includes an in-phase baseband signal of the distributed radar, and the imaginary signal includes an orthogonal baseband signal of the distributed radar.

4. The method according to claim 1, characterized in that: The first convolution layer and the second convolution layer output multiple channels, and the output channels of the first convolution layer are less than the output channels of the second convolution layer. All convolution layers adopt the SAME boundary mode. The second convolution layer outputs to the pooling layer to compress the feature map and average the value of each channel.

5. The method according to claim 1, characterized in that: The feature representation of the output of the fully connected layer is: ， in, is the feature representation of all nodes, For the The feature vector of a node, Indicates the total number of nodes, is the dimension of each feature vector, Represents the dimension of the original features.

6. The method according to claim 1, characterized in that The classifying the interference type in the signal by the signal feature comprises: For each pair of connected nodes, calculate an attention coefficient; Normalize the nodes for calculating attention coefficients; According to the normalized attention coefficient, the feature representation of each node is updated.

7. The method according to claim 6, characterized in that The calculation of the attention coefficient uses the following formula: ， in, Representation Node For Node The degree of attention, is the activation function, is the parameter vector, is the weight matrix used to linearly transform the node feature vector, and Respectively represent nodes and nodes The original feature vector of .

8. The method according to claim 7, characterized in that The normalization process for the nodes for calculating the attention coefficient uses the following formula: ， in, Is a node For Node The normalized attention coefficient of Represents the set of all neighbor nodes of a node, It is a subspace.

9. The method according to claim 8, characterized in that The updating of the feature representation of each node uses the following formula: ， in, is the updated feature vector, is a non-linear activation function.

10. The method according to claim 9, characterized in that The method further includes concatenating the updated output features using multi-head attention after updating the feature representation of each node: 。 11. The method according to claim 1, characterized in that: Decomposing node features into multiple independent subspaces, identifying whether nodes are affected by the same interference source, and then constructing a prediction adjacency matrix to distinguish different interference sources includes: The feature vector of each node Project to In different subspaces, initialize Embedding vector for channels: ， in and It is The parameters of the subspace, For the A low-dimensional projection of a subspace; For each node, its neighborhood information is used to update its decoupled representation. The calculation formula is: ， ， ， ， in, is the number of iterations, Representation Node and Between The connection probability caused by the factor is Indicated in Aggregate nodes in subspaces Relative to the node The importance of satisfying , and for all ,have and , and is a trainable trade-off coefficient that will be adjusted during the training process, and its adjustment range is within; The interference type of each node is judged. If it is judged that a certain type of interference does not exist at a certain node, the corresponding subspace output will be set to 0, otherwise it will remain unchanged; Using the formula , construct the predicted adjacency matrix, yes The matrix of is the number of nodes, is the feature dimension of each node, yes The transposed matrix of is the activation function.

12. An electronic device, characterized in that: include: one or more processors; A storage unit, used to store one or more programs, which, when executed by the one or more processors, enable the one or more processors to implement the distributed radar multi-interference source identification method according to any one of claims 1 to 11.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it can implement the distributed radar multiple interference source identification method according to any one of claims 1 to 11.

Images