CN116304553A - Interference type identification method based on deep learning under multi-modulation system - Google Patents
Interference type identification method based on deep learning under multi-modulation system Download PDFInfo
- Publication number
- CN116304553A CN116304553A CN202211708044.0A CN202211708044A CN116304553A CN 116304553 A CN116304553 A CN 116304553A CN 202211708044 A CN202211708044 A CN 202211708044A CN 116304553 A CN116304553 A CN 116304553A
- Authority
- CN
- China
- Prior art keywords
- interference
- target signal
- trained
- model
- identification
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000013135 deep learning Methods 0.000 title claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 29
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000013527 convolutional neural network Methods 0.000 claims description 26
- 230000015654 memory Effects 0.000 claims description 19
- 238000004590 computer program Methods 0.000 claims description 7
- 230000004044 response Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 3
- 239000002131 composite material Substances 0.000 abstract description 9
- 230000006870 function Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000004913 activation Effects 0.000 description 5
- 238000004891 communication Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 238000011176 pooling Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention specifically discloses an interference type identification method based on deep learning under a multi-modulation system, which comprises the following steps: acquiring a target signal; preprocessing a target signal to obtain target signal data; inputting the target signal data into a pre-trained interference recognition model to obtain the interference type of the target signal; and identifying the target signal according to the interference type of the target signal. The invention can identify the interference type of single interference/composite interference of signal superposition under a multi-modulation system, helps the selection of the subsequent anti-interference strategy, and has wide application range and strong generalization capability. Meanwhile, the invention not only can realize the identification of single interference, but also can identify partial composite interference, and has high identification accuracy and wide identification types. Finally, the method adopts an off-line network training mode, network training is not needed during on-line interference identification, and the identification speed is high.
Description
Technical Field
The invention particularly relates to an interference type identification method based on deep learning under a multi-modulation system.
Background
Along with the continuous increase of electronic products, electromagnetic signals are increasingly complex in space, time and frequency domains; meanwhile, under the development of an information-based society, various information systems are extremely dependent on wireless communication, and electronic countermeasure is particularly important in modern wireless communication. In the electronic defense of the communication system, a targeted anti-interference means is adopted to effectively ensure reliable communication. However, it is difficult to find a reliable anti-interference means in complex electromagnetic environments that can combat a wide variety of interference types. Therefore, the signal needs to be preprocessed, the interference type is identified, and an important basis is provided for the receiver to determine the anti-interference decision.
The existing interference recognition method can recognize partial interference patterns, but most of the methods are based on single interference patterns, and detection and recognition of research compound interference are few. Meanwhile, the existing interference recognition methods are all aimed at signal interference recognition under a certain specific modulation system, the interference recognition performance is obviously reduced after the modulation system is changed, the model cannot be reused, and the method cannot be used in a complex communication system with multiple modulation systems.
Disclosure of Invention
The invention aims to provide an interference type identification method based on deep learning under a multi-modulation system, so as to solve the problem of single interference/composite interference type identification under the multi-modulation system.
In order to achieve the above purpose, the invention adopts the following technical scheme:
an interference type identification method based on deep learning under a multi-modulation system comprises the following steps: acquiring a target signal; preprocessing the target signal to obtain target signal data; inputting the target signal data into a pre-trained interference recognition model to obtain the interference type of the target signal; and identifying the target signal according to the interference type of the target signal.
Further, the interference recognition model is composed of a convolutional neural network and a long-short-time memory network, wherein the convolutional neural network and the long-short-time memory network are used for extracting local features of the target signal data.
Further, the convolutional neural network comprises at least two convolutional layers.
Further, the convolutional neural network and the long-short-time memory network are connected by a full-connection layer, and the full-connection layer is used for integrating local features extracted by the convolutional neural network and the long-short-time memory network into global features.
Further, the interference recognition model is trained according to the following steps: acquiring a training sample set, wherein the training sample set comprises a sample signal and a sample identification result; inputting the sample signal into a model to be trained to obtain a recognition result corresponding to the sample signal; determining a recognition result loss value according to the sample recognition result and the recognition result; and in response to determining that the recognition result loss value does not meet the preset condition, determining that the model to be trained does not complete training, and adjusting relevant parameters in the model to be trained.
Further, the method further comprises the steps of: and determining that the model to be trained is trained, and determining the model to be trained as an interference recognition model in response to determining that the recognition result loss value meets a preset condition.
An interference type recognition device based on deep learning under a multi-modulation system, comprising: an acquisition unit configured to acquire a target signal; the processing unit is configured to preprocess the target signal to obtain target signal data; a generation unit configured to input the target signal data to a pre-trained interference recognition model to obtain an interference type of the target signal; and the identification unit is configured to identify the target signal according to the interference type of the target signal.
An electronic device comprising a processor and a memory electrically connected to the processor, the memory being adapted to store a computer program, the processor being adapted to invoke the computer program to perform the steps of the method.
A computer readable storage medium, wherein said computer readable storage medium stores a computer program, said computer program being capable of being invoked by a processor to perform the steps of said method.
Therefore, the invention can identify the interference type of single interference/composite interference of signal superposition under a multi-modulation system, and help the selection of the subsequent anti-interference strategy. Compared with the existing interference recognition algorithm, the method can adopt the same training network for various modulation system signals, and has wide application range and strong generalization capability. Meanwhile, the invention not only can realize the identification of single interference, but also can identify partial composite interference, and has high identification accuracy and wide identification types. Finally, the method adopts an off-line network training mode, network training is not needed during on-line interference identification, and the identification speed is high.
Drawings
Fig. 1 is a schematic flow chart of an interference type recognition method based on deep learning under a multi-modulation system according to some embodiments of the present invention;
FIG. 2 is a schematic diagram of the structure of an interference recognition model;
FIG. 3 is a schematic diagram of a convolutional neural network;
fig. 4 is a schematic diagram of the structure of a long and short term memory network.
Detailed Description
The advantages and features of the present invention will become more fully apparent from the following description and appended claims, taken in conjunction with the accompanying drawings and detailed description. It should be noted that the drawings are in a very simplified form and are adapted to non-precise proportions, merely for the purpose of facilitating and clearly aiding in the description of embodiments of the invention.
It should be noted that, in order to clearly illustrate the present invention, various embodiments of the present invention are specifically illustrated by the present embodiments to further illustrate different implementations of the present invention, where the various embodiments are listed and not exhaustive. Furthermore, for simplicity of explanation, what has been mentioned in the previous embodiment is often omitted in the latter embodiment, and therefore, what has not been mentioned in the latter embodiment can be referred to the previous embodiment accordingly.
As shown in fig. 1, fig. 1 is a flow chart of an interference type recognition method based on deep learning under a multi-modulation system according to some embodiments of the present invention.
Step 101, a target signal is acquired.
In some embodiments, an execution subject (e.g., a server) of a deep learning-based interference type recognition method under a multi-modulation regime may acquire a target signal.
Step 102, preprocessing the target signal to obtain target signal data.
In some embodiments, the execution body may perform preprocessing on the target signal to obtain target signal data.
Specifically, the preprocessing generally refers to cutting off the target signal, and the input size of each piece of data after cutting off is 2×n.
And step 103, inputting the target signal data into a pre-trained interference recognition model to obtain the interference type of the target signal.
In some embodiments, the executing body may input the target signal data into a pre-trained interference recognition model, so as to obtain an interference type of the target signal.
Specifically, common typical interference types include: 1) Partial band interference, which is the concentration of noise energy in a specified frequency band. The ratio of the interference to the bandwidth of the received signal in the frequency domain is broadband interference within the range of 10% -50%, and the ratio is narrowband interference within the range of 10%. 2) Single tone interference, which generally refers to audio interference having only one frequency bin, is essentially a sinusoidal signal. 3) The multi-tone interference is formed by superposing a plurality of single-tone interference time domain waveforms. 4) Pulse interference, time domain gaussian pulse interference, refers to a gaussian pulse signal with a certain pulse period and duration. 5) Sweep frequency interference, which is also called chirping interference, has the characteristic that its instantaneous frequency changes linearly with time, and can be regarded as a single-tone signal at a certain time point, and has the characteristics of broadband signal and dynamic scanning at a certain time period. 6) The forward interference refers to that the received pulse signals are directly or indirectly transmitted after being simply processed. The typical interference patterns can be used independently, and can also be overlapped to form composite interference, so that strong interference influence is carried out on signals, and difficulty is increased for interference identification.
In some optional implementations of some embodiments, the interference recognition model is composed of a convolutional neural network and a long-short-time memory network, where the convolutional neural network and the long-short-time memory network are used for local feature extraction of the target signal data.
In some alternative implementations of some embodiments, the convolutional neural network includes at least two convolutional layers.
In some optional implementations of some embodiments, the convolutional neural network and the long-short-time memory network are connected by a fully-connected layer, and the fully-connected layer is configured to integrate local features extracted by the convolutional neural network and the long-short-time memory network into global features.
Specifically, as shown in fig. 2, 3 and 4, fig. 2 is a schematic structural diagram of an interference recognition model, fig. 3 is a schematic structural diagram of a convolutional neural network, and fig. 4 is a schematic structural diagram of a long-short-term memory network.
The interference recognition model is composed of a CNN and an LSTM network, and the interference pattern recognition result is obtained after the output of the two networks are combined, and the method comprises the following steps: the original signal samples after baseband sampling respectively enter CNN and LSTM; in a CNN network, firstly, an original signal sample is formed into an input layer for output according to a time sequence; the input layer output is used as the input of a first convolution layer, the complete primary characteristics of an original signal are extracted, the extracted characteristics are subjected to maximum pooling downsampling, and then the local corresponding normalization (Local Response Normalization, LRN) layer processing is carried out, so that the generalization capability of the model is enhanced; the output of the first convolution layer is used as the input of the second convolution layer, new dimension characteristics are extracted again, the extracted characteristics are used as the input of the LRN layer, and then the maximum pooling downsampling is adopted; the input of the second convolution layer is used as the input of a full connection layer, and L2 regularization treatment is needed to be added in the layer in order to prevent overfitting; the output of the full connection layer is used as the output of the whole CNN network; in an LSTM network, firstly, original signal samples are sequentially input into the LSTM network according to a time sequence, the hidden layer output of a first LSTM node is used as input to be transmitted to a second node, and the hidden layer output of a last node comprises time domain characteristic information of signals and the like; in training mode, the state output of LSTM is passed as input to a dropout layer to prevent model over-fitting; finally, adopting a full connection layer, wherein the output of the full connection layer is used as the output of the whole LSTM network; and after the outputs of the CNN network and the LSTM network are spliced, the outputs are used as the input of a new full-connection layer, the data are re-fitted, all local features are integrated into global features, the probability of N types of interference patterns is given by selecting a normalization function softmax, and the types of the interference patterns are identified.
In some optional implementations of some embodiments, the interference recognition model is trained according to the following steps: acquiring a training sample set, wherein the training sample set comprises a sample signal and a sample identification result; inputting the sample signal into a model to be trained to obtain a recognition result corresponding to the sample signal; determining a recognition result loss value according to the sample recognition result and the recognition result; and in response to determining that the recognition result loss value does not meet the preset condition, determining that the model to be trained does not complete training, and adjusting relevant parameters in the model to be trained.
In some optional implementations of some embodiments, in response to determining that the recognition result loss value meets a preset condition, the execution body may determine that the training of the model to be trained is completed, and determine the model to be trained as the interference recognition model.
Specifically, the original signal sample is a time continuous signal under any one of multiple modulation systems such as FSK/BPSK/QPSK/MSK/GMSK/OFDM, the interference type comprises single interference and composite interference, the interference type comprises two paths of I/Q including a real part and an imaginary part, the signal is required to be cut off during network training, the input size of each section of data after the cutting off is 2 XN, and the signal is output as an interference pattern identification mark for each section of signal data. The data set is divided into a training set and a testing set according to a certain proportion, the training set is used for training network parameters, the testing set utilizes a trained network to test, and the network performance is verified.
Within a CNN network: the data size of the input layer is 2 XN, and the first convolution layer uses N conv1 ×M conv1 Convolution kernel size of S conv1 And (3) convolution kernels, namely extracting time-frequency domain characteristics of the original signals, and performing nonlinear processing by using a ReLU activation function after weighting and offset operation. After the ReLU activation function, a size N is used pool1 ×M pool1 And the Step length is Step pool11 ×Step pool12 The maximum pooling layer of (2) processes the data, and the size and step size at the moment are kept inconsistent as much as possible, so that the richness of the data is increased. The result output by the pooling layer is processed by an LRN layer, so that the network training speed and the generalization capability of the network are improved. The output of the first convolution layer serves as the input to the second convolution layer, which uses N conv2 ×M conv2 Convolution kernel size of S conv2 And (3) deeply extracting data features again by using convolution kernels, and performing nonlinear processing by using a ReLU activation function after weighting and bias operation. After the ReLU activation function, an LRN layer is processed, and a convolution kernel weighting value with obvious characteristic reaction is selected from a plurality of convolution kernels. After LRN layer, one size N is used pool2 ×M pool2 And the Step length is Step pool21 ×Step pool22 Is used for processing data. Will pass one after two convolutional layersFull-connection layer with hidden node number N local1 . To prevent the overfitting problem, an L2 regularization process is required at this layer. Finally, a non-linearization operation is performed using the ReLU activation function.
Within the LSTM network: the data of the input layer is 2 XN, and is sequentially input into a step length of N LSTM In the LSTM network, LSTM cells are fundamental components of the LSTM network, and the flow of information between LSTM cells is controlled by three gates: an input gate, a forget gate, and an output gate; each gate has a value of 0 to 1, is activated by a sigmoid function, and defines the cell state and output state of the current t cell as C t And h t The cell state and output state of each cell is transferred to its next cell, respectively. In the training state, a dropout layer needs to be added after LSTM to prevent the model from being over fitted. The output of the LSTM layer or dropout layer is used as input and needs to pass through a fully connected layer to convert the output characteristics into the required vector dimension as the output of the LSTM network.
The output of the CNN network and the output of the LSTM network are connected in series through vectors and serve as input of a full connection layer, data are subjected to re-fitting through feature weighted summation, all local features are integrated into global features, the score of each interference category is obtained, the score is mapped into the probability of each category through a softmax layer, and finally the interference type of signal superposition is judged.
In the training state, the judged interference type is required to be compared with a real result, the calculation loss of a batch of samples is calculated, the average loss of the samples is recorded as loss, an optimizer is adopted to continuously and iteratively optimize network parameters according to the minimum loss as a target, and the set training_rate is taken as a learning rate, so that the training network parameters are obtained. In the test state, the test sample uses the network parameters obtained by training to identify the signals, and the interference type obtained by final judgment is the final output of the network.
And 104, identifying the target signal according to the interference type of the target signal.
In some embodiments, the executing entity may identify the target signal according to an interference type of the target signal.
Therefore, the invention can identify the interference type of single interference/composite interference of signal superposition under a multi-modulation system, and help the selection of the subsequent anti-interference strategy. Compared with the existing interference recognition algorithm, the method can adopt the same training network for various modulation system signals, and has wide application range and strong generalization capability. Meanwhile, the invention not only can realize the identification of single interference, but also can identify partial composite interference, and has high identification accuracy and wide identification types. Finally, the method adopts an off-line network training mode, network training is not needed during on-line interference identification, and the identification speed is high.
Claims (10)
1. An interference type identification method based on deep learning under a multi-modulation system comprises the following steps:
acquiring a target signal;
preprocessing the target signal to obtain target signal data;
inputting the target signal data into a pre-trained interference identification model to obtain the interference type of the target signal;
and identifying the target signal according to the interference type of the target signal.
2. The method of claim 1, wherein the interference recognition model is comprised of a convolutional neural network and a long-short-time memory network, wherein the convolutional neural network and the long-short-time memory network are used to perform local feature extraction on the target signal data.
3. The method of claim 2, wherein the convolutional neural network comprises at least two convolutional layers.
4. The method of claim 2, wherein the convolutional neural network and the long-short-term memory network are connected by a fully-connected layer for integrating local features extracted by the convolutional neural network and the long-short-term memory network into global features.
5. The method of claim 1, wherein the interference recognition model is trained in accordance with:
acquiring a training sample set, wherein the training sample set comprises a sample signal and a sample identification result;
inputting the sample signal into a model to be trained to obtain a recognition result corresponding to the sample signal;
determining a recognition result loss value according to the sample recognition result and the recognition result;
and in response to determining that the recognition result loss value does not meet a preset condition, determining that the model to be trained does not complete training, and adjusting relevant parameters in the model to be trained.
6. The method of claim 5, wherein the method further comprises:
and in response to determining that the recognition result loss value meets a preset condition, determining that the training of the model to be trained is completed, and determining the model to be trained as an interference recognition model.
7. An interference type recognition device based on deep learning under a multi-modulation system, comprising:
an acquisition unit configured to acquire a target signal;
the processing unit is configured to preprocess the target signal to obtain target signal data;
the generating unit is configured to input the target signal data into a pre-trained interference recognition model to obtain the interference type of the target signal;
and the identification unit is configured to identify the target signal according to the interference type of the target signal.
8. The apparatus of claim 7, wherein the interference recognition model is comprised of a convolutional neural network and a long-short-time memory network, wherein the convolutional neural network and the long-short-time memory network are used to perform local feature extraction on the target signal data.
9. An electronic device comprising a processor and a memory electrically connected to the processor, the memory for storing a computer program, the processor for invoking the computer program to perform the steps of the above method.
10. A computer readable storage medium, wherein the computer readable storage medium stores a computer program that can be invoked by a processor to perform the steps of the above method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211708044.0A CN116304553A (en) | 2022-12-29 | 2022-12-29 | Interference type identification method based on deep learning under multi-modulation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211708044.0A CN116304553A (en) | 2022-12-29 | 2022-12-29 | Interference type identification method based on deep learning under multi-modulation system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116304553A true CN116304553A (en) | 2023-06-23 |
Family
ID=86834877
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211708044.0A Pending CN116304553A (en) | 2022-12-29 | 2022-12-29 | Interference type identification method based on deep learning under multi-modulation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116304553A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116743211A (en) * | 2023-08-16 | 2023-09-12 | 北京前景无忧电子科技股份有限公司 | Anti-interference method for power carrier communication |
-
2022
- 2022-12-29 CN CN202211708044.0A patent/CN116304553A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116743211A (en) * | 2023-08-16 | 2023-09-12 | 北京前景无忧电子科技股份有限公司 | Anti-interference method for power carrier communication |
| CN116743211B (en) * | 2023-08-16 | 2023-10-27 | 北京前景无忧电子科技股份有限公司 | Anti-interference method for power carrier communication |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Automatic modulation classification using CNN-LSTM based dual-stream structure | |
| Zhang et al. | An efficient deep learning model for automatic modulation recognition based on parameter estimation and transformation | |
| CN109993280A (en) | A kind of underwater sound source localization method based on deep learning | |
| CN112347871B (en) | Interference signal modulation identification method for communication carrier monitoring system | |
| CN110598530A (en) | Small sample radio signal enhanced identification method based on ACGAN | |
| Huynh-The et al. | Exploiting a low-cost CNN with skip connection for robust automatic modulation classification | |
| CN110120926A (en) | Modulation mode of communication signal recognition methods based on evolution BP neural network | |
| Li et al. | A deep convolutional network for multitype signal detection and classification in spectrogram | |
| Lin et al. | Modulation recognition using signal enhancement and multistage attention mechanism | |
| CN116304553A (en) | Interference type identification method based on deep learning under multi-modulation system | |
| CN112488294A (en) | Data enhancement system, method and medium based on generation countermeasure network | |
| Li et al. | Automatic modulation classification based on bispectrum and CNN | |
| CN115114949A (en) | Intelligent ship target identification method and system based on underwater acoustic signals | |
| CN114943245A (en) | Automatic modulation recognition method and device based on data enhancement and feature embedding | |
| Ghasemzadeh et al. | GGCNN: an efficiency-maximizing gated graph convolutional neural network architecture for automatic modulation identification | |
| Ali et al. | Modulation format identification using supervised learning and high-dimensional features | |
| CN114980122A (en) | Small sample radio frequency fingerprint intelligent identification system and method | |
| Peng et al. | A noise-robust modulation signal classification method based on continuous wavelet transform | |
| CN117110446A (en) | Method for identifying axle fatigue crack acoustic emission signal | |
| Ruan et al. | Automatic recognition of radar signal types based on CNN-LSTM | |
| CN114285545B (en) | Side channel attack method and system based on convolutional neural network | |
| CN114936570A (en) | Interference signal intelligent identification method based on lightweight CNN network | |
| CN113869238A (en) | Cognitive Internet of vehicles intelligent frequency spectrum sensing method and system | |
| CN113343924A (en) | Modulation signal identification method based on multi-scale cyclic spectrum feature and self-attention generation countermeasure network | |
| Feng et al. | FCGCN: Feature Correlation Graph Convolution Network for Few-Shot Individual Identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Yang Jian Inventor after: Zhang Wenjing Inventor after: Shang Jiadong Inventor after: Zhu Xiaoqing Inventor after: Zhang Shuo Inventor after: Jia Buyun Inventor before: Zhang Wenjing Inventor before: Yang Jian Inventor before: Shang Jiadong Inventor before: Zhu Xiaoqing Inventor before: Zhang Shuo Inventor before: Jia Buyun |