Automatic modulation identification method and device based on consistency constraint and feature prototype

Abstract

The invention discloses an automatic modulation identification method and device based on consistency constraint and a feature prototype. The method comprises the following steps: acquiring a modulation signal to be identified; inputting a to-be-recognized modulation signal into the trained automatic modulation recognition model, performing feature extraction of high signal-to-noise ratio data in the to-be-recognized modulation signal, performing feature extraction of low signal-to-noise ratio data in the to-be-recognized modulation signal, outputting a recognition probability, and obtaining a classification recognition result of the to-be-recognized modulation signal; wherein the trained automatic modulation recognition model is obtained by training the initial automatic modulation recognition model by taking data of a preset category as a training data set. The method can better deal with a low signal-to-noise ratio environment, and does not depend on expert experience any more.

Description

Technical Field

[0001] The invention belongs to the technical field of automatic modulation recognition, and in particular relates to an automatic modulation recognition method and device based on consistency constraints and feature prototypes. Background Art

[0002] With the rapid development of information technology, the requirements for wireless communications are getting higher and higher, among which the utilization and management of wireless spectrum have become crucial. Full wireless spectrum sensing technology is a technology based on software-defined radio that can monitor the wireless spectrum in real time and evaluate channel quality and signal strength, etc. Full wireless spectrum sensing technology needs to be combined with related technical tasks such as identification and classification to efficiently monitor and manage the spectrum. Among them, modulation signal classification, as one of the key tasks of wireless communication systems, has become an important means to ensure the quality and stability of wireless communication transmission. In addition, modulation signal classification can also be used in fields such as network security, spectrum allocation, and signal anti-interference in wireless communications.

[0003] However, actual communications involve various complex interferences and channel fading, which can have a serious impact on the structure and characteristics of the signal, making signal modulation recognition technology face great challenges. In order to improve recognition accuracy and reliability, expand the scope of application, and reduce computational complexity, researchers in various countries are actively researching and exploring, striving to achieve more efficient, intelligent, and reliable automatic modulation recognition technology in the future to meet people's strict requirements for communication quality and stability.

[0004] In the early stage, automatic modulation recognition mainly classified signal features manually selected based on decision boundaries, etc. The recognition effect was extremely dependent on the selected features, and manual feature selection was very complicated. When the channel is affected by noise, multipath fading, etc., the recognition efficiency will be greatly affected. Although the machine learning methods that emerged later have the advantages of high classification efficiency and high performance compared with traditional methods, the classification effect still depends on the experience of experts to a certain extent. The deep learning method is different. The deep learning network can extract the high-dimensional features of communication signal data through its powerful feature extraction ability, effectively realize the communication signal recognition and classification tasks, and no longer rely on the expert features selected manually. Experiments have shown that the modulation recognition method based on deep learning is more suitable for processing radio time series sample data, and the equivalent accuracy obtained is several times higher than that of the traditional feature-based classifier. This end-to-end training of raw data using neural networks can effectively learn discriminative features and has achieved great success in modulation recognition tasks.

[0005] Although deep learning has made some progress in the field of modulation mode recognition, its performance still needs to be significantly improved in low signal-to-noise ratio environments. With the increasing complexity of the electromagnetic environment of wireless communications, received signals frequently encounter various interference signals, which poses a challenge to the accuracy of signal processing. What is particularly serious is that in certain fields of application, the opponent may use intelligent interference sources to automatically lock and attack the specific frequency points of our signals, greatly increasing the difficulty of the receiver to distinguish between valid signals and interference in the frequency domain. In addition, the opponent may also adopt a strategy of reducing the transmitter power to reduce the possibility of its signal being intercepted by us, which will lead to an extremely low signal-to-noise ratio of the signal received by us. Given that modulation mode recognition is a key link in the demodulation process, in-depth research and improvement of modulation mode recognition capabilities under low signal-to-noise ratio conditions are of extremely important practical significance for ensuring communication security and improving the ability to obtain information in a certain scenario. Summary of the invention

[0006] In order to solve the above problems existing in the prior art, the present invention provides an automatic modulation recognition method and device based on consistency constraints and feature prototypes. The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0007] In a first aspect, the present invention provides an automatic modulation recognition method based on consistency constraints and feature prototypes, comprising:

[0008] Acquire a modulation signal to be identified;

[0009] Input the modulation signal to be identified into the trained automatic modulation recognition model, extract the features of the high signal-to-noise ratio data in the modulation signal to be identified, and extract the features of the low signal-to-noise ratio data in the modulation signal to be identified, output the recognition probability, and obtain the classification recognition result of the modulation signal to be identified;

[0010] The trained automatic modulation recognition model is obtained by training the initial automatic modulation recognition model using data of a preset category as a training data set; the data of the preset category includes a modulation signal.

[0011] In a second aspect, the present invention also includes an automatic modulation recognition device based on consistency constraints and feature prototypes, comprising:

[0012] A data acquisition module, used for acquiring a modulation signal to be identified;

[0013] A data processing module is used to input the modulation signal to be identified into the trained automatic modulation recognition model, extract the features of the high signal-to-noise ratio data in the modulation signal to be identified, and extract the features of the low signal-to-noise ratio data in the modulation signal to be identified, output the recognition probability, and obtain the classification recognition result of the modulation signal to be identified;

[0014] The trained automatic modulation recognition model is obtained by training the initial automatic modulation recognition model using data of a preset category as a training data set; the data of the preset category includes a modulation signal.

[0015] Beneficial effects of the present invention:

[0016] The present invention provides an automatic modulation recognition method and device based on consistency constraints and feature prototypes, which can better cope with low signal-to-noise ratio environments, no longer rely on expert experience, and will not cause signal features to be overly suppressed or lost. After the model of the present invention is deployed, no additional computing overhead is generated, and no threshold hyperparameters need to be manually set.

[0017] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of an automatic modulation recognition method based on consistency constraints and feature prototypes provided by an embodiment of the present invention;

[0019] Figure 2 is a schematic diagram of an automatic modulation recognition model provided by an embodiment of the present invention;

[0020] Figure 3 It is a schematic diagram of simulation experiment results provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0022] Among the existing technologies, the feature extraction-based methods have a certain dependence on expert features and perform poorly in low signal-to-noise ratio environments; the soft threshold deep learning-based methods may cause signal features to be over-suppressed or lost, and the network's generalization ability is limited; the wavelet denoising-based methods require additional wavelet denoising preprocessing of the data, which increases the computational complexity of the algorithm and is extremely sensitive to the choice of threshold.

[0023] In view of this, the present invention provides an automatic modulation recognition method based on consistency constraints and feature prototypes. Based on deep learning, the feature prototype of high signal-to-noise ratio data is calculated, and consistency constraints are imposed on the features extracted from low signal-to-noise ratio data and the feature prototype of high signal-to-noise ratio data, so that the model can autonomously learn anti-noise features and improve the recognition performance of the deep learning model in high noise scenarios.

[0024] See also Figure 1 , Figure 1 is a flow chart of an automatic modulation recognition method based on consistency constraints and feature prototypes provided by an embodiment of the present invention. Figure 2 is a schematic diagram of an automatic modulation recognition model provided by an embodiment of the present invention. An automatic modulation recognition method based on consistency constraints and feature prototypes provided by the present invention includes:

[0025] S101: Obtain a modulated signal to be identified.

[0026] S102, input the modulation signal to be identified into the trained automatic modulation recognition model, perform feature extraction of high signal-to-noise ratio data in the modulation signal to be identified, and perform feature extraction of low signal-to-noise ratio data in the modulation signal to be identified, output the recognition probability, and obtain the classification recognition result of the modulation signal to be identified.

[0027] The trained automatic modulation recognition model is obtained by training the initial automatic modulation recognition model using data of a preset category as a training data set; the data of the preset category includes a modulation signal.

[0028] Specifically, in this embodiment, before the initial automatic modulation recognition model is trained, the following steps are also included:

[0029] Construct an automatic modulation recognition model; where

[0030] The automatic modulation recognition model includes a feature extractor and a classifier, the feature extractor includes a first convolution block, a second convolution block, a third convolution block, a fourth convolution block, a fifth convolution block, a first timing layer, a second timing layer, a first fully connected layer and a second fully connected layer, and the classifier includes a classification layer; the input end of the first convolution block receives a sine wave and a cosine wave of a signal to be processed, the input end of the second convolution block receives a cosine wave of a signal to be processed, the input end of the third convolution block receives a sine wave of a signal to be processed, and the input end of the fourth convolution block receives a first signal, the first signal being a signal processed by the second convolution block and a signal processed by the third convolution block The input end of the fifth convolution block receives a connection signal, the second signal is a connection signal of the signal processed by the first convolution block and the signal processed by the fourth convolution block, the input end of the first timing layer receives the signal processed by the fifth convolution block, the input end of the second timing layer receives the signal processed by the first timing layer, the input end of the first fully connected layer receives the signal processed by the second timing layer, the input end of the second fully connected layer receives the signal processed by the first fully connected layer, the input end of the classification layer receives the signal processed by the second fully connected layer, and the output end of the classification layer outputs the prediction score of the signal to be processed.

[0031] In this embodiment, the first convolution block, the second convolution block, the third convolution block, the fourth convolution block and the fifth convolution block have the same structure and all include a convolution layer, a batch normalization layer and a RuLU activation layer; the number of convolution kernels of the convolution layer of the first convolution block is 50, the size of the convolution kernel is 2*8, the step size is 1, and the padding is 0; the number of convolution kernels of the convolution layer of the second convolution block is 50, the size of the convolution kernel is 1*8, the step size is 1, and the padding is 0; the number of convolution kernels of the convolution layer of the third convolution block is 50, the size of the convolution kernel is 1*8, the step size is 1, and the padding is 0; the number of convolution kernels of the convolution layer of the fourth convolution block is 50, the size of the convolution kernel is 1*8, the step size is 1, and the padding is 0; the number of convolution kernels of the convolution layer of the fifth convolution block is 100, the size of the convolution kernel is 2*5, the step size is 1, and the padding is 0;

[0032] The hidden layer dimension of the first temporal layer is 128, and the hidden layer dimension of the second temporal layer is 128;

[0033] The output dimension of the first fully connected layer is 128, and the output dimension of the second fully connected layer is 11.

[0034] In this embodiment, the initial automatic modulation recognition model is trained using data of a preset category as a training data set, including:

[0035] Acquire a training data set, where the training data set includes a plurality of training samples, and each training sample includes a modulated signal;

[0036] Label the types of training samples in the training data set to obtain the true labels of the types of training samples;

[0037] The training data set and the true labels of the types of training samples in the training data set are input into the initial automatic modulation recognition model for iterative training until the preset conditions are met to obtain a trained automatic modulation recognition model.

[0038] In this embodiment, the training data set and the true labels of the types of training samples in the training data set are input into the initial automatic modulation recognition model for iterative training, including:

[0039] For the first iteration training, the training data set is input into the automatic modulation recognition model, and the training samples in the training data set with a signal-to-noise ratio greater than or equal to 0 dB and correctly identified are subjected to feature extraction by the feature extractor to obtain the first feature vector of the training process; and the feature prototype of each type of training sample is obtained according to the first feature vector of each type of training sample;

[0040] In the second iteration training, the training data set is input into the automatic modulation recognition model, and the training samples in the training data set with a signal-to-noise ratio less than 0dB or greater than -10dB and correctly identified are subjected to feature extraction by the feature extractor to obtain a second feature vector of the training process; the feature prototype of each type of training sample in the first iteration training process is used to impose consistency constraints on the second feature vector of the training process to update the parameters of the current automatic modulation recognition model; the feature prototype of each type of training sample is recalculated using the updated parameters of the automatic modulation recognition model to update the feature prototype of each type of training sample;

[0041] For the third iterative training, the feature prototypes of each type of training samples updated in the previous iterative training are used to impose consistency constraints on the second feature vectors of the third iterative training process until the preset number of training times is completed.

[0042] It should be noted that the l-th iteration process needs to use the feature prototype of each type of training samples obtained in the l-1-th iteration process to impose a consistency constraint on the second feature vector in the l-th iteration process.

[0043] It should be noted that correct recognition is specifically represented by the maximum value of the probability score output after the signal sample is input into the model, which matches the true label of the sample.

[0044] In this embodiment, the preset condition includes that the loss function is minimized during a finite number of iterative training. Optionally, during the training process, the optimizer uses Adam, and the number of training epochs is 200.

[0045] Specifically, the expression of the loss function L1 in the first iteration training is:

[0046]

[0047] Among them, Y label Represented as training sample X label The corresponding prediction score is Represents the training sample X label The corresponding true label.

[0048] Specifically, the expression of the loss function L2 in the second iteration training is:

[0049]

[0050] Among them, f i represents the feature prototype of the training sample, represents the training samples whose signal-to-noise ratio is less than 0 dB and whose true label is the i-th category, n represents the total number of training sample categories, i = 1,…,n, and F(·) represents the feature extractor function.

[0051] Specifically, the feature prototype f of the i-th type of training sample i The expression is:

[0052]

[0053] Among them, Mean(·) represents the mean function, represents the training samples of the i-th class in the training data set with a signal-to-noise ratio greater than 0 dB and correct identification, and F(·) represents the feature extractor function.

[0054] In summary, the automatic modulation recognition method based on consistency constraints and feature prototypes provided by the present invention obtains the feature prototype of each type of training samples in each round of training during the training process, and uses the feature prototype of each type of samples updated in this round to impose consistency constraints on the feature vectors extracted in the next round of training process. In the process of finite iterative training, the value of the loss function is minimized, the parameters of the automatic modulation recognition model are determined, and a trained automatic modulation recognition model is constructed. This method enables the model to autonomously learn the anti-noise characteristics of the signal during the training process, and the model's recognition ability for low signal-to-noise ratio data can be improved without adding a noise reduction module in front of the model.

[0055] In an optional embodiment of the present invention, the effect of the automatic modulation recognition method based on consistency constraints and feature prototypes provided in the above embodiment is verified by simulation experiments, specifically:

[0056] 1. Simulation conditions

[0057] The hardware platform of the simulation experiment of this embodiment is as follows: the GPU is NVIDIA GeForce GTX 1060, and the video memory size is 6GB.

[0058] The software platform of the simulation experiment in this embodiment: the operating system is Windows 10.

[0059] 2. Simulation Content

[0060] In order to verify the effectiveness of the algorithm provided by the above embodiment, an automatic modulation recognition simulation experiment was carried out.

[0061] In this embodiment, the data set is trained with the RML2016.10b data set, which contains 10 different types of modulation signals, 8 of which are digital modulation signals, namely (8PSK, BPSK, CPFSK, GFSK, PAM4, 16QAM, 64QAM, QPSK), and 2 are analog modulation signals (AM-DSB, WBFM). Each modulation signal contains 20 (-20:2:18dB) signal-to-noise ratios, and a single modulation signal has 6000 data with a single signal-to-noise ratio, for a total of 1,200,000 data. 60% of the data is selected as the training set, 20% as the validation set, and 20% as the test set.

[0062] 3. Simulation Results

[0063] The simulation results are as follows Figure 3 As shown, the yellow line is the accuracy of the baseline model's signal-to-noise ratio, and the blue line is the accuracy of the model after adopting the method provided by the above embodiment of the present invention. In the low and medium signal-to-noise ratio area, the model accuracy is improved by 1% to 4%, and the overall accuracy is improved from 62.04% to 63.06%, indicating that the method proposed by the present invention has a significant improvement in low signal-to-noise ratio data processing.

[0064] Based on the same inventive concept, the present invention further provides an automatic modulation recognition device based on consistency constraints and feature prototypes, which is used to implement an automatic modulation recognition method based on consistency constraints and feature prototypes provided in the above embodiment of the present invention. The embodiment of the method is referred to above and will not be repeated here; the device comprises:

[0065] A data acquisition module, used for acquiring a modulation signal to be identified;

[0066] A data processing module is used to input the modulation signal to be identified into the trained automatic modulation recognition model, extract the features of the high signal-to-noise ratio data in the modulation signal to be identified, and extract the features of the low signal-to-noise ratio data in the modulation signal to be identified, output the recognition probability, and obtain the classification recognition result of the modulation signal to be identified;

[0067] The trained automatic modulation recognition model is obtained by training the initial automatic modulation recognition model using data of a preset category as a training data set; the data of the preset category includes a modulation signal.

[0068] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the term "include", "comprise" or any other variant is intended to cover non-exclusive inclusion, so that the article or device including a series of elements includes not only those elements, but also other elements that are not explicitly listed. In the absence of more restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the article or device including the elements. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The orientation or position relationship indicated by "up", "down", "left", "right", etc. is based on the orientation or position relationship shown in the accompanying drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention.

[0069] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification.

[0070] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. An automatic modulation recognition method based on consistency constraints and feature prototypes, characterized in that: include: Acquire a modulation signal to be identified; Inputting the modulation signal to be identified into a trained automatic modulation recognition model, extracting features of high signal-to-noise ratio data in the modulation signal to be identified, and extracting features of low signal-to-noise ratio data in the modulation signal to be identified, outputting recognition probability, and obtaining a classification recognition result of the modulation signal to be identified; The trained automatic modulation recognition model is obtained by training an initial automatic modulation recognition model using data of a preset category as a training data set; The preset category of data includes a modulated signal.

2. The automatic modulation recognition method based on consistency constraints and feature prototypes according to claim 1 is characterized in that: Prior to the initial automatic modulation recognition model training, it also includes: Construct an automatic modulation recognition model; where The automatic modulation recognition model includes a feature extractor and a classifier, the feature extractor includes a first convolution block, a second convolution block, a third convolution block, a fourth convolution block, a fifth convolution block, a first timing layer, a second timing layer, a first fully connected layer and a second fully connected layer, and the classifier includes a classification layer; the input end of the first convolution block receives a sine wave and a cosine wave of a signal to be processed, the input end of the second convolution block receives a cosine wave of a signal to be processed, the input end of the third convolution block receives a sine wave of a signal to be processed, and the input end of the fourth convolution block receives a first signal, the first signal being a connection signal of a signal processed by the second convolution block and a signal processed by the third convolution block , the input end of the fifth convolution block receives a second signal, the second signal is a connection signal of the signal processed by the first convolution block and the signal processed by the fourth convolution block, the input end of the first timing layer receives the signal processed by the fifth convolution block, the input end of the second timing layer receives the signal processed by the first timing layer, the input end of the first fully connected layer receives the signal processed by the second timing layer, the input end of the second fully connected layer receives the signal processed by the first fully connected layer, the input end of the classification layer receives the signal processed by the second fully connected layer, and the output end of the classification layer outputs the prediction score of the signal to be processed.

3. The automatic modulation recognition method based on consistency constraints and feature prototypes according to claim 2 is characterized in that: The first convolution block, the second convolution block, the third convolution block, the fourth convolution block and the fifth convolution block have the same structure and all include a convolution layer, a batch normalization layer and a RuLU activation layer; the number of convolution kernels of the convolution layer of the first convolution block is 50, the size of the convolution kernel is 2*8, the step size is 1, and the padding is 0; the number of convolution kernels of the convolution layer of the second convolution block is 50, the size of the convolution kernel is 1*8, the step size is 1, and the padding is 0; the number of convolution kernels of the convolution layer of the third convolution block is 50, the size of the convolution kernel is 1*8, the step size is 1, and the padding is 0; the number of convolution kernels of the convolution layer of the fourth convolution block is 50, the size of the convolution kernel is 1*8, the step size is 1, and the padding is 0; the number of convolution kernels of the convolution layer of the fifth convolution block is 100, the size of the convolution kernel is 2*5, the step size is 1, and the padding is 0; The hidden layer dimension of the first temporal layer is 128, and the hidden layer dimension of the second temporal layer is 128; The output dimension of the first fully connected layer is 128, and the output dimension of the second fully connected layer is 11.

4. The automatic modulation recognition method based on consistency constraints and feature prototypes according to claim 2 is characterized in that: The initial automatic modulation recognition model is trained using the preset category data as the training data set, including: Acquire a training data set, wherein the training data set includes a plurality of training samples, and each of the training samples includes a modulated signal; Labeling the types of training samples in the training data set to obtain true labels of the types of the training samples; The training data set and the true labels of the types of training samples in the training data set are input into the initial automatic modulation recognition model for iterative training until a preset condition is met, thereby obtaining the trained automatic modulation recognition model.

5. The automatic modulation recognition method based on consistency constraints and feature prototypes according to claim 4 is characterized in that: Inputting the training data set and the true labels of the types of training samples in the training data set into the initial automatic modulation recognition model for iterative training, comprising: For the first iteration training, the training data set is input into the automatic modulation recognition model, and the training samples in the training data set with a signal-to-noise ratio greater than or equal to 0 dB and correctly identified are subjected to feature extraction by the feature extractor to obtain a first feature vector of the training process; and a feature prototype of each class of the training samples is obtained according to the first feature vector of each class of the training samples; For the second iteration training, the training data set is input into the automatic modulation recognition model, and the training samples in the training data set with a signal-to-noise ratio less than 0dB or greater than -10dB and correctly identified are subjected to feature extraction by the feature extractor to obtain a second feature vector of the training process; using the feature prototype of each class of the training samples in the first iteration training process, a consistency constraint is imposed on the second feature vector of the second iteration training process to update the parameters of the current automatic modulation recognition model; using the updated parameters of the automatic modulation recognition model, the feature prototype of each class of training samples is recalculated to update the feature prototype of each class of training samples; For the third iterative training, the feature prototypes of each type of training samples updated in the previous iterative training are used to impose consistency constraints on the second feature vectors of the third iterative training process until the preset number of training times is completed.

6. The automatic modulation recognition method based on consistency constraints and feature prototypes according to claim 5, characterized in that: The preset condition includes that the loss function reaches a minimum within a preset number of training times.

7. The automatic modulation recognition method based on consistency constraints and feature prototypes according to claim 6 is characterized in that: The expression of the loss function L1 in the first iteration training is: Among them, Y label Represented as training sample X label The corresponding prediction score is Represents the training sample X label The corresponding true label.

8. The automatic modulation recognition method based on consistency constraints and feature prototypes according to claim 6, characterized in that: The expression of the loss function L2 in the second iteration training is: Among them, f i represents the feature prototype of the training sample, represents the training samples whose signal-to-noise ratio is less than 0 dB and whose true label is the i-th category, n represents the total number of training sample categories, i = 1,…,n, and F(·) represents the feature extractor function.

9. The automatic modulation recognition method based on consistency constraints and feature prototypes according to claim 8, characterized in that: The feature prototype f of the i-th type of training sample i The expression is: Among them, Mean(·) represents the mean function, represents the training samples of the i-th class in the training data set with a signal-to-noise ratio greater than 0 dB and correct identification, and F(·) represents the feature extractor function.

10. An automatic modulation recognition device based on consistency constraints and feature prototypes, characterized in that: include: A data acquisition module, used for acquiring a modulation signal to be identified; A data processing module, used for inputting the modulation signal to be identified into a trained automatic modulation recognition model, performing feature extraction of high signal-to-noise ratio data in the modulation signal to be identified, and performing feature extraction of low signal-to-noise ratio data in the modulation signal to be identified, outputting recognition probability, and obtaining a classification recognition result of the modulation signal to be identified; The trained automatic modulation recognition model is obtained by training an initial automatic modulation recognition model using data of a preset category as a training data set; The preset category of data includes a modulated signal.

Images