CN116319210A - Signal lightweight automatic modulation recognition method and system based on deep learning - Google Patents

Abstract

The invention relates to the technical field of signal processing, in particular to a signal lightweight automatic modulation recognition method and system based on deep learning, which acquire a sample data set containing a plurality of digital modulation modes and a plurality of analog modulation modes by simulating a real communication environment modulation information sequence; constructing a lightweight dense convolution long-short time memory network structure based on a lightweight convolution network, and training and optimizing the lightweight dense convolution long-short time memory network structure by utilizing a sample data set; the signal modulation recognition model is built based on the light-weight dense convolution long-short time memory network structure after training and optimization; inputting the target signal to be identified into a signal modulation identification model, and acquiring a modulation mode of the target signal to be identified by using the signal modulation identification model. The invention solves the problems of high complexity, numerous parameters, huge model and the like in the existing automatic modulation identification, and meets the deployment requirement on resource-limited equipment through the design of a lightweight model.

Description

Signal lightweight automatic modulation recognition method and system based on deep learning

technical field

The invention relates to the technical field of signal processing, in particular to a method and system for lightweight automatic modulation recognition of signals based on deep learning.

Background technique

Automatic Modulation Recognition (AMR) can obtain the modulation style of the signal, which is the prerequisite for completing signal demodulation and then obtaining information. Automatic modulation recognition is a key technology in modern communication systems and has a wide range of applications, such as spectrum interference detection, spectrum sensing, cognitive radio, etc. Researchers have carried out a lot of research in the field of automatic modulation recognition, and proposed various automatic modulation recognition methods. At present, the automatic modulation recognition methods of signals can be roughly divided into likelihood ratio-based automatic modulation recognition method (LB-AMR), feature extraction-based automatic modulation recognition method (FB-AMR) and deep learning-based automatic modulation recognition method (DL-AMR). -AMR). LB-AMR is optimal in the sense of Bayesian estimation, but it relies heavily on prior knowledge and parameter estimation, and its algorithm complexity is high. FB-AMR extracts various manual features through the expert system, such as instantaneous signal amplitude, phase and frequency, constellation diagram, time-frequency distribution characteristics, high-order cumulative quantity, cyclic spectrum, etc., and then artificial neural structure, support vector machine and decision-making Algorithms such as trees are applied to the classification process to improve AMR performance. However, the extraction of differentiated manual features of FB-AMR requires extensive domain knowledge, and the quality of feature extraction directly determines the performance of FB-AMR. DL-AMR is better than FB-AMR and LB-AMR to a certain extent.

However, the current deep learning algorithms are not designed for AMR. Some algorithm frameworks directly borrowed from the fields of image recognition and speech recognition have achieved certain results in the field of AMR, but often have many algorithm parameters and the models are large and complex. On platforms with few available resources or precious resources, such as satellite platforms, due to the constraints of volume, mass, and power consumption, as well as the influence of environmental factors such as space radiation, extreme temperature, and maintenance difficulties, the computing power and storage space of the on-board computer are limited. Compared with the ground computer, there is a very large gap. Therefore, the lightweight and low-complexity DL-AMR method has begun to be studied. How to make DL-AMR reduce the model size or speed up the calculation time while ensuring the recognition accuracy, so that it can be deployed on resource-constrained devices and become a A research and development direction in the direction of signal modulation identification.

Contents of the invention

To this end, the present invention provides a light-weight automatic modulation recognition method and system based on deep learning, which solves the problems of high complexity, numerous parameters, and huge models in existing automatic modulation recognition. Required for deployment on resource-constrained devices.

According to the design scheme provided by the present invention, a light-weight automatic modulation recognition method for signals based on deep learning is provided, including:

Obtain a sample data set including several digital modulation methods and several analog modulation methods by simulating a real communication environment modulation information sequence;

Based on the lightweight convolutional network to build a lightweight dense convolutional long-short-term memory network structure, use the sample data set to train and optimize the lightweight dense convolutional long-short-term memory network structure; and based on the optimized lightweight Dense convolution long short-term memory network structure to establish a signal modulation recognition model;

The target signal to be identified is input into the signal modulation identification model, and the modulation mode of the target signal to be identified is obtained by using the signal modulation identification model.

As the deep learning-based light-weight automatic modulation recognition method of the present invention, further, by simulating the real communication environment modulation information sequence to obtain a sample data set including several digital modulation methods and several analog modulation methods, including:

First, use the random function to randomly generate the 01-bit information sequence of the radio;

Then, use the 01-bit information sequence to simulate the real communication environment modulation information sequence, and generate N digital modulation signals and M analog modulation signals, so as to use the N digital modulation signals and M analog modulation signals to form a sample data set, Wherein, M and N are integers greater than 1, respectively.

As the light-weight automatic modulation recognition method of signals based on deep learning in the present invention, further, the 01-bit information sequence is used to simulate the real communication environment modulation information sequence, and the real communication environment influence parameters are added to the modulation and sampling process of the information sequence, To obtain the IQ sampling sequence of the analog modulated signal, wherein the real communication environment influence parameters include but not limited to: additive Gaussian noise, multipath fading, sampling rate offset and center frequency offset.

As the light-weight automatic modulation recognition method for signals based on deep learning in the present invention, further, N digital modulation signals and M analog modulation signals are used to construct a sample data set, and information points of each modulation signal are set with preset thresholds The number of information points is the sampling interval, and K information points are continuously collected each time to form a signal sample; each modulation signal samples a preset number of samples, and the samples of all modulation signal samples are combined to construct a signal sample data set, where, The preset number is set in tens of thousands.

As the light-weight automatic modulation recognition method of signals based on deep learning in the present invention, further, the light-weight dense convolution long-short-term memory network structure constructed based on light-weight convolution network includes: standardizing the input signal sequence The data input batch normalization processing unit, the densely linked convolution unit that performs convolution and splicing operations on the standardized data, and uses the long short-term memory network to extract the time series features of the spliced data and classify the time series features through the activation function Feature extraction taxa for .

As the light-weight automatic modulation recognition method based on deep learning in the present invention, further, the feature extraction and classification unit is formed by connecting two long-short-term memory network layers and a fully connected layer, and the two long-short-term memory network layers use With different numbers of hidden units, the fully connected layer activation function uses the Softmax function.

As the light-weight automatic modulation recognition method of signals based on deep learning in the present invention, further, when using the sample data set to train and optimize the light-weight dense convolution long-short-term memory network structure, the absolute cross-entropy loss function is set as the training process. The target loss function of , select the Adam optimizer to optimize the network, and set the maximum iteration round or early shutdown as the iteration termination condition for training optimization.

Further, based on the above method, the present invention also provides a signal lightweight automatic modulation recognition system based on deep learning, including: a sample building module, a model training module and a target recognition module, wherein,

The sample building block is used to obtain a sample data set including several digital modulation methods and several analog modulation methods by simulating a real communication environment modulation information sequence;

The model training module is used to build a lightweight dense convolutional long-short-term memory network structure based on a lightweight convolutional network, and uses sample data sets to train and optimize the lightweight dense convolutional long-short-term memory network structure; and based on training Optimized lightweight dense convolutional long short-term memory network structure to establish a signal modulation recognition model;

The target identification module is used to input the target signal to be identified into the signal modulation identification model, and use the signal modulation identification model to obtain the modulation mode of the target signal to be identified.

Beneficial effects of the present invention:

The present invention generates the existing 8 kinds of digital modulation signals and 2 kinds of analog modulation signals by simulating the real communication environment modulation information sequence to construct the sample data set for model training and optimization; Lightweight dense convolutional long-short-term memory network model for modulation recognition, and use sample data sets for training and optimization, use the optimized model after training to identify target signals, and solve the problem of high complexity, numerous parameters, and model problems in existing automatic modulation recognition For problems such as hugeness, the lightweight design of the model can meet the requirements of deployment on resource-constrained devices such as satellite platforms while ensuring the recognition rate, which is convenient for actual scene applications.

Description of drawings:

Fig. 1 is a schematic diagram of the signal light-weight automatic modulation recognition process based on deep learning in an embodiment;

Fig. 2 is a schematic diagram of the construction process of the signal modulation recognition model in the embodiment;

FIG. 3 is a schematic diagram of the verification loss change in the network model training process in the embodiment;

Fig. 4 is a schematic diagram of the signal modulation recognition rate results in the embodiment.

Detailed ways:

In order to make the purpose, technical solution and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and technical solutions.

The embodiment of the present invention, as shown in FIG. 1 , provides a light-weight automatic modulation recognition method for signals based on deep learning, including:

S101. Acquire a sample data set including several digital modulation methods and several analog modulation methods by simulating a real communication environment modulation information sequence.

Specifically, the 01-bit information sequence of the radio can be randomly generated using a random function at first; then, the 01-bit information sequence is used to simulate the real communication environment modulation information sequence, and N types of digital modulation signals and M types of analog modulation signals are generated to utilize the N types of digital modulation signals and M types of analog modulation signals are used to construct the sample data set, where M and N are integers greater than 1, respectively.

Among them, the 01-bit information sequence is used to simulate the real communication environment modulation information sequence, and the real communication environment influence parameter is added to the modulation and sampling process of the information sequence to obtain the IQ sampling sequence of the analog modulation signal, wherein the real communication environment influence parameter Including but not limited to: additive Gaussian noise, multipath fading, sampling rate offset and center frequency offset.

The random function is used to randomly generate 01 bit sequences to ensure the randomness of the information content, and to remove the influence of the signal content on the identification of signal modulation. Simulate the real communication environment to modulate the information sequence, and sample the quadrature in-phase IQ sequence. Add the influence of additive Gaussian noise, multipath fading, sampling rate offset and center frequency offset in the modulation and sampling process of the information sequence, and the actual The environment is similar. Some parameters are as follows: modulation rate 25KBaud/s, sampling rate 200KHz, standard deviation of each sample point in the sampling rate drift process 1Hz, maximum sampling rate offset 50Hz, maximum Doppler frequency used in fading simulation 1Hz, time delay vector [ 1,0.8,0.3], the signal-to-noise ratio is between -20dB and 18dB at 2dB intervals, and the number of sine waves used in the frequency selective fading simulation is 8. The IQ sampling sequence of the modulated signal can be obtained most, that is, each information point has two values, and one symbol has 8 information points.

It should be noted that the specific values of the above parameter settings are not within the scope of protection of the scheme of this case, and the specific values of the parameters can be adjusted according to the actual experimental environment. The above M/N can represent 8 kinds of digital modulation signals and 2 kinds of analog modulation signals, namely 8PSK, BPSK, CPFSK, GFSK, PAM4, 16QAM, 64QAM these 8 digital modulation methods and AM-DSB, WBFM these two Analog modulation method.

As a preferred embodiment, further, use N types of digital modulation signals and M types of analog modulation signals to form a sample data set, and for the information points of each type of modulation signal, take the information points of the preset threshold number as the sampling interval, and each time continuously K information points are collected to form a signal sample; each modulation signal samples a preset number of samples, and the samples of all modulation signal samples are combined to construct a signal sample data set, wherein the preset number is set in tens of thousands.

The information points of each modulation signal take 8 information points as the sampling interval, and 128 information points are continuously collected each time to form a signal sample. Each modulation signal collects 120,000 samples, and all the signals form a signal sample set. From the generated sample set, 60% of each type of modulation signal is extracted to form a training sample set, 20% is extracted from the remaining 40% to form a verification sample set, and the last 20% of the entire sample set is used as a test sample set.

Each modulation signal can collect 120,000 samples, and all signals form a signal sample set. For each modulation signal, the signal is collected at intervals of 2dB before -20dB to 18dB, and 6000 samples are collected for each signal-to-noise ratio. A total of 120,000 samples are collected for each modulation signal. All the collected signals are composed into a signal sample set, with a total of 1.2 million samples.

From the generated sample set, 60% of each type of modulation signal can be randomly selected to form a training sample set, 20% of the remaining 40% can be randomly selected to form a verification sample set, and the remaining 20% of the entire sample set can be used as a test sample set. For the 6000 signal samples under each signal-to-noise ratio under each modulation style, first randomly select 60% to add to the training set, then randomly select 20% from the remaining 40% to add to the verification sample set, and finally add The remaining 20% is added to the test sample set. Referring to Figure 2, the network model is trained, verified and tested by using the training sample set, verification sample set and test sample set respectively, so as to obtain the final model structure for modulation recognition of the target signal through training and tuning.

It should be noted that the above specific numerical values may be set according to empirical values and/or actual use environments, and for those skilled in the art, other specific numerical values may be selected.

S102. Construct a lightweight dense convolutional long-short-term memory network structure based on a lightweight convolutional network, and use the sample data set to train and optimize the lightweight dense convolutional long-short-term memory network structure; Quantitative dense convolution long short-term memory network structure to establish a signal modulation recognition model.

Specifically, the lightweight dense convolutional long-short-term memory network structure based on the lightweight convolutional network includes: a data input batch normalization processing unit for normalizing the input signal sequence; The densely linked convolution unit that performs convolution and splicing operations and the feature extraction classification unit that uses the long short-term memory network to extract the time series features of the spliced data and classify the time series features through the activation function.

The data is input into the batch normalization unit, and the IQ sequence is input into the network in a one-dimensional time series format, and the batch normalization (BN) layer is placed at the head of the network to standardize the input signal and solve the difficulty of model training. Prevent gradient disappearance and internal data distribution offset phenomenon, and strengthen the robustness of the network to various received signals; the BN layer first calculates the average and variance of the training data, and then converts the data to make it conform to the average value of 0 and the standard deviation of The standard normal distribution of 1.

The densely connected convolutional unit can be designed to contain 5 convolutional units (convolutional unit 1, convolutional unit 2, convolutional unit 3, convolutional unit 4, convolutional unit 5) and 5 concatenation layers (Concatenate1, Concatenate2, Concatenate3, Concatenate4, Concatenate5). Among them, the convolution unit is composed of convolution layer, BN layer and ReLU activation function in order to improve the nonlinear mapping ability. The role of the splicing layer is to splice the input according to the dimension.

The connection structure of the densely connected convolution unit can be designed as: BN layer → convolution unit 1 → Concatenate1, BN layer → Concatenate1; Concatenate1 → convolution unit 2 → Concatenate2, Concatenate1 → Concatenate2;

Concatenate2→Convolution Unit 3→Concatenate3, Concatenate2→Concatenate3; Concatenate3→Convolution Unit 4→Concatenate4, Concatenate3→Concatenate4; Concatenate4→Convolution Unit 5→

Concatenate5; BN layer → Concatenate5

The feature extraction classification unit can be sequentially connected by two long short-term memory layers (LSTM1, LSTM2) and a fully connected output layer. The activation function of the fully connected output layer is Softmax.

You can first set the parameters of the 6-layer convolutional layer and the parameters of the 2-layer LSTM layer in the lightweight dense convolutional long-short-term memory network, as well as the target loss function, optimizer, initial learning rate, batch training size, and training rounds And early stop mechanism; use the sample data set to train and optimize the network model after parameter initialization.

The specific parameters of the 6-layer convolutional layer and the 2-layer long-short-term memory layer can be set as follows:

The number of convolution kernels of the one-dimensional convolution layer in convolution unit 1 is 4, and the size of convolution kernels is 3.

The number of convolution kernels of the one-dimensional convolution layer in convolution unit 2 is 4, and the size of convolution kernels is 3.

The number of convolution kernels of the one-dimensional convolution layer in the convolution unit 3 is 4, and the size of the convolution kernels is 5.

The number of convolution kernels of the one-dimensional convolution layer in the convolution unit 4 is 4, and the size of the convolution kernels is 7.

The number of convolution kernels of the one-dimensional convolution layer in the convolution unit 5 is 8, and the size of the convolution kernel is 1.

The number of hidden units of LSTM1 is 32.

The number of hidden units of LSTM2 is 10.

The number of convolution kernels in the fully connected layer is 11, corresponding to the number of output categories.

When using the training set to train the lightweight dense convolution long-short-term memory network, select the Adam optimizer to optimize the network, set the initial learning rate to 0.001, train 512 samples per batch, and the maximum training rounds of the entire training sample is 200 . Use the verification set to verify the model every training round, and use the verification loss as a reference. When the verification loss does not decrease after 50 epochs, stop training the model.

Input the test sample set into the trained lightweight dense convolutional long-short-term memory network to obtain the recognition result, and compare it with the real category to count the recognition accuracy rate; at the same time, record the reasoning of the lightweight dense convolutional long-short-term memory network The time is the sample inference speed to optimize the parameters of the network model.

In model training optimization, the sequence of all samples in the training samples can be disrupted, and the training samples and verification samples can be input into the lightweight dense convolutional long-short-term memory network model to train the lightweight dense convolutional long-short-term memory network. When the maximum number of rounds of network training is reached, or the conditions of the early stopping mechanism are met, the training process of the neural network is completed, and the trained lightweight dense convolution long-short-term memory network model is obtained.

S103. Input the target signal to be identified into the signal modulation identification model, and use the signal modulation identification model to obtain the modulation mode of the target signal to be identified.

Use the trained and optimized signal modulation recognition model to complete the efficient and high-precision recognition of various modulation signals, improve the overall performance of signal modulation recognition, and facilitate the deployment of solutions on application equipment.

Further, based on the above method, the present invention also provides a signal lightweight automatic modulation recognition system based on deep learning, including: a sample building module, a model training module and a target recognition module, wherein,

The sample building block is used to obtain a sample data set including several digital modulation methods and several analog modulation methods by simulating a real communication environment modulation information sequence;

The model training module is used to build a lightweight dense convolutional long-short-term memory network structure based on a lightweight convolutional network, and uses sample data sets to train and optimize the lightweight dense convolutional long-short-term memory network structure; and based on training Optimized lightweight dense convolutional long short-term memory network structure to establish a signal modulation recognition model;

The target identification module is used to input the target signal to be identified into the signal modulation identification model, and use the signal modulation identification model to obtain the modulation mode of the target signal to be identified.

In order to verify the effectiveness of the scheme in this case, the following is a further explanation based on the experimental data:

The simulation experiment is realized on NVIDIA Quadro RTX 6000 and Keras2.6.0Tensorflow-GPU2.4.0 platform, and the simulation experiment of the present invention and the generation of modulation signal and lightweight dense convolution long short-term memory network is completed. Use the steps shown in Figure 2 to complete the experiment, and obtain the variation trend of the verification loss during the training process of the lightweight dense convolution long short-term memory network to verify the signal recognition rate and model reasoning speed of this case.

Figure 3 shows the variation of the verification loss during the training process of the lightweight dense convolution long-short-term memory network; from Figure 3, it can be seen that the verification loss decreases and converges and is stable, indicating that the training effect of the simulation experiment gradually improves with the number of training times. Fig. 4 is the result diagram of the recognition rate of the simulation experiment of this case scheme. It can be seen from the figure that the recognition rate gradually increases and stabilizes with the increase of the signal-to-noise ratio, and the highest can reach 93.7%. At the same time, the model reasoning speed is recorded for each Sample inference time 0.024 ms.

From the above simulation experiments, it can be shown that for the automatic identification of modulated signals, the scheme of this case can efficiently and accurately complete the task of automatic modulation identification, and the feasibility of the scheme has been further verified.

Relative steps, numerical expressions and numerical values of components and steps set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

Each embodiment in this specification is described in a progressive manner, each embodiment focuses on the difference from other embodiments, and the same and similar parts of each embodiment can be referred to each other. As for the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and for the related information, please refer to the description of the method part.

The units and method steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, in the above description The composition and steps of each example have been generally described in terms of functions. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution. Those of ordinary skill in the art may use different methods to implement the described functions for each particular application, but such implementation is not considered to exceed the scope of the present invention.

Those of ordinary skill in the art can understand that all or part of the steps in the above method can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium, such as: a read-only memory, a magnetic disk or an optical disk, and the like. Optionally, all or part of the steps in the above embodiments can also be implemented using one or more integrated circuits. Correspondingly, each module/unit in the above embodiments can be implemented in the form of hardware, or can be implemented in the form of software function modules. The form is realized. The present invention is not limited to any specific combination of hardware and software.

Finally, it should be noted that: the above-described embodiments are only specific implementations of the present invention, used to illustrate the technical solutions of the present invention, rather than limiting them, and the scope of protection of the present invention is not limited thereto, although referring to the foregoing The embodiment has described the present invention in detail, and those of ordinary skill in the art should understand that any person familiar with the technical field can still modify the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention Changes can be easily thought of, or equivalent replacements are made to some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of the present invention within the scope of protection. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (10)

1. The signal lightweight automatic modulation recognition method based on deep learning is characterized by comprising the following steps of:

acquiring a sample data set containing a plurality of digital modulation modes and a plurality of analog modulation modes by simulating a real communication environment modulation information sequence;

constructing a lightweight dense convolution long-short time memory network structure based on a lightweight convolution network, and training and optimizing the lightweight dense convolution long-short time memory network structure by utilizing a sample data set; the signal modulation recognition model is built based on the light-weight dense convolution long-short time memory network structure after training and optimization;

inputting the target signal to be identified into a signal modulation identification model, and acquiring a modulation mode of the target signal to be identified by using the signal modulation identification model.

2. The deep learning-based signal lightweight automatic modulation recognition method according to claim 1, wherein obtaining a sample data set including a plurality of digital modulation schemes and a plurality of analog modulation schemes by simulating a real communication environment modulation information sequence comprises:

firstly, randomly generating a 01 bit information sequence of a radio by utilizing a random function; the method comprises the steps of carrying out a first treatment on the surface of the

Then, the real communication environment modulation information sequence is simulated using the 01-bit information sequence, and N digital modulation signals and M analog modulation signals are generated to construct a sample data set using the N digital modulation signals and the M analog modulation signals, wherein M, N is an integer greater than 1, respectively.

3. The deep learning-based signal lightweight automatic modulation recognition method according to claim 2, wherein a 01-bit information sequence is utilized to simulate a real communication environment modulation information sequence, and real communication environment influence parameters are added in the process of modulating and sampling the information sequence to obtain an IQ sampling sequence of a simulated modulation signal, wherein the real communication environment influence parameters include, but are not limited to: additive gaussian noise, multipath fading, sample rate offset, and center frequency offset.

4. The deep learning-based signal lightweight automatic modulation recognition method according to claim 2, wherein a sample data set is constructed by using N digital modulation signals and M analog modulation signals, and for information points of each modulation signal, K information points are continuously acquired at each time by taking a preset threshold number of information points as sampling intervals to form a signal sample; each modulation signal samples a preset number of samples, and the samples sampled by all the modulation signals are combined to construct a signal sample data set, wherein the preset number is set in units of ten thousands.

5. The deep learning-based signal lightweight automatic modulation recognition method according to claim 1, wherein the lightweight dense convolution long-short time memory network structure constructed based on the lightweight convolution network comprises: the method comprises the steps of inputting data subjected to standardization processing to a batch normalization processing unit, carrying out convolution and splicing operation on the standardized data, and extracting time sequence features of the spliced data by using a long-short-time memory network and classifying the time sequence features by activating a function.

6. The deep learning-based signal lightweight automatic modulation recognition method according to claim 5, wherein the feature extraction classification unit is formed by connecting two long-short-time memory network layers and a full-connection layer, the two long-short-time memory network layers adopt different numbers of hidden units, and the full-connection layer activation function adopts a Softmax function.

7. The deep learning-based signal lightweight automatic modulation recognition method according to claim 1, 5 or 6, wherein when training and optimizing a lightweight dense convolution long-short-time memory network structure by using a sample data set, setting an absolute cross entropy loss function as a target loss function in a training process, selecting an Adam optimizer to optimize the network, and setting the maximum iteration round or early shutdown to manufacture as an iteration termination condition of training and optimizing.

8. A deep learning-based signal lightweight automatic modulation recognition system, comprising: a sample construction module, a model training module and a target recognition module, wherein,

the sample construction module is used for acquiring a sample data set containing a plurality of digital modulation modes and a plurality of analog modulation modes by simulating a real communication environment modulation information sequence;

the model training module is used for constructing a lightweight dense convolution long-short time memory network structure based on the lightweight convolution network, and training and optimizing the lightweight dense convolution long-short time memory network structure by utilizing the sample data set; the signal modulation recognition model is built based on the light-weight dense convolution long-short time memory network structure after training and optimization;

the target recognition module is used for inputting the target signal to be recognized into the signal modulation recognition model, and the modulation mode of the target signal to be recognized is obtained by utilizing the signal modulation recognition model.

9. An electronic device comprising a memory and a processor, said processor and said memory completing communication with each other via a bus; the memory stores program instructions executable by the processor, the processor invoking the program instructions to perform the method steps of any of claims 1-7.

10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored therein a computer program which, when executed by a processor, implements the method steps of any of claims 1-7.

Images