WO2021082469A1

WO2021082469A1 - Modulation mode identification method and apparatus

Info

Publication number: WO2021082469A1
Application number: PCT/CN2020/096565
Authority: WO
Inventors: 冯志勇; 张克终; 尉志青
Original assignee: 北京邮电大学
Priority date: 2019-10-28
Filing date: 2020-06-17
Publication date: 2021-05-06
Also published as: CN110798419A

Abstract

The embodiments of the present application provide a modulation mode identification method and apparatus, relating to the technical field of radio communications. Said method comprises: an electronic device being able to receive a signal to be identified, and determine, according to frequency components and cyclic frequency components of said signal, a cyclic spectrum of said signal and use same as a target cyclic spectrum; and then, input the target cyclic spectrum into a pre-trained modulation mode identification model, so as to obtain a modulation mode of said signal. By means of the present application, the cyclic spectrum of a signal to be identified can be completely inputted into the modulation mode identification model, so as to identify the modulation mode of said signal, that is, the modulation mode of said signal is identified by means of all the features of the cyclic spectrum of said signal, rather than some features of the cyclic spectrum, so that the accuracy of signal modulation mode identification is improved.

Description

Method and device for identifying modulation mode

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on October 28, 2019 with the application number 201911028838.0 and the invention title is "a modulation method and device for identification", the entire content of which is incorporated into this application by reference .

Technical field

This application relates to the field of radio communication technology, and in particular to a modulation method identification method and device.

Background technique

In the non-cooperative mode, that is, when the modulation method of the signal is unknown, the modulation method identification technology of the signal (especially the weak signal) has important applications in many fields. The purpose of the signal modulation method identification technology is to identify the modulation method of the received signal in the absence of transmitter information and channel information. The existing signal modulation method identification methods mainly include: calculating the cyclic spectrum of the signal to be identified, obtaining the cyclic spectrum of the signal to be identified, extracting the positions of partial peaks of the cyclic spectrum of the signal to be identified, and/or the signal to be identified Use the extracted features to identify the modulation mode of the signal to be identified by using features such as the cross-sectional information at the peak of the cyclic spectrogram.

However, when using the above modulation recognition method to recognize the modulation mode of the signal to be identified, the extracted features are only part of the cyclic spectrogram, that is, only part of the cyclic spectrogram is used to recognize the modulation mode of the signal to be recognized. This makes the recognition accuracy of the modulation mode of the signal low.

Summary of the invention

The purpose of the embodiments of the present application is to provide a modulation method identification method and device, so as to improve the accuracy of signal modulation identification. The specific technical solutions are as follows:

In the first aspect, in order to achieve the foregoing objective, an embodiment of the present application provides a modulation mode identification method, and the method includes the following steps:

Receive the signal to be identified;

Determine the cyclic spectrum of the signal to be identified according to the frequency components and the cyclic frequency components of the signal to be identified, as the target cyclic spectrum;

Input the target cyclic spectrogram into a pre-trained modulation mode recognition model to obtain the modulation mode of the signal to be recognized, wherein the modulation mode recognition model is a model obtained by training a deep neural network using sample data, and The sample data includes multiple sample signals and the label modulation mode of each sample signal.

Optionally, the step of determining the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram according to the frequency components and the cyclic frequency components of the signal to be identified includes:

Moving the frequency spectrum of the signal to be identified to a plurality of preset frequency bands to obtain signals of multiple frequency bands corresponding to the signal to be identified;

Determine the cyclic spectrogram of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal, and the cyclic spectrogram of each frequency band signal is the cyclic spectrogram corresponding to the frequency band where the signal of the frequency band is located;

Performing superposition processing on the cyclic spectrograms corresponding to the multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram.

Optionally, the step of moving the frequency spectrum of the signal to be identified to a plurality of preset frequency bands to obtain signals of multiple frequency bands corresponding to the signal to be identified includes:

Performing segmentation processing on the signal to be identified to obtain a plurality of sub-signals to be identified;

For each sub-signal to be identified, the frequency spectrum of the sub-signal to be identified is moved to multiple preset frequency bands to obtain multiple frequency band sub-signals corresponding to the sub-signal to be identified; The multiple frequency band sub-signals corresponding to the sub-signal to be identified are frequency band signals corresponding to the signal to be identified in the frequency band;

The step of determining the cyclic spectrum of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal includes:

Determine the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal;

For each frequency band, averaging is performed on the cyclic spectrograms of multiple frequency band sub-signals on the frequency band to obtain the cyclic spectrogram of the frequency band signals on the frequency band.

Optionally, the step of determining the cyclic spectrum of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal includes:

For each frequency band signal, perform segmentation processing on the frequency band signal to obtain multiple frequency band sub-signals corresponding to the frequency band signal;

Optionally, the step of determining the cyclic spectrogram of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal includes:

Fast Fourier transform is performed on each frequency band sub-signal to obtain the transformed sub-signal corresponding to each frequency band sub-signal;

According to each frequency component and each cyclic frequency component of each transformed sub-signal, the cyclic spectrum of each transformed sub-signal is determined as the cyclic spectrum of the frequency band sub-signal corresponding to each transformed sub-signal.

Optionally, the step of determining the cyclic spectrogram of each transformed sub-signal according to each frequency component and each cyclic frequency component of each transformed sub-signal includes:

Use the following formula to determine the cyclic spectrum of each transformed sub-signal:

among them,

Represents the cyclic spectrum of the k-th transformed sub-signal, N ₀ represents the signal length of the k-th transformed sub-signal, f represents the frequency component, and α represents the cyclic frequency component,

Expressed in

The k-th transformed sub-signal at frequency,

Means

The conjugate,

Expressed in

For the k-th transformed sub-signal under frequency, |·| represents the calculation of the modulus value.

Optionally, the step of performing superposition processing on the cyclic spectrograms corresponding to the multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram includes:

Use the following formula to superimpose the cyclic spectra corresponding to multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram:

|S′ _r (f,α)|=max{|S _r,1 (f,α)|,|S _r,2 (f,α)|,…,|S _r,n (f,α)| };

Where |S' _r (f,α)| represents the cyclic spectrum of the signal to be identified, |S _r,l (f,α)| represents the cyclic spectrum corresponding to the l-th frequency band, l represents the frequency band, and l =1,2,...n, max represents the maximum value, n represents the number of the multiple frequency bands, f represents the frequency component, α represents the cyclic frequency component, and |·| represents the calculation of the modulus value.

Optionally, the modulation mode recognition model is obtained by training in the following manner:

Acquiring the sample data, where the sample data includes a plurality of sample signals and a label modulation mode of each sample signal;

Determine the cyclic spectrum of each sample signal according to each frequency component and each cyclic frequency component of each sample signal;

Input the cyclic spectrogram of each sample signal into the preset deep neural network to obtain the predicted modulation mode of each sample signal;

Determine the loss value of the modulation mode recognition according to the predicted modulation mode and the labeled modulation mode of each sample signal;

Determine whether the deep neural network converges according to the loss value;

If not, adjust the parameters of the deep neural network and return to the step of inputting the cyclic spectrogram of each sample signal into the preset deep neural network to obtain the predicted modulation mode of each sample signal;

If yes, it is determined that the current deep neural network is the modulation mode recognition model.

In the second aspect, in order to achieve the foregoing objective, an embodiment of the present application provides a modulation mode identification device, the device including:

The receiving module is used to receive the signal to be identified;

The determining module is configured to determine the cyclic spectrum of the signal to be identified according to the respective frequency components and the respective cyclic frequency components of the signal to be identified, as the target cyclic spectrum;

The recognition module is used to input the target cyclic spectrogram into a pre-trained modulation mode recognition model to obtain the modulation mode of the signal to be recognized, wherein the modulation mode recognition model is obtained by training a deep neural network using sample data In the model, the sample data includes a plurality of sample signals and the label modulation mode of each sample signal.

Optionally, the determining module includes:

The moving sub-module is used to move the frequency spectrum of the signal to be identified to multiple preset frequency bands respectively to obtain multiple frequency band signals corresponding to the signal to be identified;

The first determining sub-module is used to determine the cyclic spectrum of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal, and the cyclic spectrum of each frequency band signal is the cycle corresponding to the frequency band where the signal of the frequency band is located. Spectrogram

The superimposition sub-module is used to perform superposition processing on the cyclic spectrograms corresponding to the multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram.

Optionally, the moving sub-module is specifically configured to: perform segment processing on the signal to be identified to obtain multiple sub-signals to be identified; for each sub-signal to be identified, move the frequency spectrum of the sub-signal to be identified respectively To multiple preset frequency bands, multiple frequency band sub-signals corresponding to the sub-signals to be identified are obtained; wherein, the multiple frequency band sub-signals corresponding to the multiple sub-signals to be identified in one frequency band are the to-be-identified sub-signals in the frequency band. Identify the frequency band signal corresponding to the signal;

The first determining sub-module is specifically configured to: determine the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal; for each frequency band, multiple The cyclic spectrum of the frequency band sub-signal is averaged, and the cyclic spectrum of the frequency band signal on the frequency band is obtained.

Optionally, the first determining submodule is specifically configured to: for each frequency band signal, perform segmentation processing on the frequency band signal to obtain multiple frequency band sub-signals corresponding to the frequency band signal; Frequency components and each cyclic frequency component, determine the cyclic spectrum of each frequency band sub-signal; for each frequency band, perform averaging processing on the cyclic spectrum of multiple frequency band sub-signals on the frequency band to obtain the cyclic spectrum of the frequency band signal on the frequency band Spectrogram.

Optionally, the first determining submodule is specifically configured to:

among them,

Expressed in

The k-th transformed sub-signal at frequency,

Means

The conjugate,

Expressed in

Optionally, the superposition submodule is specifically used for:

|S′ _r (f,α)|=max{|S _r,1 (f,α)|,|S _r,2 (f,α)|,…,|S _r,n (f,α)| };

Optionally, the device further includes: a training module for training to obtain the modulation mode recognition model;

The training module includes:

An obtaining sub-module, configured to obtain the sample data, the sample data including a plurality of sample signals and the labeling modulation mode of each sample signal;

The second determining sub-module is used to determine the cyclic spectrum of each sample signal according to each frequency component and each cyclic frequency component of each sample signal;

The prediction sub-module is used to input the cyclic spectrogram of each sample signal into the preset deep neural network to obtain the predicted modulation mode of each sample signal;

The third determining sub-module is used to determine the loss value of the modulation mode recognition according to the predicted modulation mode and the labeling modulation mode of each sample signal;

The fourth determination sub-module is used to determine whether the deep neural network converges according to the loss value; if not, adjust the parameters of the deep neural network, and return to execute the input of the cyclic spectrum of each sample signal The preset deep neural network obtains the step of predicting the modulation mode of each sample signal; if it is, the current deep neural network is determined to be the modulation mode recognition model.

In a third aspect, an electronic device is provided, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

Memory, used to store computer programs;

The processor is configured to implement any of the method steps described in the first aspect when executing the program stored in the memory.

In a fourth aspect, a computer-readable storage medium is provided, and a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, it implements any of the method steps described in the first aspect.

In a fifth aspect, a computer program is provided, which when running on a computer, causes the computer to execute any method step described in the first aspect.

In the modulation mode identification method and device provided by the embodiments of the present application, the electronic device inputs the complete cyclic spectrum of the signal to be identified into the modulation mode identification model to identify the modulation mode of the signal to be identified, that is, using the Recognizing all the characteristics of the cyclic spectrogram of the signal and identifying the modulation mode of the signal to be identified, instead of using part of the characteristics of the cyclic spectrogram to identify the modulation mode of the signal to be recognized, the accuracy of the recognition of the modulation mode of the signal is improved.

Of course, implementing any product or method of the present application does not necessarily need to achieve all the advantages described above at the same time.

Description of the drawings

In order to more clearly describe the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work.

FIG. 1 is a schematic flowchart of a modulation method identification method provided by an embodiment of this application;

2 is a schematic diagram of another flow chart of a modulation mode identification method provided by an embodiment of this application;

FIG. 3 is a schematic diagram of a central frequency point increase provided by an embodiment of the application;

FIG. 4 is a schematic diagram of a central frequency point reduction provided by an embodiment of the application;

FIG. 5 is a schematic diagram of a smoothing process provided by an embodiment of this application;

FIG. 6 is a schematic diagram of a picture superimposing process provided by an embodiment of this application;

FIG. 7 is a schematic flowchart of a deep neural network training method provided by an embodiment of this application;

FIG. 8 is a schematic diagram of a structure of a deep neural network provided by an embodiment of this application;

FIG. 9 is a schematic diagram of still another flow chart of a modulation mode identification method provided by an embodiment of this application;

FIG. 10 is a schematic structural diagram of a modulation mode identification device provided by an embodiment of this application;

FIG. 11 is a schematic diagram of a structure of an electronic device provided by an embodiment of the application.

Detailed ways

The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of this application.

For ease of understanding, the words appearing in the embodiments of the present application are explained below.

Signal in cooperative mode: refers to a signal whose modulation method and channel condition are known.

Signal in non-cooperative mode: Refers to a signal whose modulation method and channel condition are unknown.

Cyclic spectrum: Use frequency component as abscissa, cyclic frequency component as ordinate, or frequency component as abscissa and cyclic frequency component as ordinate to determine the cyclic spectrum of the signal. The cyclic spectrum is complex.

Cyclic spectrum: Take the modulus of the cyclic spectrum to obtain the cyclic spectrum.

The following will describe in detail a modulation method identification method provided in an embodiment of the present application in conjunction with specific implementations, as shown in FIG. 1, and the specific steps are as follows:

Step 101: Receive a signal to be identified.

In the embodiment of the present application, when the modulation mode of the signal is recognized, the electronic device may receive a signal with an unknown modulation mode as the signal to be recognized.

Step 102: Determine a cyclic spectrogram of the signal to be recognized according to each frequency component and each cyclic frequency component of the signal to be recognized, as the target cyclic spectrogram.

In the embodiment of the present application, after the electronic device receives the signal to be identified, it can determine the cyclic spectrum of the signal to be identified based on the frequency components and the cyclic frequency components of the signal to be identified, and the cyclic spectrum of the signal to be identified Figure as the target cycle spectrum.

In the embodiment of the present application, the cyclic spectrum of the signal to be identified is determined according to the frequency components and the cyclic frequency components of the signal to be identified, which can be understood as taking the frequency component as the abscissa and the cyclic frequency component as the ordinate, or Using the frequency component as the abscissa and the cyclic frequency component as the ordinate, the cyclic spectrum of the signal to be identified is determined.

Step 103: Input the target cyclic spectrogram into the pre-trained modulation mode recognition model to obtain the modulation mode of the signal to be recognized.

Wherein, the modulation recognition model is a model obtained by training a deep neural network using sample data, and the sample data includes a plurality of sample signals and a label modulation method of each sample signal.

In the embodiment of the present application, the electronic device may input the target cyclic spectrogram into the pre-trained modulation mode recognition model to obtain the modulation mode of the signal to be recognized. The modulation recognition model is a model obtained by training a deep neural network using sample data. The sample data may include multiple sample signals and a label modulation method for each sample signal. The specific training method of the modulation mode recognition model will be described in detail below, and the description will not be expanded here.

In one embodiment, the electronic device may first move the received frequency spectrum of the signal to be identified to multiple preset frequency bands. Specifically, referring to FIG. 2, FIG. 2 is a schematic diagram of another flow chart of a modulation mode identification method provided by an embodiment of the present application. The method may include the following steps.

Step 201: Receive a signal to be identified. Step 201 is consistent with step 101.

Step 202: Move the frequency spectrum of the signal to be identified to multiple preset frequency bands to obtain multiple frequency band signals corresponding to the signal to be identified.

In the embodiment of the present application, after receiving the signal to be identified, the electronic device performs spectrum shift processing on the signal to be identified. Specifically, it may be: moving the received frequency spectrum of the signal to be identified to multiple preset frequency bands, that is, changing the center frequency point f _c of the signal to be identified to obtain multiple frequency band signals corresponding to the signal to be identified.

The center frequency f _{c of} each frequency band is different. Step 202 as will be understood, to be the center frequency of the identification signal f _c is adjusted to a predetermined center frequency f _c, the identification signal to be amplitude modulated carrier signal, the frequency spectrum of the signal to be recognized without distortion to move the carrier The two sides of the preset center frequency point f _c of the signal are to adjust the signal to be identified to the frequency band where the _{preset center frequency point f c is located.} The signal obtained after the signal to be identified is adjusted to the preset frequency band is the frequency band signal.

In one embodiment, in step 202 above, the electronic device may directly move the frequency spectrum of the signal to be identified to multiple preset frequency bands to obtain signals in multiple frequency bands corresponding to the signal to be identified.

In another embodiment, the electronic device may use the following steps to obtain multiple frequency band signals corresponding to the signal to be identified. Specifically, in the above step 202, moving the frequency spectrum of the signal to be identified to multiple preset frequency bands to obtain multiple frequency band signals corresponding to the signal to be identified may include:

After receiving the signal to be identified, the electronic device may perform segmentation processing on the signal to be identified to obtain multiple sub-signals to be identified. For each sub-signal to be identified, the electronic device moves the frequency spectrum of the sub-signal to be identified to a plurality of preset frequency bands to obtain multiple frequency band sub-signals corresponding to the sub-signal to be identified. The multiple frequency band sub-signals corresponding to the multiple sub-signals to be identified in one frequency band are frequency band signals corresponding to the signal to be identified in the frequency band.

In the embodiment of the present application, a frequency band includes multiple frequency band sub-signals, and each frequency band sub-signal corresponds to one sub-signal to be identified, and the multiple sub-signals to be identified do not overlap, that is, multiple frequency band sub-signals corresponding to the multiple sub-signals to be identified The signals do not overlap. Multiple sub-signals of multiple frequency bands on a frequency band correspond to multiple sub-signals to be identified, and these multiple sub-signals to be identified form the signal to be identified. Correspondingly, multiple sub-signals of the frequency band constitute the frequency band signal corresponding to the signal to be identified on the frequency band. .

Step 203: Determine the cyclic spectrum of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal.

In the embodiment of the present application, the electronic device performs spectrum shift processing on the signal to be identified to obtain a frequency band signal, and performs calculation cyclic spectrogram processing on the frequency band signal to obtain a cyclic spectrogram of each frequency band signal. Specifically, it may be: determining the cyclic spectrum of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal. Step 203 can be understood as taking the frequency component as the abscissa and the cyclic frequency component as the ordinate, or taking the frequency component as the abscissa and the cyclic frequency component as the ordinate to determine the cyclic spectrum of each frequency band signal. Among them, for each frequency band signal, the cyclic spectrogram of the frequency band signal is the cyclic spectrogram corresponding to the frequency band where the frequency band signal is located.

In one embodiment, the electronic device moves the frequency spectrum of the signal to be identified to a plurality of preset frequency bands, and after obtaining multiple frequency band signals corresponding to the signal to be identified, for each frequency band, the electronic device can be based on the frequency band on the frequency band. Each frequency component of the signal and each cyclic frequency component determine the cyclic spectrum of the frequency band signal.

In another embodiment, the electronic device moves the frequency spectrum of the signal to be identified to a plurality of preset frequency bands, and after obtaining multiple frequency band signals corresponding to the signal to be identified, for each frequency band, the electronic device can be based on the signal in the frequency band. Perform segmentation processing to obtain multiple frequency band sub-signals corresponding to the frequency band signal; determine the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal. For each frequency band, the electronic device smoothes the cyclic spectrum of the sub-signal of each frequency band to obtain the cyclic spectrum of the frequency band signal in the frequency band.

The foregoing smoothing processing may be averaging processing, that is, performing averaging processing on the cyclic spectrograms of multiple frequency band sub-signals on the frequency band to obtain the cyclic spectrogram of the frequency band signals on the frequency band.

The above steps of determining the cyclic spectrum of each frequency sub-signal according to the respective frequency components and the respective cyclic frequency components of each frequency band sub-signal can be understood as taking the frequency component as the abscissa and the cyclic frequency component as the ordinate, or taking The frequency component is the abscissa, the cyclic frequency component is the ordinate, and the cyclic spectrum of each frequency band signal is drawn.

In another embodiment, the electronic device may first identify the signal for segmentation processing, and then perform spectrum shifting. As described above, the frequency spectrum of each sub-signal to be identified is moved to a plurality of preset frequency bands, and multiple frequency band sub-signals corresponding to each sub-signal to be identified are obtained. At this time, on each frequency band, the electronic device obtains multiple frequency band sub-signals. The electronic device can determine the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal. For each frequency band, the electronic device smoothes the cyclic spectrum of the sub-signal of each frequency band to obtain the cyclic spectrum of the frequency band signal in the frequency band.

Signals moved to different frequency bands have different corresponding cyclic spectra. When the center frequency point f _c changes, its cyclic spectrum will also change. As shown in Figure 3, when the center frequency f _c increases, the distance between the peaks of the corresponding cyclic spectrum will also increase; on the contrary, as shown in Figure 4, when the center frequency f _c decreases, the corresponding The distance between the peaks of the cyclic spectrum will also decrease. Smoothing the cyclic spectrograms of sub-signals of multiple frequency bands in one frequency band can minimize glitches caused by noise in the cyclic spectrogram.

In an embodiment, the above step of determining the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal may include:

The electronic device can perform fast Fourier transform on the sub-signals of each frequency band to obtain the transform sub-signals corresponding to the sub-signals of each frequency band. After that, the electronic device can determine the cyclic spectrogram of each transformed sub-signal according to each frequency component and each cyclic frequency component of each transformed sub-signal, and use the cyclic spectrogram of each transformed sub-signal as each transformed sub-signal The cyclic spectrum of the corresponding frequency band sub-signal.

In the embodiment of this application, the step of determining the cyclic spectrum of each transformed sub-signal according to each frequency component and each cyclic frequency component of each transformed sub-signal can be understood as taking the frequency component as the abscissa and the cyclic frequency The component is the ordinate, or the frequency component is the abscissa, the cyclic frequency component is the ordinate, and the cyclic spectrum of each transformed signal is determined.

In one embodiment, the electronic device can use the following formula to determine the cyclic spectrum of each transformed sub-signal:

among them,

Represents the cyclic spectrum of the k-th transformed sub-signal,

Represents the cyclic spectrum of the k-th transformed sub-signal, r represents the signal to be identified, N ₀ represents the signal length of the k-th transformed sub-signal corresponding to the signal to be identified, f represents the frequency component, and α represents the cyclic frequency component,

Expressed in

The k-th transformed sub-signal at frequency,

Means

The conjugate,

Expressed in

In an embodiment, for each frequency band, the electronic device may use the following formula to perform averaging processing on the cyclic spectrograms of the multiple frequency band sub-signals on the frequency band to obtain the cyclic spectrogram of the frequency band signal on the frequency band:

Among them, |S _r,l (f,α)| represents the cyclic spectrogram of the frequency band signal on the lth frequency band, N _l represents the total number of cyclic spectrograms corresponding to the lth frequency band, that is, the sub-signal of the frequency band on the lth frequency band R represents the signal to be identified, m represents the frequency band sub-signal corresponding to the signal to be identified, l represents the frequency band, |S _r,l,m (f,α)| represents the m-th frequency band sub-signal on the l-th frequency band The cyclic spectrum of S _r,l,m (f,α) represents the cyclic spectrum of the m-th frequency band sub-signal on the l-th frequency band, f represents the frequency component, α represents the cyclic frequency component, and |·| represents the modulus value Calculation.

As shown in Figure 5, the figure includes _{the cyclic spectrograms of the first, second, and N l-} th frequency band sub-signals on the _{l-th frequency band, and the above N l} cyclic spectrograms are averaged to obtain the graph 5 Means shown below. Based on this, the electronic device performs averaging processing on the cyclic spectrograms of multiple frequency band sub-signals in each frequency band, so that the glitch caused by noise in the cyclic spectrogram of the frequency band signal in each frequency band can be minimized. Reduce the influence of noise interference.

Step 204: Perform superposition processing on the cyclic spectrograms corresponding to the multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram.

In the embodiment of the present application, after obtaining the cyclic spectrogram corresponding to each frequency band, the electronic device performs image superposition processing on the cyclic spectrogram corresponding to the multiple frequency bands to obtain the target cyclic spectrogram. Specifically, it can be: superimposing the cyclic spectrograms corresponding to multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram.

In one embodiment, the electronic device may use the following formula to superimpose the cyclic spectrograms corresponding to multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram:

|S′ _r (f,α)|=max{|S _r,1 (f,α)|,|S _r,2 (f,α)|,…,|S _r,n (f,α)| };

Among them, |S' _r (f,α)| represents the cyclic spectrum of the signal to be identified, _S'r (f,α) represents the cyclic spectrum of the signal to be identified, and |S _r,l (f,α)| represents the first The cyclic spectrum corresponding to l frequency bands, S _r,l (f,α) represents the cyclic spectrum corresponding to the l-th frequency band, l represents the frequency band, l=1, 2,...n, max can represent the maximum value, n Represents the number of multiple frequency bands, f represents the frequency component, α represents the cyclic frequency component, and |·| represents the calculation of the modulus value.

As shown, in FIG comprises a central frequency f _c to be the identification signal 6 to move at 0.1f _c circulation spectrum, the signal to be recognized center frequency f _c to move at 0.3f _c circulation spectrum, to be The center frequency point f _{c of the} identification signal is moved to _{the cyclic spectrogram at 1.5f c} , etc., and the cyclic spectrograms corresponding to these different frequency bands are superimposed, and the cyclic spectrogram after the superimposition of the pictures shown in Figure 6 is obtained. , Which is the target cycle spectrum.

Based on this, the electronic device can superimpose the cyclic spectrograms corresponding to multiple frequency bands, that is, the electronic device performs the superposition processing on the cyclic spectrograms after the averaging process, that is, obtains the cyclic spectrograms of multiple frequency bands. The processing corresponding to the maximum value of the position increases the number of peaks of the target cyclic spectrogram finally obtained, that is, increases the characteristics of the modulation mode identification. Therefore, the modulation method identification based on the target cyclic spectrum with an increased number of peaks can further improve the accuracy of the modulation method identification.

Step 205: Input the target cyclic spectrum into the pre-trained modulation mode recognition model to obtain the modulation mode of the signal to be recognized. Step 205 is consistent with step 103.

In an embodiment, referring to FIG. 7, the electronic device may train to obtain the above-mentioned modulation mode recognition model through the following steps, which are specifically as follows.

Step 701: Obtain sample data, where the sample data includes a plurality of sample signals and the label modulation mode of each sample signal.

In the embodiment of the present application, the electronic device may obtain sample data, and the sample data may include a plurality of sample signals and the labeling modulation mode of each sample signal. Wherein, the modulation mode of the above-mentioned multiple sample signals is known, that is, the modulation mode is marked. For example, the modulation mode of the a-sample signal is quadrature amplitude modulation, and the modulation mode of the b-sample signal is key shift modulation.

In order to improve the accuracy of the modulation method recognition model obtained by the training training, the more sample signals included in the acquired sample data, the better.

Step 702: Determine a cyclic spectrum of each sample signal according to each frequency component and each cyclic frequency component of each sample signal.

In the embodiment of the present application, after acquiring multiple sample signals, the electronic device may determine the cyclic spectrum of each sample signal according to each frequency component and each cyclic frequency component of each sample signal. That is, taking the frequency component as the abscissa and the cyclic frequency component as the ordinate, or the frequency component as the abscissa and the cyclic frequency component as the ordinate, the cyclic spectrum of each sample signal is drawn.

Among them, the frequency component and the cyclic frequency component of each sample signal can be arbitrarily selected within a certain range of values, for example, the value range of the frequency component is 100 Hz to 1000 MHz, and the value range of the cyclic frequency component is 2000 Hz to 2000 Hz. 10000 Hz.

Step 703: Input the cyclic spectrogram of each sample signal into a preset deep neural network to obtain the predicted modulation mode of each sample signal.

In the embodiment of the present application, after determining the cyclic spectrogram of each sample signal, the electronic device can input the cyclic spectrogram into a preset deep neural network to obtain the predicted modulation mode of each sample signal. Among them, the above-mentioned deep neural network can choose ResNet50.

In one embodiment, the above-mentioned deep neural network may be ResNet50, which introduces the concept of residual error, and avoids the phenomenon of gradient dispersion and gradient disappearance in the deep neural network during the training phase. For the specific structure of ResNet50, please refer to Fig. 8. The ResNet50 shown in Fig. 8 may include 1 input layer, (3+4+6+3)*3 hidden layers and one output layer, totaling 50 layers. In addition, the above ResNet50 can be built in Pytorch, where Pytorch is a free and open source deep neural network framework.

In Figure 8, the top 7×7 convolution kernel, 64 channels and a step size of 2,2×2 pooling layer rectangular box represents the input layer; the bottom rectangular box of the global average pooling layer represents the output layer; in the input layer Each rectangular box between the output layer and the output layer represents a hidden layer. Among them, 3+4+6+3=16 groups of hidden layers are included, and each group of hidden layers includes 3 hidden layers.

Step 704: Determine the loss value identified by the modulation mode according to the predicted modulation mode and the labeled modulation mode of each sample signal.

In the embodiments of this application, specifically, a cross-entropy (English: Cross-Entropy, abbreviated: CE) function can be used as the loss function to obtain the loss value J _CE (Φ), as shown in the following formula:

Among them, C represents the set of modulation methods, c represents the c-th modulation method, P represents the sample signal set, and p represents the sample signal,

Indicates the indicator parameter,

The value of can be 0 or 1. For example, if the marked modulation mode of the sample signal p is modulation mode c,

The value of is 1, if the marked modulation mode of the sample signal p is not the modulation mode c,

The value of is 0; Φ represents the parameter of the deep neural network, O _cp represents the output of the deep neural network, that is, O _cp represents the probability that the predicted modulation mode is the modulation mode c.

In the embodiment of the present application, other methods may also be used to determine the loss value of the modulation mode identification. For example, the mean square error function is used to determine the loss value of modulation recognition. In this regard, the embodiments of the present application do not make specific limitations.

Step 705: Determine whether the deep neural network converges according to the loss value. If not, go to step 706; if yes, go to step 707.

In the embodiment of the present application, whether the deep neural network converges may be determined based on a preset loss threshold. Specifically, it may be: when the loss value is less than the preset loss threshold, the deep neural network is determined to converge; if the loss value is greater than or equal to the preset loss threshold, it is determined that the deep neural network has not converged.

In the embodiment of the present application, whether the deep neural network converges may also be determined based on a preset change threshold. Specifically: when the difference between the loss value calculated this time and the loss value calculated last time is less than the preset change threshold, the deep neural network is determined to converge; if the loss value calculated this time is different from the loss value calculated last time If the difference is greater than or equal to the preset loss threshold, it is determined that the deep neural network has not converged.

In the embodiment of the present application, it is also possible to determine whether the deep neural network converges based on the preset training times threshold. Specifically, it may be: when the number of training times for the deep neural network reaches the preset training times threshold, the deep neural network is determined to converge; if the number of training times for the deep neural network does not reach the preset training times threshold, it is determined that the deep neural network has not converged.

In the embodiment of the present application, it is also possible to determine whether the deep neural network converges based on any two or three of the preset loss threshold, the preset change threshold, and the preset training times threshold. For example, based on any two or three of the preset loss threshold and the preset training times threshold, it is determined whether the deep neural network converges. Specifically, it may be: if the loss value is less than the preset loss threshold, or the number of training times for the deep neural network reaches the preset number of training times threshold, the deep neural network is determined to converge; otherwise, it is determined that the deep neural network has not converged.

In the embodiment of the present application, the method for determining whether the deep neural network converges is not limited.

Step 706: Adjust the parameters of the deep neural network, and return to step 703.

Step 707: Determine that the current deep neural network is the modulation mode recognition model.

In the embodiments of the present application, the electronic device for training the modulation mode recognition model and the electronic device for recognizing the modulation mode may be located on the same physical machine, or may be located on different physical machines. There is no restriction on this.

In the embodiment of this application, the modulation recognition model obtained by training in the above steps 701-707 is used. The modulation recognition model is trained by inputting the cyclic spectra of signals in different modulation modes into the deep neural network, so that the modulation recognition model can learn different All the characteristics of the cyclic spectrogram of the signal under the modulation mode, and then the modulation mode recognition model is based on all the characteristics of the cyclic spectrogram of the signal under different modulation modes learned, and all the characteristics of the cyclic spectrogram of the signal to be identified are processed to determine the to-be-recognized The modulation method of the signal improves the accuracy of modulation recognition.

The following describes the modulation method identification method provided in the embodiment of the present application in detail with reference to the flowchart of the modulation method identification method shown in FIG. 9.

Training phase:

The electronic device obtains the received signal in the cooperative mode, and performs processing such as spectrum shifting, calculation of the cyclic spectrum, smoothing, and image superposition on the received signal in order to obtain the cyclic spectrum of the received signal. Among them, the frequency spectrum shift processing is to prepare for the image superposition. The electronic device inputs the cyclic spectrum of the received signal into the preset deep neural network, and after the training is completed, the parameters of the preset deep neural network are fixed to obtain the modulation mode recognition model. Among them, the processing operations such as spectrum shifting, calculation of cyclic spectrum, and picture superposition can refer to the description in the above-mentioned parts of Figures 1-8, which will not be repeated here, and the smoothing processing is the processing of calculating the average value described above. The received signal can be understood as a sample signal.

Recognition stage:

The electronic device obtains the signal to be identified, and performs the processing of spectrum shifting, calculation of the cyclic spectrum, smoothing, and image superposition of the signal to be identified in order to obtain the cyclic spectrum of the signal to be identified, and input the cyclic spectrum of the signal to be identified to the training result The modulation mode recognition model is used to recognize the picture, that is, to recognize the cyclic spectrum of the signal to be recognized, and then to identify the modulation mode of the signal to be recognized.

The description of Fig. 9 is relatively simple. For details, please refer to the description of Figs. 1-8.

In the modulation method identification method provided by the embodiments of the present application, the electronic device inputs the complete cyclic spectrum of the signal to be identified into the modulation method identification model to identify the modulation method of the signal to be identified, that is, using the signal to be identified To identify the modulation mode of the signal to be identified instead of using part of the characteristics of the cyclic spectrum to identify the modulation mode of the signal to be identified, which improves the accuracy of the recognition of the modulation mode of the signal.

Based on the same technical concept, an embodiment of the present application also provides a modulation mode identification device. As shown in FIG. 10, the device includes:

The receiving module 1001 is used to receive the signal to be identified;

The determining module 1002 is configured to determine the cyclic spectrum of the signal to be identified according to the frequency components and the cyclic frequency components of the signal to be identified, as the target cyclic spectrum;

The recognition module 1003 is used to input the target cycle spectrogram into the pre-trained modulation recognition model to obtain the modulation mode of the signal to be recognized. The modulation recognition model is a model obtained by training a deep neural network using sample data. The data includes multiple sample signals and the marked modulation method of each sample signal.

Optionally, the determining module 1002 may include:

The superposition sub-module is used to superimpose the cyclic spectrograms corresponding to multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram.

Optionally, the moving sub-module can be specifically used to: perform segment processing on the signal to be identified to obtain multiple sub-signals to be identified; for each sub-signal to be identified, move the frequency spectrum of the sub-signal to be identified to a preset Obtain multiple frequency band sub-signals corresponding to the sub-signal to be identified on multiple frequency bands; wherein, multiple frequency band sub-signals corresponding to the multiple sub-signals to be identified on one frequency band are frequency band signals corresponding to the signal to be identified on the frequency band;

The first determining sub-module can be specifically used to: determine the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal; for each frequency band, multiple frequency bands on the frequency band The cyclic spectrum of the sub-signal is averaged to obtain the cyclic spectrum of the frequency band signal in the frequency band.

Optionally, the first determining sub-module can be specifically used to: for each frequency band signal, perform segmentation processing on the frequency band signal to obtain multiple frequency band sub-signals corresponding to the frequency band signal; according to each frequency band sub-signal Frequency components and each cyclic frequency component, determine the cyclic spectrum of each frequency band sub-signal; for each frequency band, perform averaging processing on the cyclic spectrum of multiple frequency band sub-signals on the frequency band to obtain the cyclic spectrum of the frequency band signal on the frequency band Spectrogram.

Optionally, the first determining sub-module may be specifically used to: perform fast Fourier transform on each frequency band sub-signal to obtain a transformed sub-signal corresponding to each frequency band sub-signal; according to each frequency component of each transformed sub-signal With each cyclic frequency component, the cyclic spectrum of each transformed sub-signal is determined as the cyclic spectrum of the frequency band sub-signal corresponding to each transformed sub-signal.

Optionally, the first determining submodule may be specifically used for:

among them,

Represents the cyclic spectrum of the k-th transformed sub-signal corresponding to the signal to be identified,

Expressed in

The k-th transformed sub-signal corresponding to the signal to be identified under the frequency,

Means

The conjugate,

Expressed in

For the k-th transformed sub-signal corresponding to the signal to be identified at the frequency, |·| can represent the calculation of the modulus value.

Optionally, the overlay sub-module can be specifically used for:

|S′ _r (f,α)|=max{|S _r,1 (f,α)|,|S _r,2 (f,α)|,…,|S _r,n (f,α)| };

Among them, |S' _r (f,α)| represents the cyclic spectrum of the signal to be identified, |S _r,l (f,α)| represents the cyclic spectrum corresponding to the l-th frequency band, l represents the frequency band, and l=1 ,2,...n,max represents the maximum value, n represents the number of multiple frequency bands, f represents the frequency component, α represents the cyclic frequency component, and |·| can represent the calculation of the modulus value.

Optionally, the above-mentioned device may further include: a training module for training to obtain a modulation mode recognition model;

The above-mentioned training module may include:

The acquisition sub-module is used to acquire sample data, the sample data includes multiple sample signals and the label modulation mode of each sample signal;

The fourth determining sub-module is used to determine whether the deep neural network has converged according to the loss value; if not, adjust the parameters of the deep neural network, and return to execute inputting the cyclic spectrum of each sample signal into the preset deep neural network to obtain The step of predicting the modulation mode of each sample signal; if yes, determine that the current deep neural network is the modulation mode recognition model.

In the modulation method identification device provided by the embodiment of the present application, the electronic device inputs the complete cyclic spectrum of the signal to be identified into the modulation method identification model to identify the modulation method of the signal to be identified, that is, using the signal to be identified To identify the modulation mode of the signal to be identified instead of using part of the characteristics of the cyclic spectrum to identify the modulation mode of the signal to be identified, which improves the accuracy of the recognition of the modulation mode of the signal.

Based on the same technical concept, an embodiment of the present application also provides an electronic device, as shown in FIG. 11, including a processor 1101, a communication interface 1102, a memory 1103, and a communication bus 1104, wherein the processor 1101, the communication interface 1102, The memories 1103 communicate with each other through the communication bus 1104;

The memory 1103 is used to store computer programs;

The processor 1101 is configured to implement the method steps in any of the foregoing modulation mode identification method embodiments when executing the program stored in the memory 1103.

In an electronic device provided by an embodiment of the present application, the electronic device inputs the complete cyclic spectrum of the signal to be identified into the modulation method recognition model to identify the modulation method of the signal to be identified, that is, using the cyclic spectrum of the signal to be identified All the characteristics of the spectrogram identify the modulation method of the signal to be identified, instead of using part of the cyclic spectrogram to identify the modulation method of the signal to be identified, which improves the accuracy of the recognition of the signal modulation method.

The communication bus mentioned by the aforementioned network device may be a peripheral component interconnection standard (English: Peripheral Component Interconnect, abbreviated as: PCI) bus or an extended industry standard architecture (English: Extended Industry Standard Architecture, abbreviated as: EISA) bus, etc. The communication bus can be divided into address bus, data bus, control bus and so on. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

The communication interface is used for communication between the aforementioned network device and other devices.

The memory may include random access memory (English: Random Access Memory, abbreviated as: RAM), and may also include non-volatile memory (English: Non-Volatile Memory, abbreviated as: NVM), for example, at least one disk storage. Optionally, the memory may also be at least one storage device located far away from the foregoing processor.

The above-mentioned processor may be a general-purpose processor, including a central processing unit (English: Central Processing Unit, abbreviated as: CPU), a network processor (English: Network Processor, abbreviated as: NP), etc.; it may also be a digital signal processor (English: : Digital Signal Processing, abbreviation: DSP), application specific integrated circuit (English: Application Specific Integrated Circuit, abbreviation: ASIC), Field-Programmable Gate Array (English: Field-Programmable Gate Array, abbreviation: FPGA) or other programmable logic devices , Discrete gates or transistor logic devices, discrete hardware components.

Based on the same technical concept, the embodiments of the present application also provide a computer-readable storage medium in which a computer program is stored, and when the computer program is executed by a processor, any step of the above modulation method identification method is implemented.

Based on the same technical concept, the embodiments of the present application also provide a computer program, which when running on a computer, causes the computer to execute any step of the above-mentioned modulation mode identification method.

In the above-mentioned embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, it can be implemented in the form of a computer program product in whole or in part. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website, computer, server, or data center. Transmission to another website site, computer, server or data center via wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center integrated with one or more available media. The usable medium may be a magnetic medium (for example, a floppy disk, a hard disk, and a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)).

It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply one of these entities or operations. There is any such actual relationship or order between. Moreover, the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes those that are not explicitly listed Other elements of, or also include elements inherent to this process, method, article or equipment. If there are no more restrictions, the element defined by the sentence "including a..." does not exclude the existence of other identical elements in the process, method, article, or equipment that includes the element.

The various embodiments in this specification are described in a related manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, electronic equipment, computer-readable storage medium, and computer program embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for related parts, please refer to the part of the description of the method embodiments.

The foregoing descriptions are only preferred embodiments of the present application, and are not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application are all included in the protection scope of this application.

Claims

A modulation method identification method, characterized in that the method includes:

Receive the signal to be identified;

Determine the cyclic spectrum of the signal to be identified according to the frequency components and the cyclic frequency components of the signal to be identified, as the target cyclic spectrum;

Input the target cyclic spectrogram into a pre-trained modulation mode recognition model to obtain the modulation mode of the signal to be recognized, wherein the modulation mode recognition model is a model obtained by training a deep neural network using sample data, and The sample data includes multiple sample signals and the label modulation mode of each sample signal.
The method according to claim 1, wherein the step of determining the cyclic spectrum of the signal to be identified as the target cyclic spectrum according to the respective frequency components and the respective cyclic frequency components of the signal to be identified, include:

Moving the frequency spectrum of the signal to be identified to a plurality of preset frequency bands to obtain signals of multiple frequency bands corresponding to the signal to be identified;

Determine the cyclic spectrogram of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal, and the cyclic spectrogram of each frequency band signal is the cyclic spectrogram corresponding to the frequency band where the signal of the frequency band is located;

Performing superposition processing on the cyclic spectrograms corresponding to the multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram.
The method according to claim 2, wherein the step of moving the frequency spectrum of the signal to be identified to a plurality of preset frequency bands respectively to obtain signals of multiple frequency bands corresponding to the signal to be identified comprises:

Performing segmentation processing on the signal to be identified to obtain a plurality of sub-signals to be identified;

For each sub-signal to be identified, the frequency spectrum of the sub-signal to be identified is moved to multiple preset frequency bands to obtain multiple frequency band sub-signals corresponding to the sub-signal to be identified; The multiple frequency band sub-signals corresponding to the sub-signal to be identified are frequency band signals corresponding to the signal to be identified in the frequency band;

The step of determining the cyclic spectrum of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal includes:

Determine the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal;

For each frequency band, averaging is performed on the cyclic spectrograms of multiple frequency band sub-signals on the frequency band to obtain the cyclic spectrogram of the frequency band signals on the frequency band.
3. The method according to claim 2, wherein the step of determining the cyclic spectrum of each frequency band signal according to each frequency component and each cyclic frequency component of each frequency band signal comprises:

For each frequency band signal, perform segmentation processing on the frequency band signal to obtain multiple frequency band sub-signals corresponding to the frequency band signal;

Determine the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal;

For each frequency band, averaging is performed on the cyclic spectrograms of multiple frequency band sub-signals on the frequency band to obtain the cyclic spectrogram of the frequency band signals on the frequency band.
The method according to claim 3 or 4, wherein the step of determining the cyclic spectrum of each frequency band sub-signal according to each frequency component and each cyclic frequency component of each frequency band sub-signal comprises:

Fast Fourier transform is performed on each frequency band sub-signal to obtain the transformed sub-signal corresponding to each frequency band sub-signal;

According to each frequency component and each cyclic frequency component of each transformed sub-signal, the cyclic spectrum of each transformed sub-signal is determined as the cyclic spectrum of the frequency band sub-signal corresponding to each transformed sub-signal.
The method according to claim 5, wherein the step of determining the cyclic spectrum of each transformed sub-signal according to each frequency component and each cyclic frequency component of each transformed sub-signal comprises:

Use the following formula to determine the cyclic spectrum of each transformed sub-signal:

among them,
Represents the cyclic spectrum of the k-th transformed sub-signal, N 0 represents the signal length of the k-th transformed sub-signal, f represents the frequency component, and α represents the cyclic frequency component,
Expressed in
The k-th transformed sub-signal at frequency,
Means
The conjugate,
Expressed in
For the k-th transformed sub-signal at frequency, |·| represents the modulus value.
The method according to claim 2, wherein the step of performing superposition processing on the cyclic spectrograms corresponding to the multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram comprises :

Use the following formula to superimpose the cyclic spectra corresponding to multiple frequency bands to obtain the cyclic spectrogram of the signal to be identified as the target cyclic spectrogram:

|S′ r (f,α)|=max{|S r,1 (f,α)|,|S r,2 (f,α)|,…,|S r,n (f,α)| };

Where |S' r (f,α)| represents the cyclic spectrum of the signal to be identified, |S r,l (f,α)| represents the cyclic spectrum corresponding to the l-th frequency band, l represents the frequency band, and l =1,2,...n, max represents the maximum value, n represents the number of the multiple frequency bands, f represents the frequency component, α represents the cyclic frequency component, and |·| represents the modulus value.
The method according to claim 1, wherein the modulation mode recognition model is obtained by training in the following manner:

Acquiring the sample data, where the sample data includes the multiple sample signals and the labeling modulation mode of each sample signal;

Determine the cyclic spectrum of each sample signal according to each frequency component and each cyclic frequency component of each sample signal;

Input the cyclic spectrogram of each sample signal into the preset deep neural network to obtain the predicted modulation mode of each sample signal;

Determine the loss value of the modulation mode recognition according to the predicted modulation mode and the labeled modulation mode of each sample signal;

Determine whether the deep neural network converges according to the loss value;

If not, adjust the parameters of the deep neural network and return to the step of inputting the cyclic spectrogram of each sample signal into the preset deep neural network to obtain the predicted modulation mode of each sample signal;

If yes, it is determined that the current deep neural network is the modulation mode recognition model.
A modulation mode identification device, characterized in that the device includes:

The receiving module is used to receive the signal to be identified;

The determining module is configured to determine the cyclic spectrum of the signal to be identified according to the respective frequency components and the respective cyclic frequency components of the signal to be identified, as the target cyclic spectrum;

The recognition module is used to input the target cyclic spectrogram into a pre-trained modulation mode recognition model to obtain the modulation mode of the signal to be recognized, wherein the modulation mode recognition model is obtained by training a deep neural network using sample data In the model, the sample data includes a plurality of sample signals and the label modulation mode of each sample signal.
An electronic device, characterized by comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete mutual communication through the communication bus;

The memory is used to store computer programs;

The processor is configured to execute the program stored in the memory to implement the method steps of any one of claims 1-8.
A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the method steps according to any one of claims 1-8 are realized.
A computer program, characterized in that, when the computer program is executed by a processor, the method steps according to any one of claims 1-8 are realized.