Signal modulation identification model training method, signal modulation identification method and device

Abstract

The invention provides a training method of a signal modulation identification model, and a signal modulation identification method and device. The method comprises the following steps: collecting sample single side band signals of a sample communication signal in a plurality of modulation modes; extracting a sample high-order cumulant feature of the sample single-side-band signal in any modulation mode, and determining a sample modulation signal feature according to the sample high-order cumulant feature; performing modulation mode prediction on the sample modulation signal features by adopting a signal modulation identification model to obtain prediction labels of the sample modulation signal features; wherein the prediction label is used for indicating a prediction modulation mode corresponding to the sample modulation signal feature; according to the method, the signal modulation identification model is trained according to the difference between the predicted modulation mode and any modulation mode, so that the signal modulation identification model can be accurately trained, and the modulation mode of the communication signal can be accurately and efficiently identified based on the signal modulation identification model obtained through training.

Description

Technical Field

[0001] This disclosure relates to the field of communication signal processing technology, and in particular to a training method for a signal modulation recognition model, a signal modulation recognition method, and an apparatus. Background Technology

[0002] Currently, signal modulation identification refers to the process of identifying and analyzing received communication signals when prior information such as modulation details and modulation type is lacking. This process estimates the modulation scheme used in the communication signal and identifies certain modulation information, providing parameters for the receiver to correctly select the demodulation algorithm and ultimately obtain useful information. Therefore, how to implement signal modulation identification is crucial. Summary of the Invention

[0003] This disclosure aims to at least partially address one of the technical problems in the related art.

[0004] To address this, this disclosure proposes a training method for a signal modulation recognition model, a signal modulation recognition method, and an apparatus. The method determines the characteristics of a sample modulation signal by using the high-order cumulant features of the sample single-sideband signal under multiple modulation modes of a sample communication signal. Then, based on the signal modulation recognition model, the modulation mode is predicted for the sample modulation signal characteristics, resulting in a predicted label for the sample modulation signal characteristics. By accurately training the signal modulation recognition model based on the difference between the predicted modulation mode indicated by the predicted label and any given modulation mode, the trained signal modulation recognition model can accurately and efficiently identify the modulation mode of a communication signal.

[0005] The first aspect of this disclosure proposes a training method for a signal modulation recognition model, comprising: acquiring sample single-sideband signals of sample communication signals under multiple modulation schemes; extracting sample higher-order cumulant features of the sample single-sideband signals under any modulation scheme, and determining sample modulation signal features based on the sample higher-order cumulant features; using the signal modulation recognition model to predict the modulation scheme of the sample modulation signal features to obtain a predicted label of the sample modulation signal features; wherein the predicted label is used to indicate the predicted modulation scheme corresponding to the sample modulation signal features; and training the signal modulation recognition model based on the difference between the predicted modulation scheme and any modulation scheme.

[0006] The signal modulation recognition model training method of this disclosure includes: collecting sample single-sideband signals of sample communication signals under multiple modulation modes; extracting sample high-order cumulant features of the sample single-sideband signals under any modulation mode, and determining sample modulation signal features based on the sample high-order cumulant features; using the signal modulation recognition model to predict the modulation mode of the sample modulation signal features to obtain a prediction label for the sample modulation signal features; wherein the prediction label is used to indicate the predicted modulation mode corresponding to the sample modulation signal features; and training the signal modulation recognition model based on the difference between the predicted modulation mode and any modulation mode. Thus, by using the sample high-order cumulant features of the sample single-sideband signals of sample communication signals under multiple modulation modes, the sample modulation signal features are determined. Furthermore, by predicting the modulation mode of the sample modulation signal features based on the signal modulation recognition model, a prediction label for the sample modulation signal features is obtained. Based on the difference between the predicted modulation mode indicated by the prediction label and any modulation mode, the signal modulation recognition model can be accurately trained. Therefore, based on the trained signal modulation recognition model, the modulation mode of communication signals can be accurately and efficiently identified.

[0007] A second aspect of this disclosure provides a signal modulation identification method, comprising: acquiring a target communication signal; collecting target single-sideband signals of the target communication signal under multiple modulation schemes; extracting target high-order cumulant features of the target single-sideband signal under any modulation scheme, and determining target modulation signal features based on the target high-order cumulant features; and using a signal modulation identification model to predict the modulation scheme of the target modulation signal features to obtain the modulation scheme of the target communication signal.

[0008] A third aspect of this disclosure provides a training apparatus for a signal modulation recognition model, comprising: an acquisition module for acquiring sample single-sideband signals of sample communication signals under multiple modulation schemes; an extraction module for extracting sample higher-order cumulant features of the sample single-sideband signals under any modulation scheme, and determining sample modulation signal features based on the sample higher-order cumulant features; a prediction module for using the signal modulation recognition model to predict the modulation scheme of the sample modulation signal features to obtain a prediction label for the modulation signal features; wherein the prediction label is used to indicate the predicted modulation scheme corresponding to the sample modulation signal features; and a training module for training the signal modulation recognition model based on the difference between the predicted modulation scheme and any modulation scheme.

[0009] A fourth aspect of this disclosure provides a signal modulation identification device, comprising: an acquisition module for acquiring a target communication signal; an acquisition module for acquiring target single-sideband signals of the target communication signal under multiple modulation schemes; an extraction module for extracting target high-order cumulant features of the target single-sideband signal under any modulation scheme, and determining target modulation signal features based on the target high-order cumulant features; and a prediction module for predicting the modulation scheme of the target modulation signal features using a signal modulation identification model to obtain the modulation scheme of the target communication signal.

[0010] A fifth aspect of this disclosure provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a training method for a signal modulation recognition model as described in a first aspect of this disclosure, or implements a signal modulation recognition method as described in a second aspect of this disclosure.

[0011] A sixth aspect of this disclosure provides a computer-readable storage medium having a computer program stored thereon. When executed by a processor, the computer program implements a training method for a signal modulation recognition model as described in a first aspect of this disclosure, or implements a signal modulation recognition method as described in a second aspect of this disclosure.

[0012] A seventh aspect of this disclosure provides a computer program product that, when executed by an instruction processor, implements a training method for a signal modulation recognition model as described in a first aspect of this disclosure, or implements a signal modulation recognition method as described in a second aspect of this disclosure.

[0013] Additional aspects and advantages of this disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this disclosure. Attached Figure Description

[0014] The above and / or additional aspects and advantages of this disclosure will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, in which:

[0015] Figure 1 This is a schematic flowchart illustrating a training method for a signal modulation recognition model provided in an embodiment of this disclosure.

[0016] Figure 2 This is a schematic flowchart illustrating a training method for a signal modulation recognition model provided in an embodiment of this disclosure.

[0017] Figure 3 This is a schematic flowchart illustrating a training method for a signal modulation recognition model provided in an embodiment of this disclosure.

[0018] Figure 4 This is a schematic flowchart of a signal modulation recognition method provided in an embodiment of the present disclosure;

[0019] Figure 5 This is a schematic diagram of the structure of a training device for a signal modulation recognition model provided in an embodiment of this disclosure;

[0020] Figure 6 This is a schematic diagram of the structure of a signal modulation recognition device provided in an embodiment of the present disclosure;

[0021] Figure 7 This is a block diagram illustrating an electronic device for training a signal modulation recognition model or for signal modulation recognition, according to an exemplary embodiment. Detailed Implementation

[0022] Embodiments of this disclosure are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting this disclosure.

[0023] Modulation pattern identification is a process between energy detection and demodulation. Energy detection only requires a rough estimate of the signal bandwidth and center frequency, while demodulation requires precise frequency information and the signal's modulation pattern. Modulation pattern identification needs to be completed with limited prior knowledge to obtain more accurate parameter values. Current communication signals typically use different modulation patterns and parameters over a large bandwidth; therefore, identifying the modulation patterns of these signals has wide-ranging applications.

[0024] The key step in identifying the modulation scheme of a signal is the acquisition of the communication signal. Currently, some communication systems have large bandwidth and high dynamism in their communication signals (such as satellite signals). In order to ensure that the integrity of the communication signal is not lost, this places great demands on the acquisition capability of the modulation identification receiver.

[0025] Therefore, in order to address the above problems, this disclosure proposes a training method for a signal modulation recognition model, a signal modulation recognition method, and an apparatus.

[0026] The following describes, with reference to the accompanying drawings, a training method for a signal modulation recognition model, a signal modulation recognition method, and an apparatus according to embodiments of the present disclosure.

[0027] Figure 1 This is a flowchart illustrating a training method for a signal modulation recognition model provided in an embodiment of the present disclosure.

[0028] like Figure 1As shown, the training method for this signal modulation recognition model may include the following steps:

[0029] Step 101: Collect sample single-sideband signals of the sample communication signal under multiple modulation modes.

[0030] As one possible implementation of this disclosure, the sample communication signal can be obtained from a communication signal database, or a previously received communication signal can be used as the sample communication signal.

[0031] One possible implementation involves acquiring sample single-sideband signals of the communication signal under multiple modulation schemes. These multiple modulation schemes may include, but are not limited to: 2-phase shift keying (2PSK), quaternary phase shift keying (QPSK), 8-phase shift keying (8PSK), 16-phase shift keying (16PSK), 8-quadrant amplitude modulation (8QAM), 16-quadrant amplitude modulation (16QAM), 32-quadrant amplitude modulation (32QAM), 16-amplitude phase shift keying (16APSK), and 32-amplitude phase shift keying (32APSK).

[0032] As an example, a signal modulation recognition model is used to collect sample single-sideband signals of sample communication signals under multiple modulation modes.

[0033] As another example, a correlation modulator is used to acquire sample single-sideband signals of sample communication signals under multiple modulation modes.

[0034] Step 102: Extract the sample high-order cumulant features of the sample single-sideband signal under any modulation mode, and determine the sample modulation signal features based on the sample high-order cumulant features.

[0035] As an example, a signal modulation recognition model is used to process the sample single-sideband signal under any modulation mode to obtain the sample high-order cumulant features of the sample single-sideband signal under any modulation mode. Based on the sample high-order cumulant features of the sample single-sideband signal under any modulation mode, the sample modulation signal features are determined to predict the modulation mode.

[0036] As another example, a correlation feature extractor is used to extract the sample higher-order cumulant features of the sample single-sideband signal under any modulation mode, and the sample modulation signal features are determined based on the sample higher-order cumulant features.

[0037] Step 103: Use a signal modulation recognition model to predict the modulation mode of the sample modulation signal features in order to obtain the predicted label of the sample modulation signal features.

[0038] The prediction label is used to indicate the predicted modulation scheme corresponding to the features of the sample modulation signal.

[0039] As one possible implementation, a signal modulation recognition model is used to predict the modulation mode of the sample modulation signal features, thereby obtaining a predicted label for the sample modulation signal features. This predicted label can indicate the predicted modulation mode corresponding to the sample modulation signal features.

[0040] Step 104: Train the signal modulation recognition model based on the difference between the predicted modulation mode and any modulation mode.

[0041] Furthermore, based on the difference between the predicted modulation scheme and any modulation scheme, the loss function value is determined, and the signal modulation recognition model is trained based on the loss function value.

[0042] It should be noted that the above example only uses minimizing the value of the loss function as the termination condition for training the signal modulation recognition model. In actual applications, other termination conditions can also be set, such as the number of training iterations reaching a set threshold, the training duration exceeding a set duration threshold, etc. This disclosure does not impose any restrictions on these conditions.

[0043] In summary, this method involves collecting sample single-sideband signals of communication signals under multiple modulation schemes; extracting the high-order cumulant features of the sample single-sideband signals under any modulation scheme; determining the sample modulation signal features based on the high-order cumulant features; using a signal modulation recognition model to predict the modulation scheme of the sample modulation signal features to obtain predicted labels for the sample modulation signal features; and training the signal modulation recognition model based on the difference between the predicted modulation scheme and any modulation scheme. This allows for accurate training of the signal modulation recognition model, enabling accurate and efficient identification of the modulation scheme of communication signals based on the trained model.

[0044] To clearly illustrate how the above embodiments extract the sample high-order cumulant features of a sample single-sideband signal under any modulation mode, this disclosure proposes another training method for a signal modulation recognition model.

[0045] Figure 2 This is a flowchart illustrating a training method for a signal modulation recognition model provided in an embodiment of the present disclosure.

[0046] like Figure 2 As shown, the training method for this signal modulation recognition model may include the following steps:

[0047] Step 201: Collect sample single-sideband signals of the sample communication signal under multiple modulation modes.

[0048] Step 202: Obtain the sample baseband signal IQ data of the sample single-sideband signal under any modulation mode.

[0049] As one possible implementation, a Fast Fourier Transform (FFT) is performed on the sample single-sideband signal under any modulation scheme to obtain the sample center frequency and sample bandwidth of the sample single-sideband signal under any modulation scheme. The sample center frequency is then used to perform a quadrature down-conversion on the sample single-sideband signal under any modulation scheme to obtain the initial in-phase component and initial quadrature component of the sample baseband signal under any modulation scheme. Finally, the initial in-phase component and initial quadrature component of the sample baseband signal under any modulation scheme are filtered using the sample bandwidth to obtain the sample baseband signal IQ (In-phase Quadrature) data of the sample single-sideband signal under any modulation scheme.

[0050] As an example, the acquired signal is subjected to FFT to obtain spectral data, and the sample center frequency f is estimated based on the spectral data. ce and 3dB sample bandwidth B e To improve the estimated sample center frequency and 3dB sample bandwidth, the obtained spectrum is smoothed. Assuming the single-sideband signal of the sample acquired by the ADC (Analog-to-Digital Converter) is r(n), then r(n) can be expressed as:

[0051]

[0052] Among them, f c T is the carrier frequency. s T is the sampling interval. s With sampling rate f s The relationship between them is fs =1 / T s Let I(t) and Q(t) be the in-phase and quadrature components of the sampled single-sideband signal, respectively, n be the index of the sampling point, and θ be the initial phase of the carrier signal. Then r(n) can be rewritten as:

[0053] r(n)=I(n)cos(2πf c / f s +θ)-Q(n)sin(2πf c / f s +θ);

[0054] Furthermore, according to f ce The signal y is obtained by orthogonal downconversion of r(n). I (n) and y Q (n), which can be specifically represented as:

[0055]

[0056] Similarly, we can conclude that:

[0057]

[0058] Based on the sample bandwidth B e Design a low-pass filter to filter out y I (n) and y Q High-frequency components of (n) (e.g., f) c +f ce Let Δf = f c -f ce , Then the filtered signal in the same direction component z I (n) and orthogonal component z Q (n) can be represented as follows:

[0059]

[0060]

[0061] As can be seen from the above equation, when Δf and Δφ approach 0, This allows us to obtain the baseband signal IQ data z(n) = z of the sample single-sideband signal. I (n)+jz Q (n).

[0062] Step 203: Normalize the sample baseband signal IQ data of the sample single-sideband signal under any modulation mode to obtain the normalized sample baseband signal IQ data of the sample single-sideband signal under any modulation mode.

[0063] To improve the performance, generalization ability, and training efficiency of the signal modulation recognition model, as an example, the sample baseband signal IQ data of the sample single-sideband signal under any modulation mode can be normalized to unify the value range of the IQ data into a fixed interval.

[0064] Step 204: Extract the sample high-order cumulative features of the sample baseband signal IQ data of the sample single-sideband signal under any modulation mode after normalization.

[0065] As an example, the corresponding higher-order cumulants of the IQ data are calculated based on the selected higher-order cumulants order; the higher-order cumulants are then used as the values ​​of the higher-order cumulants feature. As another example, higher-order cumulants of multiple orders are combined into a single higher-order cumulants feature.

[0066] Step 205: Determine the characteristics of the sample modulation signal based on the high-order cumulant characteristics of the sample.

[0067] One possible implementation is to obtain the ratio between even-order cumulative features of multiple samples; and to determine the sample modulation signal features based on the ratio between even-order cumulative features of multiple samples.

[0068] For example, the modulation signal characteristic F1 = |C42| / |C40|, where |C42| is the absolute value of the fourth-order cumulant C42, and |C40| is the absolute value of the fourth-order cumulant C40; another example is the modulation signal characteristic... Where |C63| is the absolute value of the sixth-order cumulant C63, and |C40| is the absolute value of the fourth-order cumulant C42.

[0069] Step 206: Use a signal modulation recognition model to predict the modulation mode of the sample modulation signal features in order to obtain the predicted label of the sample modulation signal features.

[0070] The prediction label is used to indicate the predicted modulation scheme corresponding to the features of the sample modulation signal.

[0071] Step 207: Train the signal modulation recognition model based on the difference between the predicted modulation mode and any modulation mode.

[0072] In summary, by acquiring the sample baseband signal IQ data of a sample single-sideband signal under any modulation scheme; normalizing the sample baseband signal IQ data of the sample single-sideband signal under any modulation scheme to obtain normalized sample baseband signal IQ data of the sample single-sideband signal under any modulation scheme; extracting the sample high-order cumulant features of the sample baseband signal IQ data of the sample single-sideband signal under any modulation scheme after normalization; and determining the sample modulation signal features based on the sample high-order cumulant features, it is possible to reduce the signal sampling rate requirement and complexity without affecting the signal modulation recognition performance, and eliminate the influence of signal energy on parameter values ​​by using the sample high-order cumulant features as the sample modulation signal features.

[0073] To clearly illustrate how the loss function value is determined based on the difference between the predicted label corresponding to the predicted modulation mode and the labeled label corresponding to any modulation mode in the above embodiments, this disclosure proposes another training method for the signal modulation recognition model.

[0074] Figure 3 This is a flowchart illustrating a training method for a signal modulation recognition model provided in an embodiment of the present disclosure.

[0075] like Figure 3 As shown, the training method for this signal modulation recognition model may include the following steps:

[0076] Step 301: Collect sample single-sideband signals of the sample communication signal under multiple modulation modes.

[0077] Step 302: Extract the sample higher-order cumulant features of the sample single-sideband signal under any modulation mode, and determine the sample modulation signal features based on the sample higher-order cumulant features.

[0078] Step 303: Use a signal modulation recognition model to predict the modulation mode of the sample modulation signal features in order to obtain the predicted label of the sample modulation signal features.

[0079] The prediction label is used to indicate the predicted modulation scheme corresponding to the features of the sample modulation signal.

[0080] Step 304: Determine the loss function value based on the difference between the predicted label corresponding to the predicted modulation mode and the labeled label corresponding to any modulation mode.

[0081] As one possible implementation, a sub-loss function value for any modulation scheme is generated based on the difference between the prediction label corresponding to the predicted modulation scheme and the annotation label corresponding to any modulation scheme; the sub-loss function values ​​for each modulation scheme are then weighted and summed to generate the loss function value.

[0082] Step 305: Train the signal modulation recognition model based on the loss function value.

[0083] As an example, the model parameters of the signal modulation recognition model are adjusted to minimize the loss function value.

[0084] It should be noted that the above example only uses minimizing the value of the loss function as the termination condition for training the signal modulation recognition model. In actual applications, other termination conditions can also be set, such as the number of training iterations reaching a set threshold, the training duration exceeding a set duration threshold, etc. This disclosure does not impose any restrictions on these conditions.

[0085] In summary, the loss function value is determined by the difference between the predicted label corresponding to the predicted modulation mode and the labeled label corresponding to any modulation mode. Based on the loss function value, the signal modulation recognition model is trained. Thus, the signal modulation recognition model can be effectively trained, and the signal modulation recognition model based on the trained model can accurately and efficiently identify the modulation mode of communication signals.

[0086] Based on any embodiment of this disclosure, the implementation process of this disclosure mainly includes the following steps:

[0087] 1. The single-sideband signal samples with different modulation methods are acquired through the modulation recognition receiver acquisition module;

[0088] For modulation identification receiver acquisition modules, the received signal is usually a centrally symmetrical signal, and the left and right sidebands of the signal spectrum carry signal information. Now, given that the received single-carrier signal has a high frequency and large bandwidth, in order to reduce the requirements of the receiver acquisition module, the single-sideband signal of the original received signal is acquired, thereby reducing the acquisition bandwidth and the cost of the receiver acquisition module.

[0089] 2. Perform FFT processing on each acquired signal to estimate the sample center frequency and sample bandwidth of the obtained spectrum. Then, perform orthogonal downconversion on the acquired signal based on the sample center frequency, and use the sample bandwidth to design a low-pass filter to filter the same-direction component and orthogonal component obtained by orthogonal downconversion, so as to obtain the IQ data of the sample time-domain baseband signal.

[0090] 3. Normalize the acquired sample signal time-domain IQ data, extract the sample higher-order cumulant features of the normalized IQ data, and use the extracted sample higher-order cumulant features as sample modulation signal features (classification features).

[0091] Among them, the ratio of higher-order cumulants of samples is used as the sample modulation signal feature for different modulation signals, including:

[0092] F1 = |C42| / |C40|, where |C42| is the absolute value of the fourth-order cumulant C42, and |C40| is the absolute value of the fourth-order cumulant C40;

[0093] Where |C63| is the absolute value of the sixth-order cumulant C63, and |C40| is the absolute value of the fourth-order cumulant C42;

[0094] F3 = |C20| / |C21|, where |C20| is the absolute value of the second-order cumulant C20, and |C21| is the absolute value of the second-order cumulant C21;

[0095] Where |C80| is the absolute value of the eighth-order cumulant C80, and |C21| is the absolute value of the second-order cumulant C21;

[0096] Where |C63| is the absolute value of the sixth-order cumulant C63, and |C80| is the absolute value of the eighth-order cumulant C80.

[0097] 4. Label the extracted high-order cumulative features of the samples and assign labels according to the known modulation methods;

[0098] It can distinguish between nine signals: 2PSK, QPSK, 8PSK, 16PSK, 8QAM, 16QAM, 32QAM, 16APSK, and 32APSK.

[0099] 5. Input the labeled dataset into the neural network for training;

[0100] (1) Specify the training set and test set for the input network, including training data and labels, and test data and labels;

[0101] (2) Build the network structure layer by layer, and determine the network hierarchy based on the signal modulation type recognition effect of the training model;

[0102] (3) Configure the training method of the neural network and observe the recognition effect of the signal modulation type of the training model by configuring different training parameters;

[0103] (4) Perform the training process and model performance verification.

[0104] It should be noted that the neural network structure used in this disclosure may include: input layer, intermediate layer, output layer, etc.

[0105] Furthermore, it should be noted that in this embodiment, a signal acquisition device can be used to acquire 100 sets of different types of modulation signals from a signal source; the acquired signals undergo preprocessing measures such as down-conversion, low-pass filtering, and resampling; the preprocessed signals are then used to extract the higher-order cumulants of the samples using relevant formulas; the F1-F5 feature values ​​of the acquired signals are calculated; and the feature values ​​are labeled according to the signal modulation type using a one-hot encoding method, for example...

[0106] The label corresponding to 2PSK is: Label = [1,0,0,0,0,0,0,0,0];

[0107] The label corresponding to QPSK is: Label = [0,1,0,0,0,0,0,0,0];

[0108] The label corresponding to 8PSK is: Label = [0,0,1,0,0,0,0,0,0];

[0109] The label corresponding to 16PSK is: Label = [0,0,0,1,0,0,0,0,0];

[0110] The label corresponding to 8QAM is: Label = [0,0,0,0,1,0,0,0,0];

[0111] The label corresponding to 16QAM is: Label = [0,0,0,0,0,1,0,0,0];

[0112] The label corresponding to 32QAM is: Label = [0,0,0,0,0,0,1,0,0];

[0113] The label corresponding to 16APSK is: Label = [0,0,0,0,0,0,0,1,0];

[0114] The label corresponding to 32APSK is: Label = [0,0,0,0,0,0,0,0,1];

[0115] The data structure after labeling is [(F1,F2,F3,F4,F5),Label].

[0116] Repeat the above steps to process 100 sets of data for all 9 signals. Divide the processed data into training and test sets according to a ratio of 80% and 20%. The neural network is constructed with five nodes in the input layer (F1, F2, F3, F4, F5), 32 nodes in the hidden layer, and 7 nodes in the output layer. Each node represents the probability of recognizing a signal.

[0117] The training set data was trained using a neural network. After multiple rounds of training, the accuracy of the training data reached 90%. The test set data was tested using the trained neural network, and the test accuracy reached over 80%. The trained neural network model can be used to predict the modulation mode of unknown types of signals in the future.

[0118] To achieve signal modulation recognition, this disclosure proposes a signal modulation recognition method.

[0119] Figure 4 This is a schematic flowchart of a signal modulation recognition method provided in an embodiment of the present disclosure.

[0120] like Figure 4 As shown, the signal modulation identification method may include the following steps:

[0121] Step 401: Obtain the target communication signal.

[0122] As an example, a communication signal acquisition device can be used to acquire the target communication signal, or a target communication signal receiving device can be used to receive the target communication signal.

[0123] Step 402: Collect the target single-sideband signal of the target communication signal under multiple modulation modes.

[0124] As an example, a signal modulation recognition model is used to collect the target single-sideband signal of the target communication signal under multiple modulation modes.

[0125] As another example, a correlation modulator is used to acquire the target single-sideband signal of the target communication signal under multiple modulation modes.

[0126] Step 403: Extract the target high-order cumulant features of the target single-sideband signal under any modulation mode, and determine the target modulation signal features based on the target high-order cumulant features.

[0127] As an example, a signal modulation recognition model is used to process the target single-sideband signal under any modulation mode to obtain the target high-order cumulant features of the target single-sideband signal under any modulation mode. Based on the target high-order cumulant features of the target single-sideband signal under any modulation mode, the target modulation signal features are determined to predict the modulation mode.

[0128] As another example, a relevant feature extractor is used to extract the target high-order cumulant features of the target single-sideband signal under any modulation mode, and the sample modulation signal features are determined based on the target high-order cumulant features.

[0129] Step 404: Using a signal modulation recognition model, the modulation mode of the target modulation signal is predicted to obtain the modulation mode of the target communication signal.

[0130] As one possible implementation, a signal modulation recognition model is used to predict the modulation mode of the target modulation signal features, thereby obtaining a predicted label of the target modulation signal features, wherein the predicted label can indicate the predicted modulation mode of the target modulation signal.

[0131] As an example, the collected target modulation signal features are input into a trained signal modulation recognition model to obtain the modulation scheme of the target communication signal output by the signal modulation recognition model. The signal modulation recognition model is... Figures 1 to 3 The signal modulation recognition model obtained through training in the example.

[0132] In summary, by acquiring the target communication signal; collecting the target single-sideband signal of the target communication signal under multiple modulation modes; extracting the target high-order cumulant features of the target single-sideband signal under any modulation mode; and determining the target modulation signal features based on the target high-order cumulant features, a signal modulation recognition model is used to predict the modulation mode of the target modulation signal features to obtain the modulation mode of the target communication signal. Thus, by using a trained signal modulation recognition model, the modulation mode of the target communication signal can be accurately and efficiently identified.

[0133] To achieve the above Figures 1 to 3 In this embodiment, the present disclosure proposes a training device for a signal modulation recognition model.

[0134] Figure 5 This is a schematic diagram of the structure of a training device for a signal modulation recognition model provided in an embodiment of this disclosure.

[0135] like Figure 5 As shown, the training device 500 for the signal modulation recognition model includes: an acquisition module 510, an extraction module 520, a prediction module 530, and a training module 540.

[0136] The acquisition module 510 is used to acquire sample single-sideband signals of sample communication signals under multiple modulation modes; the extraction module 520 is used to extract the sample high-order cumulant features of the sample single-sideband signals under any modulation mode, and determine the sample modulation signal features based on the sample high-order cumulant features; the prediction module 530 is used to predict the modulation mode of the sample modulation signal features using a signal modulation recognition model to obtain a prediction label for the sample modulation signal features; wherein, the prediction label is used to indicate the predicted modulation mode corresponding to the modulation signal features; and the training module 540 is used to train the signal modulation recognition model based on the difference between the predicted modulation mode and any modulation mode.

[0137] As one possible implementation, the extraction module 520 is used to acquire the sample baseband signal IQ data of the sample single-sideband signal under any modulation scheme; normalize the sample baseband signal IQ data of the sample single-sideband signal under any modulation scheme to obtain the normalized sample baseband signal IQ data of the sample single-sideband signal under any modulation scheme; and extract the sample high-order cumulant features of the normalized sample baseband signal IQ data of the sample single-sideband signal under any modulation scheme.

[0138] As one possible implementation, the extraction module 520 is used to perform a fast Fourier transform on the sample single-sideband signal under any modulation mode to obtain the sample center frequency and sample bandwidth of the sample single-sideband signal under any modulation mode; to perform a quadrature down-conversion on the sample single-sideband signal under any modulation mode using the sample center frequency to obtain the initial in-phase component and initial quadrature component of the sample baseband signal of the sample single-sideband signal under any modulation mode; and to filter the initial in-phase component and initial quadrature component of the sample baseband signal of the sample single-sideband signal under any modulation mode using the sample bandwidth to obtain the sample baseband signal IQ data of the sample single-sideband signal under any modulation mode.

[0139] As one possible implementation, the extraction module 520 is also used to obtain the ratio between the even-order cumulative features of multiple samples; and to determine the sample modulation signal features based on the ratio between the even-order cumulative features of multiple samples.

[0140] As one possible implementation, the training device 500 for the signal modulation recognition model also includes a labeling module.

[0141] The annotation module is used to label the features of the sample modulation signal to obtain the annotation labels of the sample modulation signal features; the annotation labels are used to indicate the actual modulation method corresponding to the sample modulation signal features.

[0142] As one possible implementation, the training module 540 is used to determine the loss function value based on the difference between the predicted label corresponding to the predicted modulation mode and the labeled label corresponding to any modulation mode; and to train the signal modulation recognition model based on the loss function value.

[0143] As one possible implementation, the training module 540 is also used to generate a sub-loss function value for any modulation mode based on the difference between the predicted label corresponding to the predicted modulation mode and the labeled label corresponding to any modulation mode; and to perform a weighted summation of the sub-loss function values ​​for each modulation mode to generate a loss function value.

[0144] The training apparatus for the signal modulation recognition model of this disclosure collects sample single-sideband signals of sample communication signals under multiple modulation modes; extracts the sample high-order cumulant features of the sample single-sideband signals under any modulation mode, and determines the sample modulation signal features based on the sample high-order cumulant features; uses the signal modulation recognition model to predict the modulation mode of the sample modulation signal features to obtain a predicted label for the modulation signal features; wherein the predicted label is used to indicate the predicted modulation mode corresponding to the modulation signal features; and trains the signal modulation recognition model based on the difference between the predicted modulation mode and any modulation mode. Thus, by using the sample high-order cumulant features of the sample single-sideband signals of sample communication signals under multiple modulation modes, the sample modulation signal features are determined. Furthermore, by predicting the modulation mode of the sample modulation signal features based on the signal modulation recognition model, a predicted label for the sample modulation signal features is obtained. Based on the difference between the predicted modulation mode indicated by the predicted label and any modulation mode, the signal modulation recognition model can be accurately trained. Therefore, the signal modulation recognition model obtained by training can accurately and efficiently identify the modulation mode of communication signals.

[0145] It should be noted that the foregoing explanation of the training method embodiment for the signal modulation recognition model also applies to the training device for the signal modulation recognition model in this embodiment, and will not be repeated here.

[0146] To achieve the above Figure 4 In this embodiment, the present disclosure proposes a signal modulation identification device.

[0147] Figure 6 This is a schematic diagram of the structure of a signal modulation recognition device provided in an embodiment of the present disclosure.

[0148] like Figure 6 As shown, the signal modulation recognition device 600 includes: an acquisition module 610, a collection module 620, an extraction module 630, and a prediction module 640.

[0149] The system includes an acquisition module for acquiring the target communication signal; an acquisition module 620 for acquiring the target single-sideband signal of the target communication signal under multiple modulation modes; an extraction module for extracting the target high-order cumulant features of the target single-sideband signal under any modulation mode, and determining the target modulation signal features based on the target high-order cumulant features; and a prediction module 640 for predicting the modulation mode of the target modulation signal features using a signal modulation recognition model, so as to obtain the modulation mode of the target communication signal.

[0150] As one possible implementation, the extraction module 630 is used to acquire the target baseband signal IQ data of the target single-sideband signal under any modulation scheme; normalize the target baseband signal IQ data of the target single-sideband signal under any modulation scheme to obtain the normalized target baseband signal IQ data of the target single-sideband signal under any modulation scheme; and extract the target high-order cumulative quantity features of the target baseband signal IQ data of the target single-sideband signal under any modulation scheme after normalization.

[0151] As one possible implementation, the extraction module 630 is used to perform a fast Fourier transform on the target single-sideband signal under any modulation mode to obtain the target center frequency and target bandwidth of the target single-sideband signal under any modulation mode; to perform a quadrature down-conversion on the target single-sideband signal under any modulation mode using the target center frequency to obtain the initial in-phase component and initial quadrature component of the target baseband signal of the target single-sideband signal under any modulation mode; and to filter the initial in-phase component and initial quadrature component of the target baseband signal of the target single-sideband signal under any modulation mode using the target bandwidth to obtain the target baseband signal IQ data of the target single-sideband signal under any modulation mode.

[0152] As one possible implementation, the target high-order cumulant features include multiple target even-order cumulant features. The extraction module 630 is used to obtain the ratio between the multiple target even-order cumulant features; and to determine the target modulation signal features based on the ratio between the multiple target even-order cumulant features.

[0153] The signal modulation recognition device of this disclosure acquires a target communication signal; collects the target single-sideband signal of the target communication signal under multiple modulation modes; extracts the target high-order cumulant features of the target single-sideband signal under any modulation mode; determines the target modulation signal features based on the target high-order cumulant features; and uses a signal modulation recognition model to predict the modulation mode of the target modulation signal features to obtain the modulation mode of the target communication signal. Thus, by using a trained signal modulation recognition model, the modulation mode of the communication signal can be accurately and efficiently identified.

[0154] It should be noted that the foregoing explanation of the signal modulation recognition method embodiment also applies to the signal modulation recognition device of this embodiment, and will not be repeated here.

[0155] To achieve the above embodiments, this application also proposes an electronic device, such as... Figure 7 As shown, Figure 7 This is a block diagram illustrating an electronic device for training a signal modulation recognition model or for signal modulation recognition, according to an exemplary embodiment.

[0156] like Figure 7 As shown, the above-mentioned electronic device 700 includes:

[0157] The memory 710 and the processor 720 are connected by a bus 730, which connects different components (including the memory 710 and the processor 720). The memory 710 stores a computer program, and when the processor 720 executes the program, it implements the training method or signal modulation recognition method of the signal modulation recognition model described in the embodiments of this disclosure.

[0158] Bus 730 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. Examples of these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0159] Electronic device 700 typically includes a variety of electronic device readable media. These media can be any available media that can be accessed by electronic device 700, including volatile and non-volatile media, removable and non-removable media.

[0160] The memory 710 may also include computer system readable media in the form of volatile memory, such as random access memory (RAM) 740 and / or cache memory 750. The electronic device 700 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 760 can be used to read and write non-removable, non-volatile magnetic media (…). Figure 7 Not shown; usually referred to as a "hard drive"). Although Figure 7 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 730 via one or more data media interfaces. Memory 710 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of this disclosure.

[0161] A program / utility 780 having a set (at least one) of program modules 770 may be stored in, for example, memory 710. Such program modules 770 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 770 typically perform the functions and / or methods described in the embodiments of this disclosure.

[0162] Electronic device 700 can also communicate with one or more external devices 790 (e.g., keyboard, pointing device, display, etc.), and with one or more devices that enable a user to interact with electronic device 700, and / or with any device that enables electronic device 700 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through input / output (I / O) interface 792. Furthermore, electronic device 700 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) through network adapter 793. Figure 7 As shown, network adapter 793 communicates with other modules of electronic device 700 via bus 730. It should be understood that, although... Figure 7 As not shown in the diagram, other hardware and / or software modules may be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0163] The processor 720 executes various functional applications and data processing by running programs stored in the memory 710.

[0164] It should be noted that the implementation process and technical principles of the electronic device in this embodiment are explained in the foregoing embodiments, and will not be repeated here.

[0165] To implement the above embodiments, this disclosure also proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the training method or signal modulation recognition method of the signal modulation recognition model described in the above embodiments.

[0166] To implement the above embodiments, this disclosure also provides a computer program product, which, when the instruction processor in the computer program product is executed, performs the training method or signal modulation recognition method of the signal modulation recognition model described in the above embodiments.

[0167] In the description of this specification, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0168] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0169] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A training method for a signal modulation recognition model, characterized in that, include: Collect sample single-sideband signals of sample communication signals under multiple modulation modes; Extract the sample higher-order cumulant features of the sample single-sideband signal under any modulation mode, and determine the sample modulation signal features based on the sample higher-order cumulant features. The signal modulation recognition model is used to predict the modulation mode of the sample modulation signal features to obtain a predicted label for the sample modulation signal features; wherein, the predicted label is used to indicate the predicted modulation mode corresponding to the sample modulation signal features; The signal modulation recognition model is trained based on the difference between the predicted modulation scheme and any of the modulation schemes.

2. The method according to claim 1, characterized in that, The extraction of sample high-order cumulant features of sample single-sideband signals under any modulation mode includes: Obtain the sample baseband signal IQ data of the sample single-sideband signal under any of the modulation methods; The sample baseband signal IQ data of the sample single-sideband signal under any of the modulation schemes is normalized to obtain the normalized sample baseband signal IQ data of the sample single-sideband signal under any of the modulation schemes. Extract the sample high-order cumulative features of the sample baseband signal IQ data of the sample single-sideband signal under any of the normalized modulation modes.

3. The method according to claim 2, characterized in that, The acquisition of sample baseband signal IQ data of the sample single-sideband signal under any of the modulation schemes includes: Perform a Fast Fourier Transform on the sample single-sideband signal under any of the modulation schemes to obtain the sample center frequency and sample bandwidth of the sample single-sideband signal under any of the modulation schemes. The sample single-sideband signal under any modulation mode is orthogonally down-converted using the sample center frequency to obtain the initial in-phase component and initial quadrature component of the sample baseband signal of the sample single-sideband signal under any modulation mode. The initial in-phase component and initial quadrature component of the sample baseband signal of the sample single-sideband signal under any modulation mode are filtered using the sample bandwidth to obtain the sample baseband signal IQ data of the sample single-sideband signal under any modulation mode.

4. The method according to claim 1, characterized in that, The higher-order cumulant features of the samples include multiple even-order cumulant features of the samples. Determining the sample modulation signal features based on the higher-order cumulant features of the samples includes: Obtain the ratio between the even-order cumulative features of the multiple samples; The sample modulation signal features are determined based on the ratio between the even-order cumulative features of the multiple samples.

5. The method according to claim 1, characterized in that, Before using the signal modulation recognition model to predict the modulation mode of the sample modulation signal features to obtain the predicted label of the sample modulation signal features, the method further includes: The features of the sample modulation signal are labeled to obtain the labeled tags of the sample modulation signal features; The label is used to indicate the actual modulation method corresponding to the characteristics of the sample modulation signal.

6. The method according to claim 5, characterized in that, The step of training the signal modulation recognition model based on the difference between the predicted modulation scheme and any of the modulation schemes includes: The loss function value is determined based on the difference between the predicted label corresponding to the predicted modulation scheme and the labeled label corresponding to any modulation scheme. The signal modulation recognition model is trained based on the loss function value.

7. The method according to claim 6, characterized in that, The step of determining the loss function value based on the difference between the predicted label corresponding to the predicted modulation scheme and the labeled label corresponding to any modulation scheme includes: Based on the difference between the prediction label corresponding to the prediction modulation scheme and the annotation label corresponding to any modulation scheme, a sub-loss function value for any modulation scheme is generated; The sub-loss function values ​​under each modulation scheme are weighted and summed to generate the loss function value.

8. A signal modulation recognition method, characterized in that, include: Acquire target communication signals; Collect the target single-sideband signal of the target communication signal under multiple modulation modes; Extract the target high-order cumulant features of the target single-sideband signal under any modulation mode, and determine the target modulation signal features based on the target high-order cumulant features; A signal modulation recognition model is used to predict the modulation mode of the target modulation signal features in order to obtain the modulation mode of the target communication signal.

9. The method according to claim 8, characterized in that, The extraction of the target high-order cumulant features of the target single-sideband signal under any modulation mode includes: Acquire the target baseband signal IQ data of the target single-sideband signal under any of the modulation schemes; The target baseband signal IQ data of the target single-sideband signal under any of the modulation schemes is normalized to obtain the normalized target baseband signal IQ data of the target single-sideband signal under any of the modulation schemes. Extract the target high-order cumulative features of the target baseband signal IQ data of the target single-sideband signal under any of the normalized modulation modes.

10. The method according to claim 9, characterized in that, The acquisition of the target baseband signal IQ data of the target single-sideband signal under any of the modulation schemes includes: Perform a Fast Fourier Transform on the target single-sideband signal under any of the modulation schemes to obtain the target center frequency and target bandwidth of the target single-sideband signal under any of the modulation schemes. The target single-sideband signal under any modulation mode is orthogonally down-converted using the target center frequency to obtain the initial in-phase component and initial quadrature component of the target baseband signal under any modulation mode. The initial in-phase component and initial quadrature component of the target baseband signal of the target single-sideband signal under any modulation mode are filtered using the target bandwidth to obtain the target baseband signal IQ data of the target single-sideband signal under any modulation mode.

11. The method according to claim 8, characterized in that, The target higher-order cumulant features include multiple target even-order cumulant features. Determining the target modulation signal features based on the target higher-order cumulant features includes: Obtain the ratio between the multiple target even-order cumulative features; The target modulation signal features are determined based on the ratios between the multiple target even-order cumulative features.

12. A training device for a signal modulation recognition model, characterized in that, include: The acquisition module acquires sample single-sideband signals of the sample communication signal under multiple modulation modes; The extraction module is used to extract the sample higher-order cumulant features of the sample single-sideband signal under any modulation mode, and determine the sample modulation signal features based on the sample higher-order cumulant features. The prediction module is used to predict the modulation mode of the sample modulation signal features using the signal modulation recognition model, so as to obtain a prediction label of the sample modulation signal features; wherein, the prediction label is used to indicate the predicted modulation mode corresponding to the sample modulation signal features; The training module is used to train the signal modulation recognition model based on the difference between the predicted modulation scheme and any of the modulation schemes.

13. A signal modulation identification device, characterized in that, include: The acquisition module is used to acquire the target communication signal; The acquisition module is used to acquire the target single-sideband signal of the target communication signal under multiple modulation modes; The extraction module is used to extract the target high-order cumulant features of the target single-sideband signal under any modulation mode, and determine the target modulation signal features based on the target high-order cumulant features; The prediction module is used to predict the modulation mode of the target modulation signal features using a signal modulation recognition model, so as to obtain the modulation mode of the target communication signal.

14. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the training method of the signal modulation recognition model as described in any one of claims 1-7, or implements the signal modulation recognition method as described in any one of claims 8-11.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the training method of the signal modulation recognition model as described in any one of claims 1-7, or implements the signal modulation recognition method as described in any one of claims 8-11.

Images