Communication signal modulation identification method and related equipment

Abstract

The invention relates to the technical field of communication signal modulation, in particular to a communication signal modulation identification method and related equipment, and the method comprises the steps: firstly obtaining a to-be-identified communication signal, and inputting the to-be-identified communication signal into a pre-trained modulation identification model; according to the model, an MK-SVM classifier based on a genetic algorithm is adopted, optimal parameters of the classifier are obtained through genetic algorithm optimization, and the parameters are input into the MK-SVM classifier to be updated. In the identification process, the system extracts the features of the communication signals to be identified, sends the features to the updated MK-SVM classifier, and classifies the signals according to the features, thereby accurately identifying the modulation types of the communication signals. The method ensures the accuracy and high efficiency of signal modulation type identification.

Description

Technical Field

[0001] The present invention relates to the technical field of communication signal modulation, and in particular to a communication signal modulation identification method and related equipment. Background Art

[0002] In the field of communication technology, modulation recognition, as a key link in signal processing, lies between signal detection and information demodulation. Its core goal is to accurately determine the modulation type of the received signal under non-cooperative reception conditions, thereby laying a solid information foundation for subsequent communication tasks such as demodulation and interference.

[0003] Modulation recognition methods based on feature extraction primarily involve two steps: feature extraction and classification. Commonly used signal modulation features include instantaneous features, high-order cumulants, entropy features, and spectral features. Commonly used classifiers include decision tree classifiers and support vector machines (SVMs).

[0004] Existing technology provides a feature selection method based on rough set theory. This method selects four noise-insensitive features and uses these features to train a single-core SVM classifier. This method successfully recognizes 10 modulation types, including 2ASK (Amplitude Shift Keying), 4ASK, 8ASK, 2FSK (Frequency Shift Keying), 4FSK, 8FSK, 2PSK (Phase Shift Keying), 4PSK, 8PSK, and 16QAM (Quadrature Amplitude Modulation). However, the single-core SVM classifier used in this method has certain limitations in the types of modulation types it can recognize, making its applicability relatively narrow.

[0005] Taking the sum of the variances of feature parameters as the criterion for judging the sensitivity of features to noise, 10 feature parameters were selected and a decision tree was used as the classifier. Fifteen communication signal modulation modes, including BPSK (Binary Phase Shift Keying), QPSK (Quadrature Phase Shift Keying), 8PSK, π / 4-QPSK, OQPSK (Offset Quadrature Phase Shift Keying), 8QAM, 64QAM, MSK (Minimum Shift Keying), GMSK (Gaussian Minimum Shift Keying), 2ASK, 4ASK, 2FSK, 4FSK, 32APSK (32 Amplitude Phase Shift Keying), and 64APSK, were identified. However, this method faces challenges in practical applications, as it requires selecting a threshold for each branch node of the decision tree. The threshold calculation algorithm is relatively complex, and the performance of the decision tree classifier is closely related to the threshold selection, which to some extent affects the stability of the recognition method. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a communication signal modulation identification method and related equipment in response to the above-mentioned deficiencies in the prior art, so as to solve the technical problems that the current signal modulation identification method has low stability and large limitations.

[0007] The purpose of the present invention is achieved by the following technical solutions: In a first aspect, the present invention provides a communication signal modulation identification method, comprising: Obtain a communication signal to be identified, input the communication signal into a pre-trained modulation recognition model, extract features of the communication signal to be identified, classify the modulation type according to the features, and identify the modulation type of the communication signal to be identified; The modulation recognition model adopts an MK-SVM (Multi-Kernel Support Vector Machine) classifier based on a genetic algorithm. The MK-SVM classifier obtains optimal parameters according to the genetic algorithm, inputs the optimal parameters into the MK-SVM classifier, updates the MK-SVM classifier, and identifies and classifies the communication signal to be identified based on the updated MK-SVM classifier to obtain the modulation type of the communication signal.

[0008] As a further improvement of the present invention, the modulation type of the communication signal includes at least BPSK (Binary Phase Shift Keying, binary phase shift keying), QPSK (Quadrature Phase Shift Keying, orthogonal phase shift keying), 8PSK (8 Phase Shift Keying, 8 phase shift keying), / 4-QPSK (π / 4-Quadrature Phase Shift Keying), OQPSK (Offset Quadrature Phase Shift Keying), 8QAM (8 Quadrature Amplitude Modulation), 64QAM (64 Quadrature Amplitude Modulation), MSK (Minimum Shift Keying), GMSK (Gaussian Minimum Shift Keying), 2ASK (Binary Amplitude Shift Keying), 4ASK (Quadrature Amplitude Shift Keying), 2FSK (Binary Frequency Shift Keying), 4FSK (Quadrature Frequency Shift Keying), 32APSK (32 Amplitude Phase Shift Keying) and 64APSK (64 Amplitude Phase Shift Keying, 64 amplitude phase keying).

[0009] As a further improvement of the present invention, the kernel function of the MK-SVM classifier is a composite kernel function composed of a polynomial kernel function and a Gaussian kernel function, and the composite kernel function is:

[0010] Where, is a composite kernel function; is the first eigenvector; is the second eigenvector; is the weight parameter; is the inner product of the first eigenvector and the second eigenvector; is the bandwidth parameter; is the square of the Euclidean distance between the first eigenvector and the second eigenvector; is the Gaussian kernel function part; is the polynomial kernel function part; is the bias function of the polynomial kernel function; is the order of the polynomial.

[0011] As a further improvement of the present invention, the corresponding training steps of the MK-SVM classifier based on the genetic algorithm include: S1. Acquire communication signal samples of different modulation types as a sample data set, and calculate modulation characteristic parameters based on the communication signal samples in the sample data set; S2. normalizing the modulation characteristic parameters and using the normalized modulation characteristic parameters as a feature data set; and distributing the feature data set into a training set and a test set according to a set ratio; S3, constructing an initial MK-SVM classifier, and randomly generating an initial population as the parameters of the MK-SVM classifier; S4, iteratively train the initial MK-SVM classifier using the training set; S5, perform a five-fold cross validation on the current MK-SVM classifier, and obtain the average accuracy of the MK-SVM classifier as the fitness function; S6. Using a genetic algorithm to perform selection, crossover, and mutation operations on the parameters of the current MK-SVM classifier, a new generation population is obtained as the parameters of the MK-SVM classifier in the next iteration process; S7, repeating steps S5 and S6 until the iteration stop condition is met, and taking the parameters of the current MK-SVM classifier as the optimal parameters; S8. Input the optimal parameters into the MK-SVM classifier, test the MK-SVM classifier using the test set, and obtain the classification accuracy of each modulation type.

[0012] As a further improvement of the present invention, the modulation characteristic parameters include instantaneous characteristics and non-instantaneous characteristics: The instantaneous characteristics include the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the signal non-weak segment, the standard deviation of the nonlinear component of the direct instantaneous phase, and the ratio of the square of the mean value of the instantaneous phase to the variance; The non-instantaneous features include four high-order cumulant features, one power spectrum exponential entropy feature and two quadratic spectrum line features.

[0013] As a further improvement of the present invention, the instantaneous characteristics include:

[0014]

[0015]

[0016] Where, is the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the non-weak segment; is the number of non-weak segment signal values; is the average value of the instantaneous amplitude of the signal; i is the communication signal sample point; The threshold value for judging weak signals; is the decentered linear instantaneous phase; To meet the screening conditions The sum of the decentered linear instantaneous phases of ; is the standard deviation of the nonlinear component of the direct instantaneous phase; is the ratio of the square of the mean value of the instantaneous phase to the variance; is the square of the average value of the instantaneous phase; is the variance of the instantaneous phase.

[0017] As a further improvement of the present invention, the power spectrum index entropy feature is:

[0018]

[0019] Where, is the power spectrum index entropy feature; is the probability density function value of the i-th frequency point; is the index term; is the power spectrum component of the i-th frequency point; is the sum of the power spectrum components; is the total number of frequency points involved in the calculation; i is the frequency point of the communication signal.

[0020] In a second aspect, the present invention provides a communication signal modulation identification system for implementing the above-mentioned communication signal modulation identification method, comprising: A data acquisition module, used to acquire the communication signal to be identified; The modulation recognition module is used to input the communication signal into a pre-trained modulation recognition model, extract the characteristics of the communication signal to be identified, and classify it according to the characteristics to obtain the modulation type of the communication signal to be identified; The modulation recognition model adopts an MK-SVM classifier based on a genetic algorithm. The MK-SVM classifier obtains optimal parameters according to the genetic algorithm, inputs the optimal parameters into the MK-SVM classifier, updates the MK-SVM classifier, and classifies the communication signal to be identified according to the updated MK-SVM classifier, and finally obtains the modulation type of the communication signal.

[0021] In a third aspect, the present invention provides a computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions that, when executed by a computing device, enable the computing device to perform the above-mentioned communication signal modulation identification method.

[0022] In a fourth aspect, the present invention provides a computing device, comprising: One or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include means for executing the above-mentioned communication signal modulation identification method.

[0023] The present invention provides a method for communication signal modulation identification that uses a genetic algorithm to search for the optimal solution in parameter space, avoiding the local optimum problem inherent in traditional SVMs due to random parameter initialization. This allows the kernel function parameters of the MK-SVM classifier to better match signal characteristics, significantly improving the classification accuracy of modulation types, particularly in low signal-to-noise ratio (SNR) or complex channel environments. By combining the feature mapping capabilities of multiple kernel functions (such as linear, polynomial, and RBF kernels), MK-SVM can simultaneously capture both linear and nonlinear signal characteristics.

[0024] The kernel function parameters optimized by the genetic algorithm can more accurately map the original signal features to the high-dimensional space, making the features of different modulation types more separable in the high-dimensional space.

[0025] Furthermore, the modulation recognition method based on multi-core support vector machine constructs a weighted composite kernel function. By combining the advantages of the polynomial kernel and Gaussian kernel, the feature extraction results are fused at the kernel level to generate a mapping space with better separability and stronger data interpretation capability.

[0026] Furthermore, the present invention employs a genetic algorithm to optimize multiple parameters, including the composite kernel weight coefficients, kernel parameters, and SVM hyperparameters, during the multi-core SVM classifier modeling process. Compared to traditional parameter optimization methods, genetic algorithms can search the parameter space in a targeted manner, avoiding the need to evaluate all possible combinations. This makes them suitable for optimizing high-dimensional parameters, as in the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0028] Figure 1 1 is a flow chart of a method for identifying communication signal modulation according to an embodiment of the present invention; Figure 2 This is the recognition result of 15 types of modulation signals using the GA (Genetic Algorithm)-MK-SVM method in an embodiment of the present invention; Figure 3 The results are the average recognition accuracy of 15 types of modulated signals under the conditions of using GA optimized polynomial kernel SVM, GA optimized Gaussian kernel SVM and GA optimized multi-core SVM classifiers in the embodiments of the present invention respectively; Figure 4 2 is a schematic structural diagram of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose and technical solution of the present invention clearer and easier to understand, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0030] The technical solutions of the present invention will be described clearly and completely below with reference to the accompanying drawings and specific embodiments. The described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0031] Example 1 This embodiment provides a communication signal modulation identification method, and the specific implementation steps are as follows.

[0032] First, in this embodiment, the communication signal to be identified is obtained first, and the communication signal is input into a pre-trained modulation recognition model. By extracting the features of the communication signal to be identified, the modulation type is classified according to the features, and the modulation type of the communication signal to be identified is identified.

[0033] The modulation recognition model uses an MK-SVM classifier based on a genetic algorithm. Specifically, the MK-SVM classifier obtains optimal parameters based on the genetic algorithm. After the optimal parameters are input into the MK-SVM classifier, the MK-SVM classifier is updated. The communication signal to be identified is then classified using the kernel function in the updated MK-SVM classifier to determine the modulation type of the communication signal.

[0034] In this embodiment, the modulation recognition model can recognize the modulation types of communication signals, including at least BPSK, QPSK, 8PSK, / 4-QPSK, OQPSK, 8QAM, 64QAM, MSK, GMSK, 2ASK, 4ASK, 2FSK, 4FSK, 32APSK and 64APSK.

[0035] The kernel function of the MK-SVM classifier is a composite kernel function composed of a polynomial kernel function and a Gaussian kernel function. The composite kernel function is:

[0036] Where, is a composite kernel function; is the first eigenvector; is the second eigenvector; is the weight parameter; is the inner product of the first eigenvector and the second eigenvector; is the bandwidth parameter; is the square of the Euclidean distance between the first eigenvector and the second eigenvector; is the Gaussian kernel function part; is the polynomial kernel function part; is the bias function of the polynomial kernel function; is the order of the polynomial.

[0037] The training process of the modulation recognition model specifically includes: S1. Acquire communication signal samples of different modulation types as a sample data set, and calculate modulation characteristic parameters based on the communication signal samples in the sample data set.

[0038] Among them, this embodiment uses software simulation to generate communication signal samples of 15 modulation types, including BPSK, QPSK, 8PSK, π / 4-QPSK, OQPSK, 8QAM, 64QAM, MSK, GMSK, 2ASK, 4ASK, 2FSK, 4FSK, 32APSK, and 64APSK under different signal-to-noise ratios.

[0039] The modulation characteristic parameters include instantaneous characteristics and non-instantaneous characteristics. Specifically, the instantaneous characteristics include the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the signal in the non-weak segment, the standard deviation of the nonlinear component of the direct instantaneous phase, and the ratio of the square of the mean value of the instantaneous phase to the variance. The calculation expression of the instantaneous characteristics is:

[0040]

[0041]

[0042] Where, is the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the non-weak segment; is the number of non-weak segment signal values; is the average value of the instantaneous amplitude of the signal; i is the communication signal sample point; The threshold value for judging weak signals; is the decentered linear instantaneous phase; To meet the screening conditions The sum of the decentered linear instantaneous phases of ; is the standard deviation of the nonlinear component of the direct instantaneous phase; is the ratio of the square of the mean value of the instantaneous phase to the variance; is the square of the average value of the instantaneous phase; is the variance of the instantaneous phase.

[0043] The non-instantaneous features include four high-order cumulant features, one power spectrum exponential entropy feature and two quadratic spectral line features.

[0044] The calculation expressions of the four high-order cumulant features are:

[0045] in, is a high-order cumulant with order 4 and conjugate position 0; is the fourth-order mixing moment; is the second-order mixed moment; is a high-order cumulant with order 4 and conjugate position 2; is the fourth-order mixing moment; is the second-order mixed moment; is a high-order cumulant with an order of 6 and a conjugate position of 1; is the fourth-order mixing moment; is a high-order cumulant with an order of 6 and a conjugate position of 3; is the 6th order mixing moment, is a high-order cumulant, represents the order of the higher-order cumulant, represents the conjugate position, for of The specific expression is:

[0046] Where, Used to find the expected value, * represents the complex conjugate, is a complex random signal, is the complex conjugate of the complex random signal, represents the order of the higher-order cumulant, represents the conjugated position.

[0047] The calculation expression of the power spectrum index entropy feature is:

[0048]

[0049] Where, is the power spectrum index entropy feature; is the probability density function value of the i-th frequency point; is the index term; is the power spectrum component of the i-th frequency point; is the sum of the power spectrum components; is the total number of frequency points involved in the calculation; i is the frequency point of the communication signal.

[0050] The calculation expressions of the two quadratic spectral line characteristics are:

[0051] Where, represents the discrete Fourier transform of the complex signal of the quadratic spectrum, Indicates the maximum amplitude of the spectrum line in the left area of ​​the quadratic spectrum line, It is the maximum amplitude of the spectrum line in the right area of ​​the quadratic spectrum line.

[0052] S2. Normalize the modulation characteristic parameters and use the normalized modulation characteristic parameters as a feature data set; and distribute the feature data set into a training set and a test set according to a set ratio.

[0053] In this embodiment, the normalized modulation feature parameters are scaled to the range of [-1, +1]. The normalized feature data set is randomly distributed, with 70% of the samples used as a training set and 30% of the samples used as a test set.

[0054] S3. Construct the initial MK-SVM classifier and randomly generate the initial population as the parameters of the MK-SVM classifier. The initial population includes the penalty coefficient C of the SVM (Support Vector Machine) optimization function and the composite kernel function parameters (such as weight parameters , bandwidth parameters , bias function of polynomial kernel function , the order of the polynomial ), and also set the genetic algorithm population size, genetic generations, crossover rate, mutation rate and other parameter values, using real number coding.

[0055] S4, iteratively train the initial MK-SVM classifier using the training set; S5, perform a five-fold cross validation on the current MK-SVM classifier, and obtain the average accuracy of the MK-SVM classifier as the fitness function value; S6. Use a genetic algorithm to select, crossover, and mutate the parameters of the current MK-SVM classifier to obtain a new generation of population as the parameters of the MK-SVM classifier in the next iteration process; specifically, the following steps are performed: sorting the fitness in ascending order, selecting the individual with the highest fitness as the optimal individual, replacing the odd-numbered individuals in the population as the parent population, and keeping the even-numbered individuals unchanged as the mother population; using the optimal individual to crossover with individuals in the mother population to generate a new population; performing multi-point mutation on the new population to generate a sub-population, and calculating its fitness value; sorting the sub-population in ascending order of fitness, and merging it with the parent population, sorting the individuals in the merged population in ascending order of fitness value, and selecting individuals with the highest population size to constitute the next generation of population.

[0056] S7, repeating steps S5 and S6 until the iteration stop condition is met, and taking the parameters of the current MK-SVM classifier as the optimal parameters; S8. Input the optimal parameters into the MK-SVM classifier, test the MK-SVM classifier using the test set, and obtain the classification accuracy of each modulation type.

[0057] Example 2 In order to better verify the effectiveness of the communication signal modulation identification method in Example 1, this embodiment takes a simulated communication signal as an example to specifically illustrate Example 1. The specific steps of the example in this embodiment are as follows.

[0058] Simulate and generate BPSK, QPSK, 8PSK, Communication signals using 15 modulation types, including 4-QPSK, OQPSK, 8QAM, 64QAM, MSK, GMSK, 2ASK, 4ASK, 2FSK, 4FSK, 32APSK, and 64APSK, were generated using binary random symbol sequences with a symbol length of 128, a sampling multiple of 16, a symbol rate of 1 kbps, a carrier frequency of 2 kHz, a raised cosine roll-off factor of 0.35, and a sampling rate of 16 kHz. The channel model used a Gaussian white noise channel with a signal-to-noise ratio (SNR) range of -10 to 30 dB (in 2 dB steps). The number of samples for each modulation type at each SNR was 500, resulting in a total of 157,500 samples for the 15 modulation types at 21 SNRs.

[0059] Taking a BPSK modulated signal sample with a signal-to-noise ratio of 10 dB as an example, 10 modulation features are calculated.

[0060] Calculate the instantaneous characteristics 、 、 .

[0061] Calculate the high-order cumulant characteristics 、 、 、 .

[0062] Calculated 、 .

[0063] We traverse 21 signal-to-noise ratio conditions under 15 modulation modes and obtain a 10-dimensional feature vector for each sample.

[0064] All eigenvector values ​​are normalized, and the normalized feature data sets are randomly allocated, with 70% as the training set, that is, the number of samples is 110,250, and 30% as the test set, that is, the number of samples is 94,500.

[0065] Step 3, MK-SVM parameter optimization based on GA: (1) Randomly generate the initial population of MK-SVM model parameters 、 、 、 and ,in C is the penalty coefficient of the SVM optimization function, 、 、 、 and are the composite kernel function parameters, and the genetic algorithm population size is set to 10, the genetic generation number is 50, the crossover rate is 0.8, and the mutation rate is 0.01.

[0066] (2) Perform 5-fold cross validation on MK-SVM using the training set; (3) The average classification accuracy of MK-SVM is used as the fitness function; (4) When the number of iterations reaches 50, exit the iteration and get the optimal parameter output , , , , =1.02; otherwise continue; (5) Perform selection, crossover, and mutation operations on the current population to obtain a new generation of population as the MK-SVM parameters, and jump to step (2).

[0067] The GA-MK-SVM classifier is constructed using the optimized parameters, and the test samples are sent to the classifier to obtain the classification accuracy of 15 modulation types, as shown in the following figure: Figure 2 shown.

[0068] Example 3 This embodiment provides a communication signal modulation identification system for implementing the communication signal modulation identification method described in Embodiments 1 and 2. The system mainly includes: A data acquisition module, used to acquire the communication signal to be identified; The modulation recognition module is used to input the communication signal into a pre-trained modulation recognition model, extract the characteristics of the communication signal to be identified, and classify it according to the characteristics to obtain the modulation type of the communication signal to be identified; The modulation recognition model adopts the MK-SVM classifier based on the genetic algorithm. The MK-SVM classifier obtains the optimal parameters according to the genetic algorithm, inputs the optimal parameters into the MK-SVM classifier, updates the MK-SVM classifier, and classifies the communication signal to be identified according to the updated MK-SVM classifier, and finally obtains the modulation type of the communication signal.

[0069] In this embodiment, the modulation recognition model can recognize the modulation types of communication signals, including at least BPSK, QPSK, 8PSK, / 4-QPSK, OQPSK, 8QAM, 64QAM, MSK, GMSK, 2ASK, 4ASK, 2FSK, 4FSK, 32APSK and 64APSK.

[0070] The kernel function of the MK-SVM classifier is a composite kernel function composed of a polynomial kernel function and a Gaussian kernel function. The composite kernel function is:

[0071] Where, is a composite kernel function; is the first eigenvector; is the second eigenvector; is the weight parameter; is the inner product of the first eigenvector and the second eigenvector; is the bandwidth parameter; is the square of the Euclidean distance between the first eigenvector and the second eigenvector; is the Gaussian kernel function part; is the polynomial kernel function part; is the bias function of the polynomial kernel function; is the order of the polynomial.

[0072] In addition, this embodiment also includes a model training module, and the processing steps of this module include: S1. Acquire communication signal samples of different modulation types as a sample data set, and calculate modulation characteristic parameters based on the communication signal samples in the sample data set.

[0073] Among them, this embodiment uses software simulation to generate communication signal samples of 15 modulation types, including BPSK, QPSK, 8PSK, π / 4-QPSK, OQPSK, 8QAM, 64QAM, MSK, GMSK, 2ASK, 4ASK, 2FSK, 4FSK, 32APSK, and 64APSK under different signal-to-noise ratios.

[0074] The modulation characteristic parameters include instantaneous characteristics and non-instantaneous characteristics. Specifically, the instantaneous characteristics include the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the signal in the non-weak segment, the standard deviation of the nonlinear component of the direct instantaneous phase, and the ratio of the square of the mean value of the instantaneous phase to the variance. The calculation expression of the instantaneous characteristics is:

[0075]

[0076]

[0077] Where, is the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the non-weak segment; is the number of non-weak segment signal values; is the average value of the instantaneous amplitude of the signal; i is the communication signal sample point; The threshold value for judging weak signals; is the decentered linear instantaneous phase; To meet the screening conditions The sum of the decentered linear instantaneous phases of ; is the standard deviation of the nonlinear component of the direct instantaneous phase; is the ratio of the square of the mean value of the instantaneous phase to the variance; is the square of the average value of the instantaneous phase; is the variance of the instantaneous phase.

[0078] The non-instantaneous features include four high-order cumulant features, one power spectrum exponential entropy feature and two quadratic spectral line features.

[0079] The calculation expressions of the four high-order cumulant features are:

[0080] in, is a high-order cumulant with order 4 and conjugate position 0; is the fourth-order mixing moment; is the second-order mixed moment; is a high-order cumulant with order 4 and conjugate position 2; is the fourth-order mixing moment; is the second-order mixed moment; is a high-order cumulant with an order of 6 and a conjugate position of 1; is the fourth-order mixing moment; is a high-order cumulant with an order of 6 and a conjugate position of 3; is the 6th order mixing moment, is a high-order cumulant, represents the order of the higher-order cumulant, represents the conjugate position, for of The specific expression is:

[0081] Where, is the expected value calculator, and * represents the complex conjugate.

[0082] The calculation expression of the power spectrum index entropy feature is:

[0083]

[0084] Where, is the power spectrum index entropy feature; is the probability density function value of the i-th frequency point; is the index term; is the power spectrum component of the i-th frequency point; is the sum of the power spectrum components; is the total number of frequency points involved in the calculation; i is the frequency point of the communication signal.

[0085] The calculation expressions of the two quadratic spectral line characteristics are:

[0086] Where, represents the discrete Fourier transform of the complex signal of the quadratic spectrum, Respectively represent the maximum amplitude of the spectrum line in the left and right areas of the quadratic spectrum line, where is the maximum amplitude of the spectrum line in the left area of ​​the quadratic spectrum line, It is the maximum amplitude of the spectrum line in the right area.

[0087] S2. Normalize the modulation characteristic parameters and use the normalized modulation characteristic parameters as a feature data set; and distribute the feature data set into a training set and a test set according to a set ratio.

[0088] In this embodiment, the normalized modulation feature parameters are scaled to the range of [-1, +1]. The normalized feature data set is randomly distributed, with 70% of the samples used as a training set and 30% of the samples used as a test set.

[0089] S3. Construct the initial MK-SVM classifier and randomly generate the initial population as the parameters of the MK-SVM classifier. The initial population includes the penalty coefficient C of the SVM optimization function and the composite kernel function parameters (such as the weight parameter , bandwidth parameters , bias function of polynomial kernel function , the order of the polynomial ), and also set the genetic algorithm population size, genetic generations, crossover rate, mutation rate and other parameter values.

[0090] S4, using the training set to iteratively train the composite kernel function in the initial MK-SVM classifier; S5, perform a five-fold cross validation on the current MK-SVM classifier, and obtain the average accuracy of the MK-SVM classifier as the fitness function; S6. Using a genetic algorithm to perform selection, crossover, and mutation operations on the parameters of the current MK-SVM classifier, a new generation population is obtained as the parameters of the MK-SVM classifier in the next iteration process; S7, repeating steps S5 and S6 until the iteration stop condition is met, and taking the parameters of the current MK-SVM classifier as the optimal parameters; S8. Input the optimal parameters into the MK-SVM classifier, test the MK-SVM classifier using the test set, and obtain the classification accuracy of each modulation type.

[0091] Example 4 In another embodiment of the present invention, a computer-readable storage medium is provided as a storage component within a terminal device, the function of which is to store programs and data. It should be noted that the computer-readable storage medium herein encompasses not only the built-in storage component of the terminal device, but also the extended storage component supported by the device. Essentially, it is a tangible medium that can contain or store programs that can be called by, or run in conjunction with, an instruction execution system, device, or component. This storage medium provides a storage area for the terminal's operating system and stores one or more instructions suitable for loading and executing by the processor. These instructions can constitute one or more computer programs containing program code.

[0092] In particular, examples (a non-exclusive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable magnetic disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, optical fiber, a portable optical disc read-only memory, an optical storage device, a magnetic storage device, or any reasonable combination of the foregoing.

[0093] The storage medium may also include a data signal transmitted as part of a baseband portion or carrier wave, which carries readable program code. Such a transmitted data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any reasonable combination of the two. In addition, computer-readable storage media may also refer to other readable media other than traditional readable storage media, which are capable of sending, transmitting, or transmitting programs for use by or in conjunction with an instruction execution system, device, or device. The program code on the storage medium may be transmitted via any suitable medium, including but not limited to wireless, wired, optical cable, or any reasonable combination thereof.

[0094] The program code used to implement the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as "C." The program code can be executed entirely on the user's computing device, partially on the user's device as a standalone software package, partially distributed across the user's device and a remote computing device, or entirely on a remote computing device or server. When a remote computing device is involved, the device may be connected to the user's computing device via any type of network, such as a local area network or wide area network, or connected to an external computing device via the Internet through an Internet service provider.

[0095] The processor is capable of loading and executing one or more instructions stored in a computer-readable storage medium to implement corresponding steps of the communication signal modulation identification method described in Example 1.

[0096] Example 5 Figure 4 A schematic diagram of a computer device provided in one embodiment of the present invention.

[0097] See also Figure 4 The terminal device is a computer device. The computer device 60 of this embodiment includes: a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable by the processor 61. When the computer program 63 is executed by the processor 61, the communication signal modulation identification method of the embodiment is implemented. To avoid repetition, the details are not repeated here. Alternatively, when the computer program 63 is executed by the processor 61, the functions of each model / unit in the computing system of the communication signal modulation identification method of the embodiment are implemented. To avoid repetition, the details are not repeated here.

[0098] The computer device 60 may be a desktop computer, a notebook computer, a PDA, a cloud server, or other computing devices. The computer device 60 may include, but is not limited to, a processor 61 and a memory 62. It will be understood by those skilled in the art that Figure 4 This is only an example of the computer device 60 and does not constitute a limitation of the computer device 60. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0099] The processor 61 may be a central processing unit (CPU), other general-purpose processors, central processing units (CPUs), graphics processors (GPUs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), other programmable logic devices, discrete gate or transistor logic devices, quantum computing-based data processing logic, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0100] The memory 62 may be an internal storage unit of the computer device 60, such as a hard disk or memory of the computer device 60. The memory 62 may also be an external storage device of the computer device 60, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 60.

[0101] Furthermore, the memory 62 may include both an internal storage unit of the computer device 60 and an external storage device. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 may also be used to temporarily store data that has been output or is about to be output.

[0102] Any reference to memory, database, or other media used in the various embodiments provided herein may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0103] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

Claims

1. A communication signal modulation identification method, characterized in that: include: Obtain a communication signal to be identified, input the communication signal into a pre-trained modulation recognition model, extract features of the communication signal to be identified, classify the modulation type according to the features, and identify the modulation type of the communication signal to be identified; The modulation recognition model adopts an MK-SVM classifier based on a genetic algorithm. The MK-SVM classifier obtains optimal parameters according to the genetic algorithm, inputs the optimal parameters into the MK-SVM classifier, updates the MK-SVM classifier, and identifies and classifies the communication signal to be identified according to the updated MK-SVM classifier to obtain the modulation type of the communication signal.

2. The communication signal modulation identification method according to claim 1, characterized in that: The modulation type of the communication signal includes at least BPSK, QPSK, 8PSK, / 4-QPSK, OQPSK, 8QAM, 64QAM, MSK, GMSK, 2ASK, 4ASK, 2FSK, 4FSK, 32APSK and 64APSK.

3. The communication signal modulation identification method according to claim 1, characterized in that: The kernel function of the MK-SVM classifier is a composite kernel function composed of a polynomial kernel function and a Gaussian kernel function. The composite kernel function is: Where, is a composite kernel function; is the first eigenvector; is the second eigenvector; is the weight parameter; is the inner product of the first eigenvector and the second eigenvector; is the bandwidth parameter; is the square of the Euclidean distance between the first eigenvector and the second eigenvector; is the Gaussian kernel function part; is the polynomial kernel function part; is the bias function of the polynomial kernel function; is the order of the polynomial.

4. The communication signal modulation identification method according to claim 1, characterized in that: The corresponding training steps of the MK-SVM classifier based on genetic algorithm include: S1. Acquire communication signal samples of different modulation types as a sample data set, and calculate modulation characteristic parameters based on the communication signal samples in the sample data set; S2. normalizing the modulation characteristic parameters and using the normalized modulation characteristic parameters as a feature data set; and distributing the feature data set into a training set and a test set according to a set ratio; S3, constructing an initial MK-SVM classifier, and randomly generating an initial population as the parameters of the MK-SVM classifier; S4, iteratively train the initial MK-SVM classifier using the training set; S5, perform a five-fold cross validation on the current MK-SVM classifier, and obtain the average accuracy of the MK-SVM classifier as the fitness function; S6. Using a genetic algorithm to perform selection, crossover, and mutation operations on the parameters of the current MK-SVM classifier, a new generation population is obtained as the parameters of the MK-SVM classifier in the next iteration process; S7, repeating steps S5 and S6 until the iteration stop condition is met, and taking the parameters of the current MK-SVM classifier as the optimal parameters; S8. Input the optimal parameters into the MK-SVM classifier, test the MK-SVM classifier using the test set, and obtain the classification accuracy of each modulation type.

5. The communication signal modulation identification method according to claim 4, characterized in that: The modulation characteristic parameters include instantaneous characteristics and non-instantaneous characteristics: The instantaneous characteristics include the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the signal non-weak segment, the standard deviation of the nonlinear component of the direct instantaneous phase, and the ratio of the square of the mean value of the instantaneous phase to the variance; The non-instantaneous features include four high-order cumulant features, one power spectrum exponential entropy feature and two quadratic spectrum line features.

6. The communication signal modulation identification method according to claim 5, characterized in that: The instantaneous characteristics include: Where, is the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the non-weak segment; is the number of non-weak segment signal values; is the average value of the instantaneous amplitude of the signal; i is the communication signal sample point; The threshold value for judging weak signals; is the decentered linear instantaneous phase; To meet the screening conditions The sum of the decentered linear instantaneous phases of ; is the standard deviation of the nonlinear component of the direct instantaneous phase; is the ratio of the square of the mean value of the instantaneous phase to the variance; is the square of the average value of the instantaneous phase; is the variance of the instantaneous phase.

7. The communication signal modulation identification method according to claim 5, characterized in that: The power spectrum index entropy feature is: Where, is the power spectrum index entropy feature; is the probability density function value of the i-th frequency point; is the index term; is the power spectrum component of the i-th frequency point; is the sum of the power spectrum components; is the total number of frequency points involved in the calculation; i is the frequency point of the communication signal.

8. A communication signal modulation identification system, used to implement the communication signal modulation identification method according to any one of claims 1 to 7, characterized in that: include: A data acquisition module, used to acquire the communication signal to be identified; The modulation recognition module is used to input the communication signal into a pre-trained modulation recognition model, extract the characteristics of the communication signal to be identified, and classify it according to the characteristics to obtain the modulation type of the communication signal to be identified; The modulation recognition model adopts an MK-SVM classifier based on a genetic algorithm. The MK-SVM classifier obtains optimal parameters according to the genetic algorithm, inputs the optimal parameters into the MK-SVM classifier, updates the MK-SVM classifier, and classifies the communication signal to be identified according to the updated MK-SVM classifier, and finally obtains the modulation type of the communication signal.

9. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform the communication signal modulation identification method according to any one of claims 1 to 7.

10. A computing device, characterized in that include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include a method for executing the communication signal modulation identification method according to any one of claims 1 to 7.

Images