Adaptive integrated feature selection-based aliasing signal modulation identification method

Abstract

The invention relates to the technical field of modulation identification, in particular to an adaptive integrated feature selection-based aliasing signal modulation identification method, which comprises the following steps of: segmenting a received signal into a plurality of small signals by using a sliding window, determining an aliasing position and an aliasing ending position of the received signal based on a power matrix of each small signal, and extracting an aliasing signal; performing feature extraction on the aliasing signals, and constructing a feature set of the aliasing signals; selecting each feature in the feature set by using five feature selection methods including minimum redundancy and maximum correlation criterion selection, multi-cluster feature selection, correlation feature selection, recursive feature elimination selection and random forest feature importance selection, and performing preliminary screening to obtain five stable feature subsets; performing adaptive optimization on the five stable feature subsets by using an HGSO algorithm to obtain an optimal feature subset; and performing modulation identification based on the optimal feature subset to obtain a modulation identification result. According to the method, the efficiency and accuracy of aliasing signal modulation identification are effectively improved.

Description

Technical Field

[0001] Embodiments of the present application relate to the technical field of modulation recognition of communication signals, and particularly to a method for modulating and recognizing aliased signals with adaptive integrated feature selection. Background Art

[0002] As a key technology in a non-cooperative communication environment, modulation recognition of communication signals has a wide range of applications in many fields. Most of the current modulation recognition algorithms are aimed at the recognition of single-user modulation methods. However, with the booming development of the wireless communication industry, the surge in the number of access devices and the demand for spectrum resources has made the current electromagnetic environment often in a highly dense and interleaved situation in the time domain, frequency domain, and spatial domain. Signals from different transmitters overlap in the time-frequency domain. For example, both parties communicating using the Paired Carrier Multiple Access (PCMA) technology send signals using the same frequency band, time slot, and codeword at the same time. The third listening party can only receive signals that are aliased in the time-frequency domain. At this time, the modulation recognition algorithm for single signals cannot be directly applied to the modulation recognition of aliased signals. Therefore, how to design an efficient and low-complexity algorithm to correctly identify the modulation method of aliased signals has important engineering practical significance.

[0003] Li, Han, Zaerin and others analyzed the higher-order statistics and instantaneous characteristics of aliased signals, manually selected features with high discrimination based on the calculated theoretical values, and designed a classifier using the Support Vector Machine (SVM). The maximum number of aliased dual-signal modulation methods that can be recognized is 21. After Wu et al. performed fractional Fourier transform on aliased signals, they constructed features based on the detail components of different-layer wavelet decompositions, and set appropriate thresholds to identify the modulation methods of different aliased signals using a decision tree. However, the computational complexity of the features of this method is relatively high. Zhang et al. considered the different power ratios and small sample situations between the components of aliased signals, and added an attention layer and an LSTM layer on the basis of the capsule neural network to better capture the time-frequency characteristics and hierarchical structure of the signals, achieving a higher recognition accuracy at low signal-to-noise ratios. However, the complexity of the model constructed by this method is relatively high.

[0004] In summary, the inventors of the present application found that the currently proposed methods for modulating and recognizing aliased signals still have the following defects and deficiencies.

[0005] First, the aliased signals based on traditional feature extraction and analysis are single, and the aliasing patterns are relatively fixed. Most methods analyze two aliased signal components, and the maximum number of aliased signal components analyzed is three. In addition, the explored aliasing patterns are mainly complete aliasing in the time domain and frequency domain. Only a small part considers partially aliased signals in the frequency domain, and there is almost no analysis of partially aliased signals in the time domain.

[0006] Second, the classifiers designed by the recognition methods based on traditional feature extraction have poor universality. In fact, there are many factors affecting the characteristics of the modulation mode of aliased signals, including the types of modulation modes of component signals, aliasing patterns, power ratios between component signals, carrier frequency offsets, phase offsets, and signal-to-noise ratios, etc. Most of the currently proposed methods only consider the influence of a single factor, and the designed classifiers can only recognize aliased signals in fixed forms.

[0007] Third, the aliased signals analyzed by the recognition methods based on deep learning lack the application of expert knowledge. Most of them consider transforming the modulation recognition problem of aliased signals into the field of image recognition. Summary of the Invention

[0008] To solve the above technical problems, the embodiments of the present application propose a method for identifying the modulation of aliased signals with adaptive integrated feature selection, which can automatically locate the positions where the aliased signals start and end aliasing in the time domain, is applicable to the modulation mode recognition of time-domain aliased signals with multiple aliasing patterns and multi-factor changes, and effectively improves the efficiency and accuracy of the modulation recognition of aliased signals.

[0009] To achieve the above object, the embodiments of the present application propose a method for identifying the modulation of aliased signals with adaptive integrated feature selection. The method includes: using a sliding window to divide the received signal into several small signals, determining the positions where the received signal starts and ends aliasing based on the power matrix of each small signal, and extracting the aliased signal; performing feature extraction on the aliased signal to construct a feature set of the aliased signal. The features in the feature set of the aliased signal include 10 combined high-order moment features, 16 combined high-order cumulant features, 8 high-order moment features, 8 high-order cumulant features, and 4 instantaneous features; using five feature selection methods, namely minimum redundancy maximum correlation criterion selection, multi-cluster feature selection, relevant feature selection, recursive feature elimination selection, and random forest feature importance selection, to select each feature in the feature set, and initially screening to obtain five stable feature subsets; using the HGSO algorithm to perform adaptive optimization on the five stable feature subsets to obtain an optimal feature subset; performing modulation recognition based on the optimal feature subset to obtain a modulation recognition result.

[0010] To achieve the above object, an embodiment of the present application further provides an aliased signal modulation recognition system with adaptive integrated feature selection. The system includes: a sliding window positioning module, configured to divide the received signal into several small signals by using a sliding window, determine the position where the received signal starts to be aliased and the position where the aliasing ends based on the power matrix of each small signal, and extract the aliased signal; a feature set construction module, configured to extract features from the aliased signal and construct a feature set of the aliased signal. The features in the feature set of the aliased signal include 10 combined high-order moment features, 16 combined high-order cumulant features, 8 high-order moment features, 8 high-order cumulant features, and 4 instantaneous features; an integrated feature selection module, configured to use five feature selection methods, namely minimum redundancy maximum correlation criterion selection, multi-cluster feature selection, correlation feature selection, recursive feature elimination selection, and random forest feature importance selection, to select each feature in the feature set, and initially screen to obtain 5 stable feature subsets; an adaptive optimization module, configured to use the HGSO algorithm to adaptively optimize the 5 stable feature subsets to obtain an optimal feature subset; and a modulation recognition module, configured to perform modulation recognition based on the optimal feature subset to obtain a modulation recognition result.

[0011] To achieve the above object, an embodiment of the present application further provides an electronic device. The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor is enabled to execute an aliased signal modulation recognition method with adaptive integrated feature selection as described above.

[0012] To achieve the above object, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, it can implement an aliased signal modulation recognition method with adaptive integrated feature selection as described above.

[0013] An aliased signal modulation recognition method based on adaptive integrated feature selection proposed in this application first accurately locates the position where the received signal is aliased and the position where the aliasing ends by using a sliding window, so as to accurately extract the aliased signal. Next, a feature set containing 46 different features is constructed for the aliased signal. High-order moment features and high-order cumulant features are used to improve the discrimination, and combined high-order moment features and combined high-order cumulant features are used to eliminate the influence of power factors and phase jitters on high-order moments and high-order cumulants. After that, the advantages of three feature selection methods, namely filter, wrapper, and embedded, are comprehensively considered. Five feature selection methods, namely minimum redundancy maximum correlation criterion selection, multi-cluster feature selection, correlation feature selection, recursive feature elimination selection, and random forest feature importance selection, are used for feature selection, and the HGSO algorithm is used for adaptive optimization to obtain the optimal feature subset. Finally, modulation recognition is performed based on the optimal feature subset to obtain the modulation recognition result. The designed aliased signal modulation recognition method can quickly and accurately realize the modulation recognition of aliased signals when factors such as the type of modulation method of the aliased signal, the aliasing pattern, the power ratio between component signals, carrier frequency offset, phase offset, and signal-to-noise ratio change simultaneously, and has high stability and universality.

[0014] Optionally, the received signal is segmented into several small signals by using a sliding window. Based on the power matrix of each small signal, the position where the received signal is aliased and the position where the aliasing ends are determined, and the aliased signal is extracted, including:

[0015] Set the window size and sliding step of the sliding window according to the length of the received signal, so as to segment the received signal into n win small signals by using the set sliding window;

[0016] Calculate the power matrix of each small signal respectively, and reduce the number n win of small signals to a preset fixed value n fixed , and take the upper quantile for every n win / n fixed small signals to obtain a new power matrix;

[0017] Perform unsupervised clustering on the new power matrix by using the spectral density clustering algorithm;

[0018] Determine the position where the category changes, that is, the position Loc_p where the power changes in the new power matrix, and invert the position Loc_win of the original window according to Loc_p, so as to obtain the position where the received signal is aliased and the position where the aliasing ends;

[0019] Extract the aliased signal from the received signal according to the position where the aliasing occurs and the position where the aliasing ends.

[0020] Optionally, the 8 high-order moment features are m 20, m 40 , m 60 , m 80 , m 21 , m 42 , m 63 and m 84 , the eight high - order cumulant features are c 20 , c 21 , c 40 , c 41 , c 42 , c 60 , c 63 and c 8o ;

[0021] The calculation formulas of high - order moment features and high - order cumulant features are respectively:

[0022] m kx (τ1, …, τ k-1 ) = E{x(t)x(t + τ1)…x(t + τ k-1 )};

[0023] c kx (τ1, …, τ k-1 ) = cum[x(t), x(t + τ1), …, x(t + τ k-1 )];

[0024] Among them, x(t) represents a stationary random process, k represents the order, τ represents the time delay, and cum(·) represents the cumulant function.

[0025] Optionally, the combined high - order moment features are denoted as F moma , a = 1, 2, …, 10, and the combined high - order cumulant features are denoted as F cumb , b = 1, 2, …, 16;

[0026] The calculation formulas of the 10 combined high - order moment features are respectively:

[0027]

[0028] The calculation formulas of the 16 combined high - order cumulant features are respectively:

[0029] F cum1 = |c 40 | / |c 41 |; F cum2 = |c 40 | / |c 42 |; F cum3 = |c 41 | / |c 42 |;

[0030]

[0031] Optionally, the four instantaneous features are respectively the second-order statistical feature m of the instantaneous amplitude a , the maximum value feature γ of the instantaneous amplitude spectral density max , the standard deviation feature σ of the zero-centered normalized instantaneous amplitude of non-weak signals da and the standard deviation feature σ of the absolute value of the zero-centered normalized instantaneous amplitude aa ;

[0032] The calculation formulas for the four instantaneous features are respectively:

[0033]

[0034] γ max = max|DFT[A cn (i)]| 2 / N;

[0035]

[0036] where n is the number of sampling points, A(i) is the instantaneous amplitude of the signal, A cn (i)=A(i)-1, A cn (i) is the zero-centered normalized instantaneous amplitude of the signal, DFT(·) represents the discrete Fourier transform, a i is the threshold of the preset non-weak signal, and C is the number of non-weak signals.

[0037] Optionally, the stable feature subset screened by using the minimum redundancy maximum correlation criterion selection method is denoted as S mRMR , the stable feature subset screened by using the multi-cluster feature selection method is denoted as S MCFS , the stable feature subset screened by using the correlation feature selection method is denoted as S CBFS , the stable feature subset screened by using the recursive feature elimination selection method is denoted as S RFE , the stable feature subset screened by using the random forest feature importance selection method is denoted as S RT ;

[0038] After initially screening to obtain five stable feature subsets, the method further includes:

[0039] Using the random forest algorithm to determine the verification accuracy of each of the five stable feature subsets, the verification accuracy of S mRMR is denoted as Acc mRMR , the verification accuracy of S MCFS is denoted as Acc MCFS , the verification accuracy of S CBFS is denoted as Acc CBFS , the verification accuracy of S RFEThe verification accuracy is denoted as Acc RFE , S RT The verification accuracy is denoted as Acc RT .

[0040] Optionally, before using the HGSO algorithm to adaptively optimize the 5 stable feature subsets to obtain the optimal feature subset, the method further includes:

[0041] Based on the verification accuracy of each stable feature subset, assign an initial weight to each stable feature subset, S mRMR The initial weight of is denoted as w mRMR , S MCFS The initial weight of is denoted as w MCFS , S CBFS The initial weight of is denoted as w CBFS , S RFE The initial weight of is denoted as w RFE , S RT The initial weight of is denoted as w RT ;

[0042] The calculation formulas for the initial weights of each stable feature subset are respectively:

[0043] w mRMR = Acc mRMR / (Acc mRMR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT );

[0044] w MCFS = Acc MCFS / (Acc mRmR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT );

[0045] w CBFS = Acc CBFS / (Acc mRMR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT );

[0046] w RFE = Acc RFE / (Acc mRMR + Acc MCFS + Acc CBFS + Acc RFE + AccRT )

[0047] w RT = Acc RT / (Acc mRMR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT ) BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the related art, the following will briefly introduce the drawings required to be used in the description of the embodiments of the present application or the related art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1 is a flowchart of a method for identifying aliased signal modulation with adaptive integrated feature selection provided in an embodiment of the present application;

[0050] Figure 2 is a schematic diagram of an aliasing scenario provided in an embodiment of the present application;

[0051] Figure 3 is a design diagram of an adaptive integrated feature selection algorithm provided in an embodiment of the present application;

[0052] Figure 4 is a schematic diagram showing the change of the recognition accuracy rate with the signal-to-noise ratio after 28 kinds of aliased signals adopt the method proposed in the present application in an embodiment of the present application;

[0053] Figure 5 is a schematic diagram showing the average recognition accuracy rate of the adaptive integrated feature selection algorithm for 10 different component signal power ratios in an embodiment of the present application;

[0054] Figure 6 is a schematic diagram showing the comparison of the average recognition accuracy rate of the adaptive integrated feature selection algorithm for 10 different component signal power ratios with common algorithms in an embodiment of the present application;

[0055] Figure 7 is a schematic diagram showing the recognition accuracy rate when the number of aliased signal components changes for the method proposed in the present application in an embodiment of the present application;

[0056] Figure 8 is a schematic structural diagram of a system for identifying aliased signal modulation with adaptive integrated feature selection provided in another embodiment of the present application;

[0057] Figure 9 It is a schematic structural diagram of an electronic device provided in another embodiment of the present application. Detailed implementation manners

[0058] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the embodiments of the present application will be elaborated in detail below with reference to the accompanying drawings. In various embodiments of the present application, many technical details are proposed to help readers better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in the present application can still be implemented. The following division of the embodiments is only for convenience of description and should not constitute any limitation to the specific implementation manners of the present application. The various embodiments can be combined and cross-referenced with each other on the premise of not being contradictory.

[0059] To more clearly elaborate the technical solutions proposed in the present application, the relevant content of aliased signals is introduced first here.

[0060] Generally speaking, when there is only one signal in a single channel, the signal is a single-carrier signal, and its mathematical model is:

[0061]

[0062] where N v is the number of symbols, P i is the power of the independent modulation signal, a i (n) is the information symbol sequence, q i (t) is the shaping pulse, T si is the symbol period, f ci is the carrier frequency of the modulation signal, φ i is the initial phase of the carrier, and ξT si is the timing error.

[0063] Assume that within the allowable bandwidth f re of a single-channel receiver, within a reception time T re there are R independent narrowband signals falling within the receiver's bandwidth range, and the signals are statistically independent of each other. The duration of each component signal within the entire reception time is different, but all are completely received by the receiver. In addition, noise pollution must also exist. Then the mathematical model of the single-channel time-frequency aliased signal is expressed as:

[0064]

[0065] where s i (t, T si ) is the i-th component signal in the aliased signal, and T siIndicates the duration of each component signal. n(t) is stationary additive white Gaussian noise, and the component signals in the aliased signal are statistically independent of each other, and each signal component in the aliased signal is statistically independent of the noise.

[0066] To more comprehensively and intuitively measure the aliasing situation of the aliased signal, from the perspective of the component signal, the influences in the time domain, frequency domain, and spatial domain are comprehensively considered. Among them, the spatial domain is reflected by the power change, and the time-domain aliasing degree, frequency-domain aliasing degree, and power ratio of the aliased signal are defined.

[0067] First, define the time-domain aliasing degree of the signal. Assume that the duration of the aliased signal composed of R component signals is T re , and the duration of each component signal is T i (0 < T i ≤T re ), T ij represents the duration of aliasing between the i-th signal and the j-th signal (T ij = T ji ), then the time-domain aliasing ratio P t of the R component signals is:

[0068]

[0069] For the definition of the frequency-domain aliasing ratio, assume that the bandwidth of each component signal is f i (0 < f i < f re ), f ij represents the frequency band where the i-th signal and the j-th signal are aliased (f ij = f ji ), then the frequency-domain aliasing ratio P f is:

[0070]

[0071] The definition of the power ratio of the aliased signal is relatively simple. The power of each component signal is expressed as E i . Assume that the aliased signal is composed of component signal 1 and component signal 2, then the power ratio between component signal 1 and component signal 2 is E1:E2.

[0072] An embodiment of the present application proposes an aliased signal modulation recognition method based on adaptive integrated feature selection, which is applied to an electronic device. Among them, the electronic device can be a terminal or a server. In this embodiment and the following embodiments, the electronic device is described by taking the server as an example. The implementation details of an aliased signal modulation recognition method based on adaptive integrated feature selection proposed in this embodiment are specifically described below. The following content is only implementation details provided for convenient understanding and is not necessary for implementing this solution.

[0073] The specific process of an aliased signal modulation recognition method based on adaptive integrated feature selection proposed in this embodiment can be as follows Figure 1 shown, including:

[0074] Step 101: Use a sliding window to divide the received signal into several small signals. Based on the power matrix of each small signal, determine the position where the received signal starts to be aliased and the position where the aliasing ends, and extract the aliased signal.

[0075] In specific implementation, the aliasing scenario is as follows Figure 2 shown. The server needs to extract the aliased signal from the received signal, which is achieved by means of a sliding window. The server uses the sliding window to divide the received signal into several small signals. Based on the power matrix of each small signal, determine the position where the received signal starts to be aliased and the position where the aliasing ends, and extract the aliased signal.

[0076] In an example, the server first needs to set a suitable window size win_lengtth and a sliding step length step_length for the sliding window according to the length Signal_length of the received signal. Then, use the set sliding window to divide the received signal into N win small signals. Subsequently, calculate the power matrix of each small signal respectively, and reduce the number N of small signals win to a preset fixed value N fixed . Take the upper quantile for every N win / N fixed small signals to obtain a new power matrix. Next, the server needs to perform unsupervised clustering on the new power matrix using the spectral density clustering algorithm. Finally, determine the position where the category changes, that is, the position Loc_p where the power changes in the new power matrix, and reverse-infer the position Loc_win of the original window according to Loc_p, so as to obtain the position where the received signal starts to be aliased and the position where the aliasing ends. Then, according to the position where the aliasing starts and the position where the aliasing ends, extract the aliased signal from the received signal.

[0077] Step 102: Extract features from the aliased signal to construct a feature set of the aliased signal. The features in the feature set of the aliased signal include 10 combined high-order moment features, 16 combined high-order cumulant features, 8 high-order moment features, 8 high-order cumulant features, and 4 instantaneous features.

[0078] In specific implementation, the instantaneous feature statistic of the aliased signal is essentially to analyze the second-order statistical characteristics of the characteristics of the instantaneous amplitude, phase, and frequency of the signal over a period of time. The existence of its equivalence and multiplicity makes it unable to identify the minimum-phase system, and it is sensitive to noise in actual communication, resulting in a reduction in the discrimination of features under non-ideal conditions. To overcome the influence of these factors, higher-order statistics must be selected, collectively referred to as higher-order statistics. The most commonly used higher-order statistics are higher-order moments and higher-order cumulants. Based on this, the server extracts features from the aliased signal to construct a feature set of the aliased signal. The features in the feature set of the aliased signal include 10 combined higher-order moment features, 16 combined higher-order cumulant features, 8 higher-order moment features, 8 higher-order cumulant features, and 4 instantaneous features. It should be noted that the power factor and phase jitter will have an impact on the higher-order moments and higher-order cumulants. Therefore, in this embodiment, different-order higher-order moments and higher-order cumulants are used in the form of ratios to construct new combined higher-order moments and combined higher-order cumulants, thereby eliminating the adverse effects brought by the power factor and phase jitter.

[0079] In one example, the 8 selected higher-order moment features are m 20 、m 40 、m 60 、m 80 、m 21 、m 42 、m 63 and m 84 , and the 8 higher-order cumulant features are c 20 、c 21 、c 40 、c 41 、c 42 、c 60 、c 63 and c 80 . The calculation formulas for the higher-order moment features and higher-order cumulant features are as follows:

[0080] m kx (τ1,…,τ k-1 ) = E{x(t)x(t + τ1)…x(t + τ k-1 )};

[0081] c kx (τ1,…,τ k-1 ) = cum[x(t),x(t + τ1),…,x(t + τ k-1 )];

[0082] where x(t) represents a stationary random process, k represents the order, τ represents the time delay, and cum(·) represents the cumulative function.

[0083] In one example, the combined higher-order moment feature is denoted as Fmoma where \(a = 1, 2, \ldots, 10\), and the combined high - order cumulant feature is denoted as \(F\). cumb where \(b = 1, 2, \ldots, 16\).

[0084] The calculation formulas for 10 combined high - order moment features are as follows:

[0085]

[0086] The calculation formulas for 16 combined high - order cumulant features are as follows:

[0087] \(F\) cum1 =\(\vert c\) 40 \vert / \vert c\) 41 \vert\); \(F\) cum2 =\(\vert c\) 40 \vert / \vert c\) 42 \vert\); \(F\) cum3 =\(\vert c\) 41 \vert / \vert c\) 42 \vert\);

[0088]

[0089] In an example, the 4 instantaneous features are the second - order statistical feature \(m\) of the instantaneous amplitude a , the maximum value feature \(\gamma\) of the instantaneous amplitude spectrum density max , the standard deviation feature \(\sigma\) of the zero - centered normalized instantaneous amplitude of non - weak signals da and the standard deviation feature \(\sigma\) of the zero - centered normalized absolute value of the instantaneous amplitude aa .

[0090] The calculation formulas for the 4 instantaneous features are as follows:

[0091]

[0092] \(\gamma\) max =\(\max\vert DFT[A\) cn (i)]\vert\) 2 / N;

[0093]

[0094] where \(N\) is the number of sampling points, \(A(i)\) is the instantaneous amplitude of the signal, \(A\) cn (i)=A(i) - 1, \(A\) cn (i) is the zero - centered normalized instantaneous amplitude of the signal, \(DFT(\cdot)\) represents the discrete Fourier transform, \(a\) i is the threshold of the preset non - weak signal, and \(C\) is the number of non - weak signals.

[0095] Step 103: Use five feature selection methods, namely minimum redundancy maximum correlation criterion selection, multi-cluster feature selection, correlation-based feature selection, recursive feature elimination selection, and random forest feature importance selection, to select each feature in the feature set, and initially screen to obtain five stable feature subsets.

[0096] There is a conversion relationship between the higher-order moments and higher-order cumulants of the signal. Therefore, the M-dimensional feature set I composed of higher-order moments, higher-order cumulants, combined higher-order moments, and combined higher-order cumulants feature has a high degree of correlation and redundancy. In addition, when the power ratio of the component signals of the aliased signal changes, it is equivalent to performing a function perturbation on the original feature set. Therefore, the target classification task c of the modulation method of the aliased signal feature requires a feature subset S with high discrimination and high stability feature to achieve. feature To this end, this embodiment integrates three types of feature selection methods: filter type, wrapper type, and embedded type, and uses five feature selection methods, namely minimum redundancy maximum correlation criterion selection (minimal-Redundancy-Maximal-Relevance criterion, mRMR), multi-cluster feature selection (Multi-Cluster Feature Selection, MCFS), correlation-based feature selection (Correlation BasedFeature Selection, CBFS), recursive feature elimination selection (Recursive Feature Elimination, RFE), and random forest feature importance selection (Recursive Feature Elimination, RFE), to select each feature in the feature set, and initially screen to obtain five stable feature subsets. The design of the adaptive integrated feature selection algorithm is as

[0097] shown. Figure 3 shown.

[0098] In an example, the stable feature subset screened by using the minimum redundancy maximum correlation criterion selection method is denoted as S mRMR , the stable feature subset screened by using the multi-cluster feature selection method is denoted as S MCFS , the stable feature subset screened by using the correlation-based feature selection method is denoted as S CBFS , the stable feature subset screened by using the recursive feature elimination selection method is denoted as S RFE , and the stable feature subset screened by using the random forest feature importance selection method is denoted as S RT .

[0099] In one example, after the server obtains 5 stable feature subsets through preliminary screening, it is also necessary to use the random forest algorithm to determine the verification accuracy of each of the 5 stable feature subsets respectively. Among them, the verification accuracy of S mRMR is denoted as Acc mRMR , S MCFS 's verification accuracy is denoted as Acc MCFS , S CBFS 's verification accuracy is denoted as Acc CBFS , S RFE 's verification accuracy is denoted as Acc RFE , S RT 's verification accuracy is denoted as Acc RT .

[0100] Step 104, use the HGSO algorithm to perform adaptive optimization on the 5 stable feature subsets to obtain the optimal feature subset.

[0101] In a specific implementation, there are still correlations and redundancies among the 5 stable feature subsets. Therefore, in order to further adaptively optimize the sorted features to the optimal classification effect and the smallest feature subset, it is necessary to search the rearranged feature set. Compared with using global optimal search, the efficiency of heuristic search is high. In this embodiment, the meta-heuristic Henry Gas Solubility Optimization (HGSO) algorithm is used to perform adaptive optimization on the 5 stable feature subsets to obtain the optimal feature subset.

[0102] When using the Henry Gas Solubility Optimization algorithm for feature selection, different positions of the gas in the feature space represent different feature subsets. The main idea is to regard the feature set as the given liquid, and different feature subsets represent the partial pressure of the gas. The optimal feature subset is selected by exploring the solubility of the gas in different feature spaces.

[0103] In one example, to improve the stability and generalization ability of the extracted features, before the server uses the HGSO algorithm to perform adaptive optimization on the 5 stable feature subsets to obtain the optimal feature subset, it is also necessary to assign different initial weights to each stable feature subset based on the verification accuracy of each stable feature subset. Among them, the initial weight of S mRMR is denoted as w mRMR , S MCFS 's initial weight is denoted as w MCFS , S CBFS 's initial weight is denoted as w CBFS , S RFE 's initial weight is denoted as w RFE , S RT 's initial weight is denoted as w RT .

[0104] In one example, the calculation formulas for the initial weights of each stable feature subset are respectively:

[0105] w mRMR = Acc mRMR / (Acc mRMR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT );

[0106] w MCFS = Acc MCFS / (Acc mRMR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT );

[0107] w CBFS = Acc CBFS / (Acc mRMR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT );

[0108] w RFE = Acc RFE / (v mRMR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT );

[0109] w RT = Acc RT / (Acc mRMR + Acc MCFS + Acc CBFS + Acc RFE + Acc RT ).

[0110] Step 105: Perform modulation recognition based on the optimal feature subset to obtain the modulation recognition result.

[0111] An aliased signal modulation recognition method based on adaptive integrated feature selection proposed in this embodiment first accurately locates the aliasing position and the end of aliasing of the received signal by using a sliding window, so as to accurately extract the aliased signal. Next, a feature set containing 46 different features is constructed for the aliased signal. High-order moment features and high-order cumulant features are used to improve the discrimination, and combined high-order moment features and combined high-order cumulant features are used to eliminate the influence of power factors and phase jitters on high-order moments and high-order cumulants. Subsequently, the advantages of three feature selection methods, namely filter, wrapper, and embedded, are comprehensively considered. Five feature selection methods, namely minimum redundancy maximum correlation criterion selection, multi-cluster feature selection, correlation feature selection, recursive feature elimination selection, and random forest feature importance selection, are used for feature selection, and the HGSO algorithm is used for adaptive optimization to obtain the optimal feature subset. Finally, modulation recognition is performed based on the optimal feature subset to obtain the modulation recognition result. The designed aliased signal modulation recognition method can quickly and accurately realize the modulation recognition of aliased signals when factors such as the modulation mode type, aliasing pattern, power ratio between component signals, carrier frequency offset, phase offset, and signal-to-noise ratio of aliased signals change simultaneously, and has high stability and universality.

[0112] The step division of the above various methods is only for clear description. When implemented, they can be combined into one step or some steps can be split into multiple steps. As long as the same logical relationship is included, it is within the protection scope of this application; adding insignificant modifications to the algorithm or process or introducing insignificant designs, but not changing the core design of its algorithm and process, are within the protection scope of this application.

[0113] In one embodiment, in order to verify the effectiveness of an aliased signal modulation recognition method based on adaptive integrated feature selection proposed in this application, relevant simulation experiments were carried out. Figure 4 It shows a schematic diagram of the recognition accuracy rate of 28 aliased signals changing with the signal-to-noise ratio after using the method proposed in this application. As the signal-to-noise ratio increases, the recognition accuracy is significantly improved. Figure 5 It shows the average recognition accuracy rate of the adaptive integrated feature selection algorithm for 10 different power ratios of component signals. As the signal-to-noise ratio increases, the recognition accuracy rates are significantly improved. Figure 6 It shows the comparison of the average recognition accuracy rate of the adaptive integrated feature selection algorithm for 10 different power ratios of component signals with common algorithms. The random forest algorithm has the best comprehensive recognition rate. Figure 7 It shows the recognition accuracy rate of the method proposed in this application when the number of component signals of the aliased signal changes. When the signal-to-noise ratio is 10 dB, the modulation recognition accuracy rates of the aliased signals with 3 and 4 component signals can reach 85% and 90% respectively.

[0114] Another embodiment of the present application proposes an aliased signal modulation recognition system with adaptive integrated feature selection. The implementation details of the aliased signal modulation recognition system with adaptive integrated feature selection proposed in this embodiment will be specifically described below. The following content is only the implementation details provided for convenient understanding and is not necessary for implementing this example.

[0115] Figure 8 FIG. 4 is a schematic structural diagram of an aliased signal modulation recognition system with adaptive integrated feature selection proposed in this embodiment. The system includes: a sliding window positioning module 201, a feature set construction module 202, an integrated feature selection module 203, an adaptive optimization module 204, and a modulation recognition module 205.

[0116] The sliding window positioning module 201 is configured to divide the received signal into several small signals by using a sliding window, determine the position where the received signal starts to be aliased and the position where the aliasing ends based on the power matrix of each small signal, and extract the aliased signal.

[0117] The feature set construction module 202 is configured to extract features from the aliased signal and construct a feature set of the aliased signal. The features in the feature set of the aliased signal include 10 combined high-order moment features, 16 combined high-order cumulant features, 8 high-order moment features, 8 high-order cumulant features, and 4 instantaneous features.

[0118] The integrated feature selection module 203 is configured to use five feature selection methods, namely minimum redundancy maximum correlation criterion selection, multi-cluster feature selection, correlation feature selection, recursive feature elimination selection, and random forest feature importance selection, to select each feature in the feature set and preliminarily screen to obtain five stable feature subsets.

[0119] The adaptive optimization module 204 is configured to adaptively optimize the five stable feature subsets by using the HGSO algorithm to obtain an optimal feature subset.

[0120] The modulation recognition module 205 is configured to perform modulation recognition based on the optimal feature subset to obtain a modulation recognition result.

[0121] It is worth mentioning that each module involved in this embodiment is a logical module. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or can be implemented by a combination of multiple physical units. In addition, in order to highlight the innovative part of the present application, units that are not closely related to solving the technical problems proposed in the present application are not introduced in this embodiment, but this does not mean that there are no other units in this embodiment.

[0122] It is not difficult to find that this embodiment is a system embodiment corresponding to the above method embodiment, and this embodiment can be implemented in cooperation with the above method embodiment. The relevant technical details and technical effects mentioned in the above embodiments are still valid in this embodiment. To avoid repetition, they will not be elaborated here. Correspondingly, the relevant technical details mentioned in this embodiment can also be applied to the above embodiments.

[0123] Another embodiment of the present application proposes an electronic device, and its specific structure is as Figure 9 shown, including: at least one processor 301; and a memory 302 communicatively connected to the at least one processor 301; wherein, the memory 302 stores instructions executable by the at least one processor 301, and the instructions are executed by the at least one processor 301 to enable the at least one processor 301 to execute an aliased signal modulation recognition method of adaptive integrated feature selection as described in the above method embodiment.

[0124] Among them, the memory and the processor can be connected in a bus manner. The bus can include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be further described herein. The bus interface is responsible for providing an interface between the bus and the transceiver. The transceiver can be an element or multiple elements, such as multiple receivers and transmitters, and provides a unit for communicating with various other devices on the transmission medium. The data processed by the processor is transmitted on the wireless medium through the antenna. Further, the antenna also receives data and transmits the data to the processor.

[0125] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. The memory can be used to store the data used by the processor when executing operations.

[0126] Another embodiment of the present application proposes a computer-readable storage medium storing a computer program, which when executed by a processor, can implement an aliased signal modulation recognition method of adaptive integrated feature selection as described in the above method embodiment.

[0127] That is, those skilled in the art can understand that all or part of the steps in the methods of the above embodiments can be completed by instructing relevant hardware through a program. The program is stored in a storage medium, including several instructions to enable a device (such as a single-chip microcomputer, a chip, etc.) or a processor to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, ROM (Read-Only Memory), RAM (Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0128] Those of ordinary skill in the art can understand that the above embodiments are specific embodiments for implementing the present application. In actual applications, various changes can be made to them in form and details without departing from the spirit and scope of the present application.

Claims

1. A method for identifying aliased signal modulation by adaptive integrated feature selection, characterized in that: include: The received signal is divided into several small signals by using a sliding window, and the position where aliasing occurs and the position where aliasing ends of the received signal are determined based on the power matrix of each small signal, so as to extract the aliased signal; The aliasing signal is feature extracted to construct a feature set of the aliasing signal. The features in the feature set of the aliasing signal include 10 combined high-order moment features, 16 combined high-order cumulant features, 8 high-order moment features, 8 high-order cumulant features and 4 instantaneous features. Five feature selection methods, including minimum redundancy and maximum correlation criterion selection, multi-cluster feature selection, correlation feature selection, recursive feature elimination selection, and random forest feature importance selection, were used to select each feature in the feature set, and five stable feature subsets were obtained through preliminary screening. The HGSO algorithm is used to adaptively optimize the five stable feature subsets to obtain the optimal feature subset; Modulation recognition is performed based on the optimal feature subset to obtain a modulation recognition result.

2. The aliasing signal modulation recognition method of adaptive integrated feature selection according to claim 1 is characterized in that: The received signal is divided into several small signals by using a sliding window. Based on the power matrix of each small signal, the position where the aliasing of the received signal occurs and the position where the aliasing ends are determined, and the aliasing signal is extracted, including: The window size and sliding step size of the sliding window are set according to the length of the received signal, so as to divide the received signal into N win A small signal; Calculate the power matrix of each small signal separately, and convert the number of small signals N win Reduced to a preset fixed value N fixed , each N win / N fixed Take the upper quantile of each small signal and get a new power matrix; The new power matrix is ​​clustered unsupervisedly using the spectral density clustering algorithm; Determine the position where the category changes, that is, the position Loc_p where the power changes in the new power matrix, and reversely infer the position Loc_win of the original window based on Loc_p, so as to obtain the position where the aliasing of the received signal occurs and the position where the aliasing ends; The aliased signal is extracted from the received signal according to the position where the aliasing occurs and the position where the aliasing ends.

3. The aliasing signal modulation recognition method of adaptive integrated feature selection according to claim 1, characterized in that: The 8 high-order moment features are m 20 、m 40 、m 60 、m 80 、m 21 、m 42 、m 63 and m 84 , the 8 high-order cumulative features are c 20 、c 21 、c 40 、c 41 、c 42 、c 60 、c 63 and c 80 ; The calculation formulas for high-order moment characteristics and high-order cumulant characteristics are: m kx (τ1,…,τ k-1 )=E{x(t)x(t+τ1)…x(t+τ k-1 )}; c kx (τ1,…,τ k-1 )=cum[x(t),x(t+τ1),…,x(t+τ k-1 )]; Here, x(t) represents a stationary random process, k represents the order, τ represents the time delay, and cum(·) represents the cumulative function.

4. The aliasing signal modulation identification method of adaptive integrated feature selection according to claim 3 is characterized in that: The combined high-order moment feature is denoted as F moma , a=1,2,…,10, the combined high-order cumulative feature is recorded as F cumb , b=1,2,…,16; The calculation formulas for the 10 combined high-order moment features are: The calculation formulas for the 16 combined high-order cumulant features are: F cum1 =|c 40 | / |c 41 |;F cum2 =|c 40 | / |c 42 |;F cum3 =|c 41 | / |c 42 |; 5. The aliasing signal modulation identification method of adaptive integrated feature selection according to claim 4 is characterized in that: The four instantaneous features are the second-order statistical features m of the instantaneous amplitude a , the maximum value characteristic of instantaneous amplitude spectrum density γ max , the standard deviation characteristic σ of the instantaneous amplitude of the zero-centered normalized non-weak signal da and the standard deviation characteristic σ of the absolute value of the instantaneous amplitude normalized to the zero center aa ; The calculation formulas of the four instantaneous features are: γ max =max|DFT[A cn (i)]| 2 / N; Where N is the number of sampling points, A(i) is the instantaneous amplitude of the signal, and A cn (i) = A(i)-1, A cn (i) is the zero-centered normalized instantaneous amplitude of the signal, DFT(·) represents discrete Fourier transform, a i is the preset non-weak signal threshold, and C is the number of non-weak signals.

6. The aliased signal modulation identification method according to any one of claims 1 to 5, characterized in that: The stable feature subset is obtained by screening using the minimum redundancy and maximum correlation criterion selection method and is denoted as S mRMR , the stable feature subset is obtained by multi-cluster feature selection method and is recorded as S MCFS , the stable feature subset is obtained by using the relevant feature selection method and is recorded as S CBFS , the stable feature subset is obtained by recursive feature elimination selection method and is recorded as S RFE , the random forest feature importance selection method is used to select a stable feature subset, which is denoted as S RT ; After obtaining five stable feature subsets through preliminary screening, the method further comprises: The random forest algorithm is used to determine the verification accuracy of each of the five stable feature subsets, S mRMR The verification accuracy is recorded as Acc mRMR , S MCFS The verification accuracy is recorded as Acc MCFS , S CBFS The verification accuracy is recorded as Acc CBFS , S RFE The verification accuracy is recorded as Acc RFE , S RT The verification accuracy is recorded as Acc RT .

7. The aliasing signal modulation identification method of adaptive integrated feature selection according to claim 6, characterized in that: Before using the HGSO algorithm to adaptively optimize the five stable feature subsets to obtain the optimal feature subset, the method further includes: Based on the verification accuracy of each stable feature subset, an initial weight is assigned to each stable feature subset, S mRMR The initial weight is denoted as w mRMR , S MCFS The initial weight is denoted as w MCFS , S CBFS The initial weight is denoted as w CBFS , S RFE The initial weight is denoted as w RFE , S RT The initial weight is denoted as w RT ; The calculation formulas for the initial weights of each stable feature subset are: w mRMR =Acc mRMR / (Acc mRMR +Acc MCFS +Acc CBFS +Acc RFE +Acc RT ); w MCFS =Acc MCFS / (Acc mRMR +Acc MCFS +Acc CBFS +Acc RFE +Acc RT ); w CBFS =Acc CBFS / (Acc mRMR +Acc MCFS +Acc CBFS +Acc RFE +Acc RT ); w RFE =Acc RFE / (Acc mRMR +Acc MCFS +Acc CBFS +Acc RFE +Acc RT ); w RT =Acc RT / (Acc mRMR +Acc MCFS +Acc CBFS +Acc RFE +Acc RT )。 8. An aliased signal modulation recognition system with adaptive integrated feature selection, characterized in that: include: A sliding window positioning module is used to divide the received signal into several small signals by using a sliding window, determine the position where aliasing occurs and the position where aliasing ends in the received signal based on the power matrix of each small signal, and extract the aliased signal; A feature set construction module is used to extract features of aliased signals and construct a feature set of aliased signals. The features in the feature set of aliased signals include 10 combined high-order moment features, 16 combined high-order cumulant features, 8 high-order moment features, 8 high-order cumulant features and 4 instantaneous features. The integrated feature selection module is used to select the features in the feature set using five feature selection methods: minimum redundancy maximum correlation criterion selection, multi-cluster feature selection, correlation feature selection, recursive feature elimination selection, and random forest feature importance selection. Five stable feature subsets are initially screened; The adaptive optimization module is used to adaptively optimize the five stable feature subsets using the HGSO algorithm to obtain the optimal feature subset; The modulation recognition module is used to perform modulation recognition based on the optimal feature subset to obtain a modulation recognition result.

9. An electronic device, characterized in that: include: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute an aliasing signal modulation identification method with adaptive integrated feature selection as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it can implement an aliased signal modulation identification method of adaptive integrated feature selection according to any one of claims 1 to 7.

Images