CN116756637A - Wireless signal intelligent detection and identification method and computer readable storage medium - Google Patents
Wireless signal intelligent detection and identification method and computer readable storage medium Download PDFInfo
- Publication number
- CN116756637A CN116756637A CN202311006207.5A CN202311006207A CN116756637A CN 116756637 A CN116756637 A CN 116756637A CN 202311006207 A CN202311006207 A CN 202311006207A CN 116756637 A CN116756637 A CN 116756637A
- Authority
- CN
- China
- Prior art keywords
- signal
- generator
- noise
- model
- approximation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000001514 detection method Methods 0.000 title claims abstract description 20
- 239000011159 matrix material Substances 0.000 claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 14
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 12
- 238000013434 data augmentation Methods 0.000 claims description 9
- 238000013527 convolutional neural network Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 7
- 230000008485 antagonism Effects 0.000 abstract description 6
- 238000009826 distribution Methods 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 238000013256 Gubra-Amylin NASH model Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 241001647769 Mirza Species 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0475—Generative networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/094—Adversarial learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
The invention discloses a wireless signal intelligent detection and identification method and a computer readable storage medium, wherein the method comprises the following steps: combining the standard signal model, and performing feature decomposition on the mixed radio signals in the data set of the standard signal model to obtain an approximation coefficient and approximation noise; generating countermeasure learning on the approximation coefficient and the approximation noise, wherein the generated countermeasure learning is generated separately and in combination; generating a hybrid radio signal in combination with the standard signal model using the coefficients and noise generated by the trained generator; and combining the generated mixed radio signal with the data set of the quasi-signal model, putting the combined mixed radio signal into a wireless signal recognition model for training, and recognizing the mixed radio signal by the trained wireless signal recognition model. The method combines non-Negative Matrix Factorization (NMF) technology and Generation Antagonism Network (GAN) technology, and amplifies the data set, thereby effectively improving the identification precision of the mixed radio signal.
Description
Technical Field
The invention relates to the field of radio signal identification, in particular to a wireless signal intelligent detection and identification method and a computer readable storage medium.
Background
The difficulty of identifying mixed radio signals is greatly increased compared to the identification of single radio signals for the following reasons: firstly, to identify the signals mixed by various radio signals, a huge data set is needed to learn accurately, but in reality, the two are difficult to achieve, and due to other interference factors such as noise, the manually collected data set has the problems of improper data distribution and data quality. The above causes that the existing identification method is insufficient in identification accuracy for the hybrid radio signal.
For the problem of insufficient data sets, there are two solutions:
one is to use classification methods with good predictive power, such as nearest neighbor classifier [1], support Vector Machine (SVM) [2], which are mostly conventional machine learning algorithms, in the case of insufficient data sets, often fail to achieve good classification accuracy.
Another is to generate new samples to expand the data set to improve classification performance.
In practical applications, many data are difficult to collect, and the performance of the traditional machine learning method of the first solution is not satisfactory. Therefore, the data augmentation of the second solution is now used by more people.
Data augmentation using existing data sets is currently one very effective approach.
Generating an antagonism network has many advantages in deep learning. The GAN may be used as unsupervised learning, semi-supervised learning, or supervised learning. The GAN may generate additional knowledge as a priori knowledge based on the data set without adding a priori knowledge, and is feasible in experimental effect. Huang L4 et al uses GAN technology to amplify data in wireless modulation in the field of image and voice recognition, improving the performance of classifier. Tang [5] et al propose to use a Generated Antagonism Network (GAN) as a data augmentation tool to address the problem of under training in the absence of sufficient data samples. The data enhancement process can be implemented using Nash equalization of the generator and discriminator. S. Kacprzak [6] et al studied several Domain Adaptive (DA) antagonism models and their impact on the task of classifying acoustic scenes, the studied models including several Generated Antagonism Networks (GANs) with different loss functions, and so-called cyclic GAN models consisting of two interconnected GAN models. H. Zhou [7] et al introduced GAN into the field of wireless signal classification. They propose a GAN-combined SSL framework that can directly process raw in-phase and quadrature (IQ) signal data for data augmentation purposes.
After the GAN technology is introduced into the technical scheme, the identification precision of the radio signals is improved to a certain extent, but the network model has a complex structure and high training difficulty, and the identification precision still cannot be better on the identification of the mixed radio signals consisting of a plurality of radio signals.
Document [1]: astam M W, zhuZ, nandi A K. Automatic digital modulation classification using genetic programming with K-nearest neighbor [ C ]//2010-MILCOM 2010 Military Communications Conference, IEEE, 2010:1731-1736.
Document [2]: sunX, su S, zuo Z, et al Modulation classification using compressed sensing and decision tree-support vector machine in cognitiveradio systems, sensors, 2020, 20 (5): 1438.
Document [3]: goodFe low I J, pouget-Abadie J, mirza M, et al Generative adversarial networks arXiv preprint arXiv:1406.2661, 2014.
Document [4]: huang L, pan W, zhang Y, et al Data augmentation for deep learningbased radio modulation classification IEEE Access, 2019, 8:1498-1506.
Document [5]: Z.Tang, M.Tao, J.Su, Y.Gong, Y.Fan and T.Li, "Data Augmentation for Signal Modulation Classification using Generative Adverse Network,"2021 IEEE 4th International Conference on ElectronicInformation and Communication Technology (ICEICT), 2021, pp. 450-453, doi: 10.1109/ICEICT53123.2021.9531296.
Document [6 ]: s. Kacprpzakand K.Kowalczyk, "Adversarial Domain Adaptation with Paired Examples for Acoustic Scene Classification onDifferent Recording Devices," 2021 29th European Signal Processing Conference (EUSIPCO), 2021, pp.1030-1034, doi:10.23919/EUSIPCO54536.2021.9616321.
Document [7]: H.Zhou, L.Jiao, S.Zheng, L.Yang, W.shen and X.Yang, "Generative adversarial network-based electromagnetic signal classification: A semisupervised learning framework," in ChinaCommunications, vol.17, no.10, pp.157-169, oct.2020, doi:10.23919/JCC.2020.10.011.
Disclosure of Invention
The invention aims to overcome the defects and shortcomings of the prior art and provides a wireless signal intelligent detection and identification method, which combines a non-Negative Matrix Factorization (NMF) technology and a generation countermeasure network (GAN) technology to amplify a data set, thereby effectively improving the identification precision of a hybrid radio signal.
The aim of the invention is achieved by the following technical scheme:
a wireless signal intelligent detection and identification method comprises the following steps:
s1, combining a standard signal model, and performing feature decomposition on a mixed radio signal in a data set of the standard signal model to obtain an approximation coefficient and approximation noise;
s2, generating countermeasure learning for the approximation coefficient and the approximation noise, wherein the generated countermeasure learning is generated separately and used for countermeasure in combination;
s3, generating a mixed radio signal by combining the coefficient and noise generated by the trained generator with a standard signal model;
and S4, combining the generated mixed radio signal with the data set of the quasi-signal model, putting the combined radio signal into a wireless signal recognition model for training, and recognizing the mixed radio signal by the trained wireless signal recognition model.
The step S1 specifically comprises the following steps:
s11, performing non-negative matrix factorization on the mixed radio signals in the data set of the standard signal model to obtain approximate coefficients:
（1）；
wherein ,representing n real mixed radio signals, each signal having a length m; />Representing an approximation coefficient matrix obtained by non-negative matrix decomposition; />A standard signal model is represented, and k represents the number of signal types in the mixed radio signal;
s12, combining the original signal, the approximation coefficient and the standard signal model to obtain the approximation noise：
（2）。
In step S1, the feature decomposition is non-negative matrix decomposition; or other means of feature extraction.
The step S2 specifically includes:
s21, performing data augmentation by using a generation countermeasure network, performing supervised training on an approximation coefficient and approximation noise by using a first generator and a second generator respectively, wherein the input of the first generator and the input of the second generator are random parameters, and the output of the first generator and the second generator are the generation coefficient and the generation noise respectively;
s22, the results generated by the first generator and the second generator are jointly placed into a discriminator for discrimination, and meanwhile, the approximation coefficient and the approximation noise obtained by nonnegative matrix factorization are jointly placed into the discriminator for discrimination; the discriminator judges whether the generated result is output 0 or not and judges whether the approximated result of non-negative matrix factorization is output 1 or not;
s23, the output of the discriminator is fed back to the generator and the first and second discriminators for learning, and the final discriminator does not output whether the input is the generated result or the approximate result of the non-negative matrix factorization, so that the generator completes training.
The first generator and the second generator are built by convolutional neural networks or other supervised network models.
The discriminator is built by using convolutional nerves or other supervised network models.
The generating a loss function of the countermeasure network is:
（3）；
wherein the formula representsFor generators, for cross entropy of two classes and />The right side of the formula takes the minimum value for the +.>The right side of the formula takes the maximum value; />Indicating desire(s)>Representing data complianceProbability distribution->Is (are) desirable to be (are)>Representing the discrimination result of the discriminator on the data X, and between 0 and 1;
indicating when the data is created by the generator-> and />Generating compliance-> and />When distributed jointlyIs (are) desirable to be (are)>Representation generator->Is a generator->And so on.
The step S3 specifically includes:
s31, generating coefficients and noise by using a trained generator, wherein the input and output of a generator model are normalization results, and the size of the generated results is adjusted according to the data range of a data set;
s32, combining the standard signal model to generate a mixed signal：
（4）；
wherein ,、/>respectively representing the generation coefficient, the generation noise, < >>A standard signal model is represented, and k represents the number of signal types in the mixed radio signal; m represents the signal length, and N represents the number of generations.
Meanwhile, the invention provides:
the server comprises a processor and a memory, wherein at least one section of program is stored in the memory, and the program is loaded and executed by the processor to realize the wireless signal intelligent detection and identification method.
A computer-readable storage medium having stored therein at least one program loaded and executed by a processor to implement the wireless signal intelligent detection and identification method described above.
Compared with the prior art, the invention has the following advantages and beneficial effects:
the present invention is based on non-Negative Matrix Factorization (NMF) and generation of a countermeasure network (GAN) model, eliminating the traditional method of directly generating signals, generating a hybrid radio signal from learning coefficients and noise in combination with a standard signal model. Due to the uniqueness and correlation of coefficients and noise in the hybrid radio signal, a learning method of dual generator training and single discriminant joint discrimination is used. And adding the generated mixed radio signals into the data set, and then putting the data set into a recognition model for recognition, so that the recognition accuracy is greatly improved.
Drawings
Fig. 1 is a flowchart of a wireless signal intelligent detection and recognition method according to the present invention.
Fig. 2 is a network architecture diagram corresponding to the intelligent detection and identification method for wireless signals according to the present invention.
Fig. 3 is a graph of data generation ratio versus recognition accuracy for different signal-to-noise ratios under a CNN model.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but embodiments of the present invention are not limited thereto.
Referring to fig. 1, a wireless signal intelligent detection and identification method comprises the following steps:
s1, combining a standard signal model, and performing feature decomposition on a mixed radio signal in a data set of the standard signal model to obtain an approximation coefficient and approximation noise;
s11, performing non-negative matrix factorization on the mixed radio signals in the data set of the standard signal model to obtain approximate coefficients:
（1）；
wherein ,representing n real mixed radio signals, each signal having a length m; />Representing an approximation coefficient matrix obtained by non-negative matrix decomposition; />A standard signal model is represented, and k represents the number of signal types in the mixed radio signal;
s12, combining the original signal, the approximation coefficient and the standard signal model to obtain the approximation noise：
（2）。
The role of non-Negative Matrix Factorization (NMF) is feature extraction (decomposition) and may be replaced by other feature extraction methods.
S2, generating countermeasure learning for the approximation coefficient and the approximation noise, wherein the generated countermeasure learning is generated separately and used for countermeasure in combination;
s21, performing data augmentation by using a generation countermeasure network, performing supervised training on an approximation coefficient and approximation noise by using a first generator and a second generator respectively, wherein the input of the first generator and the input of the second generator are random parameters, and the output of the first generator and the second generator are the generation coefficient and the generation noise respectively;
s22, the results generated by the first generator and the second generator are jointly placed into a discriminator for discrimination, and meanwhile, the approximation coefficient and the approximation noise obtained by nonnegative matrix factorization are jointly placed into the discriminator for discrimination; the discriminator judges whether the generated result is output 0 or not and judges whether the approximated result of non-negative matrix factorization is output 1 or not;
s23, because the data distribution difference of the radio signal coefficient and the noise is large, the two generators are used for generating respectively, the mutual influence is reduced, and because the coefficient and the noise have certain correlation, the joint judgment is carried out in the discriminator. The output of the discriminator is fed back to the generator and the first and second discriminators for learning, and the final discriminator does not output whether the input is the generated result or the approximate result of the non-negative matrix factorization, so that the generator completes training.
The network models of the first and second generators and discriminators of the present invention are convolutional neural network models, and may be replaced with other supervised network models.
The generating a loss function of the countermeasure network is:
（3）；
（3）；
wherein the formula is expressed as a two-class cross entropy for the generator and />The right side of the formula takes the minimum value for the +.>The right side of the formula takes the maximum value; />Indicating desire(s)>Representing data complianceProbability distribution->Is (are) desirable to be (are)>Representing the discrimination result of the discriminator on the data X, and between 0 and 1; />Indicating when the data is created by the generator-> and />Generating compliance-> and />When distributed jointlyIs (are) desirable to be (are)>Representation generator->Is a generator->And so on.
S3, generating a mixed radio signal by combining the coefficient and noise generated by the trained generator with a standard signal model;
s31, generating coefficients and noise by using a trained generator, wherein the input and output of a generator model are normalization results, and the size of the generated results is adjusted according to the data range of a data set;
s32, combining the standard signal model to generate a mixed signal：
（4）；
wherein ,、/>respectively representing the generation coefficient, the generation noise, < >>A standard signal model is represented, and k represents the number of signal types in the mixed radio signal; m represents the signal length, and N represents the number of generations.
And S4, combining the generated mixed radio signal with the data set of the quasi-signal model, putting the combined radio signal into a wireless signal recognition model for training, and recognizing the mixed radio signal by the trained wireless signal recognition model.
Meanwhile, the invention provides:
the server comprises a processor and a memory, wherein at least one section of program is stored in the memory, and the program is loaded and executed by the processor to realize the wireless signal intelligent detection and identification method.
A computer-readable storage medium having stored therein at least one program loaded and executed by a processor to implement the wireless signal intelligent detection and identification method described above.
Compared with the prior art, the invention mainly specifically comprises the following distinguishing points:
1. the invention combines non-Negative Matrix Factorization (NMF) into generating an antagonism network, and provides a method for decomposing sequence data into characteristics and weights and then learning and generating.
2. The invention provides a joint learning method of a double generator single discriminator by using a double generator in generating an reactance network.
3. The present invention eliminates the traditional method of directly generating signals, and generates hybrid radio signals by combining learning coefficients and noise with a standard signal model.
In fig. 2, real data represents a Real data set, standard signals represents a Standard signal model, and random noise is used as an input of a generator, subject to uniform distribution or gaussian distribution. The result of the generator and the NMF decomposed result enter a discriminator to discriminate, and Loss feedback is indicated by Loss feedback to the generator and the discriminator. The final training completed generator generates a hybrid radio signal in combination with standard signatures.
In fig. 3, the data generation ratio is represented as a ratio of the number of generated data to the number of original data. As can be seen from fig. 3, when the signal-to-noise ratio is unchanged, the greater the data generation ratio is, the higher the recognition accuracy is, and after the data generation ratio reaches 30, the recognition accuracy reaches a peak.
The above examples are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above examples, and any other changes, modifications, substitutions, combinations, and simplifications that do not depart from the spirit and principle of the present invention should be made in the equivalent manner, and the embodiments are included in the protection scope of the present invention.
Claims (10)
1. The intelligent detection and identification method for the wireless signals is characterized by comprising the following steps of:
s1, combining a standard signal model, and performing feature decomposition on a mixed radio signal in a data set of the standard signal model to obtain an approximation coefficient and approximation noise;
s2, generating countermeasure learning for the approximation coefficient and the approximation noise, wherein the generated countermeasure learning is generated separately and used for countermeasure in combination;
s3, generating a mixed radio signal by combining the coefficient and noise generated by the trained generator with a standard signal model;
and S4, combining the generated mixed radio signal with the data set of the quasi-signal model, putting the combined radio signal into a wireless signal recognition model for training, and recognizing the mixed radio signal by the trained wireless signal recognition model.
2. The method for intelligent detection and identification of wireless signals according to claim 1, wherein the step S1 specifically comprises:
s11, performing non-negative matrix factorization on the mixed radio signals in the data set of the standard signal model to obtain approximate coefficients:
（1）；
wherein ,representing n real mixed radio signals, each signal having a length m; />Representing an approximation coefficient matrix obtained by non-negative matrix decomposition; />A standard signal model is represented, and k represents the number of signal types in the mixed radio signal;
s12, combining the original signal, the approximation coefficient and the standard signal model to obtain the approximation noise：
（2）。
3. The intelligent detection and recognition method for wireless signals according to claim 1, wherein in step S1, the feature decomposition is a non-negative matrix decomposition; or other means of feature extraction.
4. The method for intelligent detection and identification of wireless signals according to claim 1, wherein the step S2 specifically comprises:
s21, performing data augmentation by using a generation countermeasure network, performing supervised training on an approximation coefficient and approximation noise by using a first generator and a second generator respectively, wherein the input of the first generator and the input of the second generator are random parameters, and the output of the first generator and the second generator are the generation coefficient and the generation noise respectively;
s22, the results generated by the first generator and the second generator are jointly placed into a discriminator for discrimination, and meanwhile, the approximation coefficient and the approximation noise obtained by nonnegative matrix factorization are jointly placed into the discriminator for discrimination; the discriminator judges whether the generated result is output 0 or not and judges whether the approximated result of non-negative matrix factorization is output 1 or not;
s23, the output of the discriminator is fed back to the generator and the first and second discriminators for learning, and the final discriminator does not output whether the input is the generated result or the approximate result of the non-negative matrix factorization, so that the generator completes training.
5. The method of claim 4, wherein the first and second generators are built using convolutional neural networks or other supervised network models.
6. The method of claim 4, wherein the discriminator is constructed using convolutional neural or other supervised network models.
7. The method of claim 4, wherein the generating a loss function against the network is:
（3）；
wherein the formula is expressed as a two-class cross entropy for the generator and />The right side of the formula takes the minimum value for the discriminatorThe right side of the formula takes the maximum value; />Indicating desire(s)>Representing data complianceProbability distribution->Is (are) desirable to be (are)>Representing the discrimination result of the discriminator on the data X, and between 0 and 1; />Indicating when the data is created by the generator-> and />Generating compliance-> and />When distributed jointlyIs (are) desirable to be (are)>Representation generator->Is a generator->And so on.
8. The method for intelligent detection and identification of wireless signals according to claim 1, wherein the step S3 specifically comprises:
s31, generating coefficients and noise by using a trained generator, wherein the input and output of a generator model are normalization results, and the size of the generated results is adjusted according to the data range of a data set;
s32, combining the standard signal model to generate a mixed signal：
（4）；
wherein ,、/>respectively representing the generation coefficient, the generation noise, < >>A standard signal model is represented, and k represents the number of signal types in the mixed radio signal; m represents the signal length, and N represents the number of generations.
9. A server comprising a processor and a memory, wherein the memory stores at least one program, and the program is loaded and executed by the processor to implement the wireless signal intelligent detection and identification method according to any one of claims 1 to 8.
10. A computer-readable storage medium, wherein at least one program is stored in the storage medium, and the program is loaded and executed by a processor to implement the wireless signal intelligent detection and identification method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311006207.5A CN116756637B (en) | 2023-08-10 | 2023-08-10 | Wireless signal intelligent detection and identification method and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311006207.5A CN116756637B (en) | 2023-08-10 | 2023-08-10 | Wireless signal intelligent detection and identification method and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116756637A true CN116756637A (en) | 2023-09-15 |
| CN116756637B CN116756637B (en) | 2023-12-05 |
Family
ID=87953528
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311006207.5A Active CN116756637B (en) | 2023-08-10 | 2023-08-10 | Wireless signal intelligent detection and identification method and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116756637B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180060758A1 (en) * | 2016-08-30 | 2018-03-01 | Los Alamos National Security, Llc | Source identification by non-negative matrix factorization combined with semi-supervised clustering |
| CN111650655A (en) * | 2020-06-17 | 2020-09-11 | 桂林电子科技大学 | Non-negative matrix factorization supervised transient electromagnetic signal noise reduction method |
| US20230121812A1 (en) * | 2021-10-15 | 2023-04-20 | International Business Machines Corporation | Data augmentation for training artificial intelligence model |
| CN116230001A (en) * | 2023-03-10 | 2023-06-06 | 中国农业银行股份有限公司 | Mixed voice separation method, device, equipment and storage medium |
-
2023
- 2023-08-10 CN CN202311006207.5A patent/CN116756637B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180060758A1 (en) * | 2016-08-30 | 2018-03-01 | Los Alamos National Security, Llc | Source identification by non-negative matrix factorization combined with semi-supervised clustering |
| CN111650655A (en) * | 2020-06-17 | 2020-09-11 | 桂林电子科技大学 | Non-negative matrix factorization supervised transient electromagnetic signal noise reduction method |
| US20230121812A1 (en) * | 2021-10-15 | 2023-04-20 | International Business Machines Corporation | Data augmentation for training artificial intelligence model |
| CN116230001A (en) * | 2023-03-10 | 2023-06-06 | 中国农业银行股份有限公司 | Mixed voice separation method, device, equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| HUAJI ZHOU: "《Generative Adversarial Network-Based Electromagnetic Signal Classification: A SemiSupervised Learning Framework》", IEEE, pages 157 - 169 * |
| 孟子轩等: "基于非负矩阵分解的次声信号分类方法", 《应用声学》, pages 627 - 636 * |
| 张梦阳: "《基于非负矩阵分解的频谱感知技术研究》", 《信息系统与网络》, pages 1 - 7 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116756637B (en) | 2023-12-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Salem et al. | Anomaly generation using generative adversarial networks in host-based intrusion detection | |
| CN109727246B (en) | Comparative learning image quality evaluation method based on twin network | |
| WO2018176889A1 (en) | Method for automatically identifying modulation mode for digital communication signal | |
| CN110113288B (en) | Design and demodulation method of OFDM demodulator based on machine learning | |
| CN111626371B (en) | Image classification method, device, equipment and readable storage medium | |
| CN110968845B (en) | Detection method for LSB steganography based on convolutional neural network generation | |
| CN109887047B (en) | Signal-image translation method based on generation type countermeasure network | |
| CN112014801B (en) | SPWVD and improved AlexNet based composite interference identification method | |
| Huang et al. | Hierarchical digital modulation classification using cascaded convolutional neural network | |
| CN116756637B (en) | Wireless signal intelligent detection and identification method and computer readable storage medium | |
| CN116383719A (en) | MGF radio frequency fingerprint identification method for LFM radar | |
| CN116471048A (en) | Real-time and efficient DDoS attack detection method and system for Internet of things | |
| CN112347921B (en) | PDW sequence preprocessing method, system, computer equipment and storage medium | |
| CN113420791B (en) | Access control method and device for edge network equipment and terminal equipment | |
| CN115392325A (en) | Multi-feature noise reduction modulation identification method based on cycleGan | |
| Tang et al. | Data augmentation for signal modulation classification using generative adverse network | |
| CN115913792B (en) | DGA domain name identification method, system and readable medium | |
| CN114997299B (en) | Radio frequency fingerprint identification method in resource limited environment | |
| Li et al. | Solving the data imbalance problem in network intrusion detection: A MP-CVAE based method | |
| Yuan et al. | Adversarial attack with adaptive gradient variance for deep fake fingerprint detection | |
| Perera et al. | A joint representation learning and feature modeling approach for one-class recognition | |
| CN117118765B (en) | IPV6 identity security authentication method and system | |
| CN117807526B (en) | Electromagnetic signal identification method based on cyclic spectrum feature selection and fusion mechanism | |
| Changbo et al. | Multimodal Feature Fusion Recognition of Modulated Signals Based on Image and Waveform Domain | |
| Zhao et al. | Feature Re-Balancing for Long-Tailed Visual Recognition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |