CN108616470A - Modulation Signals Recognition method based on convolutional neural networks - Google Patents
Modulation Signals Recognition method based on convolutional neural networks Download PDFInfo
- Publication number
- CN108616470A CN108616470A CN201810253650.5A CN201810253650A CN108616470A CN 108616470 A CN108616470 A CN 108616470A CN 201810253650 A CN201810253650 A CN 201810253650A CN 108616470 A CN108616470 A CN 108616470A
- Authority
- CN
- China
- Prior art keywords
- signal
- model
- data set
- neural networks
- training
- 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.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/0012—Modulated-carrier systems arrangements for identifying the type of modulation
-
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Digital Transmission Methods That Use Modulated Carrier Waves (AREA)
Abstract
The invention belongs to communication signal recognition fields to be not necessarily to manual extraction signal characteristic to propose the signal identification model of High-efficiency Sustainable.Under different signal-to-noise ratio, there is good recognition performance.The present invention, the Modulation Signals Recognition method based on convolutional neural networks, steps are as follows：(1) three-layer coil of one robust of training accumulates neural network model；(2) it uses Software Radio platform to generate modulated signal data set, the completeness of data is paid attention to when generating data set, avoids model over-fitting；(3) training dataset and test data set are built, and the sequence of adjustment data set ensures the robustness of model, training pattern and preservation model structure and parameter at random；(4) signal is acquired, the model preserved in load (3) can complete the automatic identification of signal, real-time signal cognitive system can be built based on this.Present invention is mainly applied to communication signal recognition occasions.
Description
Technical field
The invention belongs to communication signal recognition field, deep learning field devises a kind of based on convolutional neural networks
Signal identification model.
Background technology
National strategic studies report points out that China is a resource critical shortage and the serious development China of environmental pollution
Family, the development of WeiLai Technology have to the constraint in face of resource and energy consumption.Therefore advanced deployment, the related reply future development of development
The theory and technology research of bottleneck has become great demand.The electromagnetic environment of complicated isomery brings huge choose for signal processing
War, the technology that there is an urgent need to develop civil-military inosculation special applications, public's system is combined with dedicated system, to effectively improve frequency spectrum
Utilization ratio, improvement environment, collaboration coexist.It can be docked by the identification and analyzing processing to signal of communication in the communications field
The information processing got off provides more information with application, has in military surveillance, electronic countermeasure, wireless network secure etc.
Very important meaning.
The modulation system of signal of communication is its more important technical characteristic, and identification signal of communication modulation type will be
The further analyzing processing of signal provides foundation.In non-cooperating communication, signal modulation pattern-recognition is as signal detection and letter
Number demodulation intermediate steps, be the key that one step of signal analysis and processing.With growing, the kind of signal of communication of the communication technology
Class is more and more various, and signal environment also becomes increasingly complex.Therefore find a kind of automatic signal identification technology of High-efficiency Sustainable at
For research hotspot.
Deep learning is a new field of machine learning research, it imitates the mechanism of human brain to be solved to data
It releases, made breakthrough progress in the research of multiple application fields such as language identification and computer identification.By deep learning
Combine with communication signal recognition so that the study and identify the tune for receiving signal that signal detection apparatus can be adaptive
Molding formula increases its stability, improving information handling rate and national defense and military level has to improving communication signal recognition efficiency
Important role.Convolutional neural networks have with the special construction that its local weight is shared in terms of speech recognition and image procossing
Unique superiority, layout shares the complexity for reducing network closer to actual biological neural network, weights.
Convolutional neural networks and general neural network difference lies in, convolutional neural networks contain one by convolutional layer and
The feature extractor that sub-sampling layer is constituted.In the convolutional layer of convolutional neural networks, a neuron is only neural with part adjacent bed
Member connection.In a convolutional layer of CNN (convolutional neural networks), several characteristic planes (featureMap) are generally comprised,
Each characteristic plane is made of the neuron of some rectangular arrangeds, and the neuron of same characteristic plane shares weights, here altogether
The weights enjoyed are exactly convolution kernel.Convolution kernel initializes generally in the form of random decimal matrix, is rolled up in the training process of network
Study is obtained rational weights by product core.The direct benefit that shared weights (convolution kernel) are brought is the company reduced between each layer of network
It connects, while reducing the risk of over-fitting again.Sub-sampling is also referred to as pond (pooling), usually there is mean value sub-sampling (mean
) and two kinds of forms of maximum value sub-sampling (max pooling) pooling.Sub-sampling is considered as a kind of special convolution process.
Convolution sum sub-sampling enormously simplifies model complexity, reduces the parameter of model.
Invention content
In order to overcome the deficiencies of the prior art, the present invention is directed to propose the signal identification model of High-efficiency Sustainable, is based on convolution
The powerful ability in feature extraction of neural network trains the three-layer coil of a robust to accumulate nerve net using original real part imaginary part I/Q data
Network model is not necessarily to manual extraction signal characteristic.Under different signal-to-noise ratio, there is good recognition performance.For this purpose, the present invention passes through
It adopts the following technical scheme that and is achieved, the Modulation Signals Recognition method based on convolutional neural networks, steps are as follows：
(1) based on the powerful ability in feature extraction of convolutional neural networks, one is trained using original real part imaginary part I/Q data
The three-layer coil of robust accumulates neural network model, and selection includes the suitable hyper parameter of convolution kernel, activation primitive and object function, with
Obtain preferred network model and parameter more new strategy；
(2) it uses Software Radio platform to generate modulated signal data set, the completeness of data is paid attention to when generating data set,
Avoid model over-fitting；
(3) training dataset and test data set are built, and the sequence of adjustment data set ensures the robust of model at random
Property, training pattern and preservation model structure and parameter；
(4) signal is acquired, the model preserved in load (3) can complete the automatic identification of signal, can be built based on this
Real-time signal cognitive system.
Model includes continuous three convolutional layers and a full articulamentum, wherein two-dimentional real part imaginary part I/Q signal be processed into 2 ×
400 size inputs network, and three convolutional layers separately include 256,128,80 convolution kernels, first convolutional layer convolution kernel it is big
Small is 2 × 7, and second convolutional layer convolution kernel size is 1 × 5, and third convolutional layer convolution kernel size is 1 × 3, last full connection
Layer exports 6 vectors, indicates 6 classifications of modulated signal.
6 kinds of modulated signal binary phase shift keying BPSK, quaternary phase-shift keying (PSK)s are generated using Software Radio platform
QPSK, octal system phase-shift keying (PSK) 8PSK, quadrature amplitude modulation 16QAM, frequency shift keying fsk, quaternary frequency shift keying 4FSK conducts
Data set；The processing procedure for emitting signal includes modulation, shaping filter, up-sampling and fixed point, is emitted by radio-frequency module；
Receiving terminal receives signal in such a way that feeder line is direct-connected or aerial radiation, output base band letter after downconverted, sampling, quantization
Number；It often acquires a frame just to be stored, changes different modulation systems and noise parameter, it is emitted just to obtain various modulated signals
Data after reception, software radio receiver sample frequency are set as 1.92MHz.
Training pattern comprises the concrete steps that each convolutional layer of model is using the linear unit R eLU (rectified of amendment
Linear it) is used as activation primitive, the activation primitive using more classification activation primitive softmax as last layer, the mesh of model
Scalar functions are to intersect entropy function, and expression formula is as follows：
H (p, q)=∑xp(x)logq(x) (1)
Wherein, p indicates that the distribution of authentic signature, q are then the predictive marker distribution of the model after training, and p (x) is distribution p
Entropy, q (x) is the entropy of prediction distribution q, and cross entropy loss function H (p, q) can weigh the similitude of p and q.
Over-fitting is prevented using dropout technologies, adds dropout layers below at every layer, setting dropout is 0.5, is adopted
With propagated forward and backpropagation techniques come training pattern, weight is updated, using stochastic gradient descent SGD methods -- Adam
(Adaptive Moment Estimation), by the way of batch training, batch size is 256, model training 30 on GPU
Period.
The features of the present invention and advantageous effect are：
The deep learning model of the present invention can well solve communication signal recognition problem, and in different signal-to-noise ratio
Under have very high accuracy of identification.Fig. 2 illustrates the training process of model, it can be seen that model is received rapidly in 5 epoch
It holds back, model has tended towards stability after 15 epoch.
In order to probe into influence of the different sample data length to recognition performance, this experiment produces different sample lengths
(200,400,800) 3 data sets are trained respectively under same model and experiment condition.The results are shown in Figure 3, can
To see that in the case of low signal-to-noise ratio, sample length can have an impact recognition performance.The case where wherein sample length is 800
Under, recognition performance is best.But in high s/n ratio, influence of the sample length to recognition performance is little.All in all, signal-to-noise ratio
When more than -4dB, model can reach 90% or more recognition accuracy.
Fig. 4 and Fig. 5 illustrates the recognition accuracy of different modulated signals under the conditions of -9dB and 10dB signal-to-noise ratio.In -9dB
Under conditions of, it can be seen that 2FSK and 4FSK, which has some, to be obscured, 8PSK and QPSK have it is slight obscure, but totality or standard
Really prediction.Under conditions of 10dB, it can be seen that gem-pure diagonal line, recognition performance are preferable.
Description of the drawings：
Fig. 1 deep learning models.
Fig. 2 model trainings loss (loss function).
The performance of Fig. 3 difference sample lengths compares.
The recognition performance (SNR=-9dB) of Fig. 4 different modulated signals.
The recognition performance (SNR=10dB) of Fig. 5 different modulated signals.
Specific implementation mode
The present invention devises the signal identification model based on convolutional neural networks, has reached fine in Modulation Signals Recognition
Effect.
The signal identification model for the High-efficiency Sustainable that the present invention designs, based on the powerful feature extraction energy of convolutional neural networks
Power trains the three-layer coil of a robust to accumulate neural network model, is not necessarily to manual extraction signal characteristic using original I Q data.Not
Under same signal-to-noise ratio, there is good recognition performance to can reach 90% recognition accuracy when wherein signal-to-noise ratio is more than -4dB.
(1) model structure
The implementation model of the present invention is as shown in Figure 1.
Model includes continuous three convolutional layers (Conv) and a full articulamentum (Dense), wherein at I/Q signal (two dimension)
Manage into 2 × 400 size input network.Three convolutional layers separately include 256,128,80 convolution kernels, first convolutional layer volume
The size of product core is 2 × 7, and second convolutional layer convolution kernel size is 1 × 5, and third convolutional layer convolution kernel size is 1 × 3, most
Full articulamentum exports 6 vectors afterwards, indicates modulated signal classification (6 class).
(2) data set generates
The present invention using Software Radio platform produce 6 kinds of modulated signals (BPSK, QPSK, 8PSK, 16QAM, FSK,
4FSK) it is used as data set.The processing procedure for emitting signal includes modulation, shaping filter, up-sampling and fixed point, passes through radio frequency mould
Block emits.In receiving terminal, signal is received in such a way that feeder line is direct-connected or aerial radiation, it is defeated after downconverted, sampling, quantization
Go out baseband signal.It often acquires a frame just to be stored, changes different modulation systems and noise parameter, various modulation can be obtained
Data after the emitted reception of signal, software radio receiver sample frequency are set as 1.92MHz.
In storing process, each sample separately includes each N number of sampled point in the roads I and the roads Q, and early period, training pattern was needed to each
Sample adds label, such as BPSK, QPSK etc., using the dictionary storage organization of Python, dictionary key corresponding label (BPSK etc.),
The value of dictionary corresponds to M sample.In order to improve the reliability of the adjustment model, different random noises is added in each sample.Data set includes
6 kind modulated signal data of the signal-to-noise ratio from -9dB to 10dB, 86400 samples, each sample storage form are 2 × 400 altogether
I/Q data.43200 samples are upset at random as training set, remaining 43200 samples are upset at random as test set.
(3) model training
Each convolutional layer of model is used as activation primitive using rectified linear (ReLU), compared to sigmoid
With tanh functions, ReLU has huge acceleration for the convergence of SGD.This may be linear, unsaturated characteristic by it
It is caused.ReLU only needs a threshold value to can be obtained by activation value, and (index) operation of a lot of complexity is calculated without spending.It adopts
Use softmax as the activation primitive of last layer.The object function of model is to intersect entropy function, and expression formula is as follows：
H (p, q)=∑xp(x)logq(x) (1)
In order to avoid over-fitting, this model prevents over-fitting using dropout technologies, and dropout is added below at every layer
Layer, setting dropout are 0.5.Using propagated forward and backpropagation techniques come training pattern, weight is updated.The present invention uses
Stochastic gradient descent (SGD) method -- the Adam (Adaptive Moment Estimation) of optimization.SGD needs artificial adjust
Whole many parameters, such as learning rate, convergence criterion etc..In addition, it is the method for sequence, it is unfavorable for GPU parallel or at distribution
Reason.And Adam adjusts the learning rate of each parameter using the single order moments estimation and second order moments estimation dynamic of gradient.The advantages of Adam
It essentially consists in after bias correction, iterative learning rate has a determining range each time so that parameter is more steady.Using batch
Trained mode, batch size are 256, model 30 epoch of training on the GPU of GTX1080.
The general steps of the present invention are as follows：
(1) based on the powerful ability in feature extraction of convolutional neural networks, the three of a robust are trained using original I Q data
Layer convolutional neural networks model, is not necessarily to manual extraction signal characteristic.Wherein needing to select suitable hyper parameter, (such as convolution kernel swashs
Function living, object function etc.) to obtain optimal network model and optimal parameter more new strategy.
(2) use Software Radio platform generate 6 kinds of modulated signal data sets (BPSK, QPSK, 8PSK, 16QAM, FSK,
4FSK).The completeness that data are paid attention to when generating data set, avoids model over-fitting.
(3) training dataset and test data set are built, and the sequence of adjustment data set ensures the robust of model at random
Property.Training pattern and preservation model structure and parameter.
(4) signal is acquired, the model preserved in load (3) can be completed the automatic identification of signal, reality can be built based on this
When signal cognitive system.
Claims (5)
1. a kind of Modulation Signals Recognition method based on convolutional neural networks, characterized in that steps are as follows：
(1) based on the powerful ability in feature extraction of convolutional neural networks, a robust is trained using original real part imaginary part I/Q data
Three-layer coil accumulate neural network model, selection include convolution kernel, activation primitive and object function suitable hyper parameter, to obtain
Preferred network model and parameter more new strategy；
(2) it uses Software Radio platform to generate modulated signal data set, the completeness of data is paid attention to when generating data set, is avoided
Model over-fitting；
(3) training dataset and test data set are built, and the sequence of adjustment data set ensures the robustness of model, instruction at random
Practice model and preservation model structure and parameter；
(4) signal is acquired, the model preserved in load (3) can complete the automatic identification of signal, can be built in real time based on this
Signal cognitive system.
2. the Modulation Signals Recognition method based on convolutional neural networks as described in claim 1, characterized in that model includes to connect
Continue three convolutional layers and a full articulamentum, wherein two-dimentional real part imaginary part I/Q signal is processed into 2 × 400 size input network,
Three convolutional layers separately include 256,128,80 convolution kernels, and the size of first convolutional layer convolution kernel is 2 × 7, second volume
Lamination convolution kernel size is 1 × 5, and third convolutional layer convolution kernel size is 1 × 3, and last full articulamentum exports 6 vectors,
Indicate 6 classifications of modulated signal.
3. the Modulation Signals Recognition method based on convolutional neural networks as described in claim 1, characterized in that using software without
Line level platform generates 6 kinds of modulated signal binary phase shift keying BPSK, quaternary phase-shift keying (PSK) QPSK, octal system phase-shift keying (PSK)
8PSK, quadrature amplitude modulation 16QAM, frequency shift keying fsk, quaternary frequency shift keying 4FSK are as data set；Emit the place of signal
Reason process includes modulation, shaping filter, up-sampling and fixed point, is emitted by radio-frequency module；It is direct-connected using feeder line in receiving terminal
Or the mode of aerial radiation receives signal, and baseband signal is exported after downconverted, sampling, quantization；A frame is often acquired just to carry out
Storage, changes different modulation systems and noise parameter, just obtains the data after the emitted reception of various modulated signals, software without
Line electricity receiver sample frequency is set as 1.92MHz.
4. the Modulation Signals Recognition method based on convolutional neural networks as described in claim 1, characterized in that training pattern
It comprises the concrete steps that, each convolutional layer of model is used as activation letter using linear unit R eLU (rectified linear) is corrected
The object function of number, the activation primitive using more classification activation primitive softmax as last layer, model is cross entropy letter
Number, expression formula are as follows：
H (p, q)=∑xp(x)logq(x) (1)
Wherein, p indicates that the distribution of authentic signature, q are then the predictive marker distribution of the model after training, and p (x) is the entropy of distribution p,
Q (x) is the entropy of prediction distribution q, and cross entropy loss function H (p, q) can weigh the similitude of p and q.
5. the Modulation Signals Recognition method based on convolutional neural networks as described in claim 1, characterized in that use
Dropout technologies prevent over-fitting, every layer below add dropout layer, setting dropout be 0.5, using propagated forward with
Backpropagation techniques carry out training pattern, update weight, using stochastic gradient descent SGD methods -- Adam (Adaptive Moment
Estimation), by the way of batch training, batch size is 256, model 30 periods of training on GPU.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810253650.5A CN108616470A (en) | 2018-03-26 | 2018-03-26 | Modulation Signals Recognition method based on convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810253650.5A CN108616470A (en) | 2018-03-26 | 2018-03-26 | Modulation Signals Recognition method based on convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108616470A true CN108616470A (en) | 2018-10-02 |
Family
ID=63658756
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810253650.5A Pending CN108616470A (en) | 2018-03-26 | 2018-03-26 | Modulation Signals Recognition method based on convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108616470A (en) |
Cited By (42)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109302238A (en) * | 2018-11-29 | 2019-02-01 | 武汉邮电科学研究院有限公司 | A kind of smooth I/Q modulator parameter adjusting method and system |
| CN109361635A (en) * | 2018-11-23 | 2019-02-19 | 泰山学院 | Subsurface communication Modulation Mode Recognition method and system based on depth residual error network |
| CN109462564A (en) * | 2018-11-16 | 2019-03-12 | 泰山学院 | Subsurface communication Modulation Mode Recognition method and system based on deep neural network |
| CN109472360A (en) * | 2018-10-30 | 2019-03-15 | 北京地平线机器人技术研发有限公司 | Update method, updating device and the electronic equipment of neural network |
| CN109471074A (en) * | 2018-11-09 | 2019-03-15 | 西安电子科技大学 | Recognition Method of Radar Emitters based on singular value decomposition Yu one-dimensional CNN network |
| CN109523023A (en) * | 2018-11-16 | 2019-03-26 | 泰山学院 | A kind of deep learning network and system for subsurface communication Modulation Identification |
| CN109547374A (en) * | 2018-11-23 | 2019-03-29 | 泰山学院 | A kind of depth residual error network and system for subsurface communication Modulation Identification |
| CN109614952A (en) * | 2018-12-27 | 2019-04-12 | 成都数之联科技有限公司 | A kind of target echo detection recognition methods based on Waterfall plot |
| CN109657604A (en) * | 2018-12-17 | 2019-04-19 | 中国人民解放军战略支援部队信息工程大学 | Satellite width phase signals identification demodulation method and device based on Recognition with Recurrent Neural Network |
| CN109787927A (en) * | 2019-01-03 | 2019-05-21 | 荆门博谦信息科技有限公司 | Modulation Identification method and apparatus based on deep learning |
| CN109890043A (en) * | 2019-02-28 | 2019-06-14 | 浙江工业大学 | A kind of wireless signal noise-reduction method based on production confrontation network |
| CN109922019A (en) * | 2019-02-26 | 2019-06-21 | 天津大学 | Intelligent communication method based on deep learning |
| CN109936423A (en) * | 2019-03-12 | 2019-06-25 | 中国科学院微电子研究所 | A kind of training method, device and the recognition methods of fountain codes identification model |
| CN110266620A (en) * | 2019-07-08 | 2019-09-20 | 电子科技大学 | 3D MIMO-OFDM system channel estimation method based on convolutional neural networks |
| CN110515096A (en) * | 2019-08-20 | 2019-11-29 | 东南大学 | Satellite navigation interference signal identification device and its method based on convolutional neural networks |
| CN110659684A (en) * | 2019-09-23 | 2020-01-07 | 中国人民解放军海军航空大学 | Convolutional neural network-based STBC signal identification method |
| CN110798417A (en) * | 2019-10-24 | 2020-02-14 | 北京邮电大学 | Signal modulation identification method and device based on cyclic residual error network |
| CN110889496A (en) * | 2019-12-11 | 2020-03-17 | 北京工业大学 | Human brain effect connection identification method based on confrontation generation network |
| WO2020087293A1 (en) * | 2018-10-30 | 2020-05-07 | 华为技术有限公司 | Communication receiver and method for processing signal |
| CN111259861A (en) * | 2020-02-18 | 2020-06-09 | 西北工业大学 | Underwater acoustic communication signal modulation mode identification method under data set unbalanced condition |
| CN111464469A (en) * | 2020-03-12 | 2020-07-28 | 南京航空航天大学 | Hybrid digital modulation mode identification method based on neural network |
| CN111553248A (en) * | 2020-04-24 | 2020-08-18 | 中国人民解放军海军航空大学 | STBC signal-oriented identification network STBCCNN |
| CN111585925A (en) * | 2020-04-18 | 2020-08-25 | 西北工业大学 | Robust real-time radio frequency signal modulation identification method based on deep learning |
| CN111585922A (en) * | 2020-03-23 | 2020-08-25 | 成都奥特为科技有限公司 | Modulation mode identification method based on convolutional neural network |
| CN111585923A (en) * | 2020-03-23 | 2020-08-25 | 成都奥特为科技有限公司 | Modulation mode recognition device and system based on convolutional neural network |
| CN111628833A (en) * | 2020-06-10 | 2020-09-04 | 桂林电子科技大学 | MIMO antenna number estimation method based on convolutional neural network |
| CN111988252A (en) * | 2020-08-24 | 2020-11-24 | 成都华日通讯技术股份有限公司 | Signal modulation mode identification method based on deep learning |
| CN112364753A (en) * | 2020-11-09 | 2021-02-12 | 北京计算机技术及应用研究所 | Low-voltage power carrier abnormal signal detection method |
| CN112565128A (en) * | 2020-11-28 | 2021-03-26 | 西北工业大学 | Radio signal modulation recognition network based on hybrid neural network and implementation method |
| CN112731522A (en) * | 2020-12-14 | 2021-04-30 | 中国地质大学（武汉） | Intelligent recognition method, device and equipment for seismic stratum and storage medium |
| CN113726350A (en) * | 2021-08-09 | 2021-11-30 | 哈尔滨工程大学 | Deep neural network-based strong correlation self-interference cancellation method |
| CN113723353A (en) * | 2021-09-13 | 2021-11-30 | 上海交通大学 | Modulated signal identification method based on CBD network under random multipath interference condition |
| CN113723556A (en) * | 2021-09-08 | 2021-11-30 | 中国人民解放军国防科技大学 | Modulation mode identification method based on entropy weighting-multi-mode domain antagonistic neural network |
| CN114002334A (en) * | 2021-09-29 | 2022-02-01 | 西安交通大学 | Structural damage acoustic emission signal identification method and device and storage medium |
| CN114285701A (en) * | 2021-11-30 | 2022-04-05 | 西安电子科技大学重庆集成电路创新研究院 | Method, system, equipment and terminal for identifying transmitting power of master user |
| CN114615118A (en) * | 2022-03-14 | 2022-06-10 | 中国人民解放军国防科技大学 | Modulation identification method based on multi-terminal convolution neural network |
| CN114900407A (en) * | 2022-07-12 | 2022-08-12 | 南京科伊星信息科技有限公司 | Modulation mode automatic identification and countermeasure method based on data enhancement |
| CN115277324A (en) * | 2022-07-25 | 2022-11-01 | 电信科学技术第五研究所有限公司 | FSK signal identification method based on convolutional neural network |
| CN115296758A (en) * | 2022-07-07 | 2022-11-04 | 中山大学 | Method, system, computer device and storage medium for identifying interference signal |
| CN116132235A (en) * | 2023-01-29 | 2023-05-16 | 北京林业大学 | Continuous phase modulation signal demodulation method based on deep learning |
| CN117354106A (en) * | 2023-12-06 | 2024-01-05 | 中国海洋大学 | Communication signal modulation identification method and system based on heavy parameter causal convolution network |
| CN112364753B (en) * | 2020-11-09 | 2024-06-28 | 北京计算机技术及应用研究所 | Method for detecting abnormal signal of low-voltage power carrier |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160307072A1 (en) * | 2015-04-17 | 2016-10-20 | Nec Laboratories America, Inc. | Fine-grained Image Classification by Exploring Bipartite-Graph Labels |
| CN107220606A (en) * | 2017-05-22 | 2017-09-29 | 西安电子科技大学 | The recognition methods of radar emitter signal based on one-dimensional convolutional neural networks |
| CN107342962A (en) * | 2017-07-03 | 2017-11-10 | 北京邮电大学 | Deep learning intelligence Analysis On Constellation Map method based on convolutional neural networks |
| CN107682216A (en) * | 2017-09-01 | 2018-02-09 | 南京南瑞集团公司 | A kind of network traffics protocol recognition method based on deep learning |
-
2018
- 2018-03-26 CN CN201810253650.5A patent/CN108616470A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160307072A1 (en) * | 2015-04-17 | 2016-10-20 | Nec Laboratories America, Inc. | Fine-grained Image Classification by Exploring Bipartite-Graph Labels |
| CN107220606A (en) * | 2017-05-22 | 2017-09-29 | 西安电子科技大学 | The recognition methods of radar emitter signal based on one-dimensional convolutional neural networks |
| CN107342962A (en) * | 2017-07-03 | 2017-11-10 | 北京邮电大学 | Deep learning intelligence Analysis On Constellation Map method based on convolutional neural networks |
| CN107682216A (en) * | 2017-09-01 | 2018-02-09 | 南京南瑞集团公司 | A kind of network traffics protocol recognition method based on deep learning |
Cited By (58)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109472360B (en) * | 2018-10-30 | 2020-09-04 | 北京地平线机器人技术研发有限公司 | Neural network updating method and updating device and electronic equipment |
| WO2020087293A1 (en) * | 2018-10-30 | 2020-05-07 | 华为技术有限公司 | Communication receiver and method for processing signal |
| US11328180B2 (en) | 2018-10-30 | 2022-05-10 | Beijing Horizon Robotics Technology Research And Development Co., Ltd. | Method for updating neural network and electronic device |
| CN109472360A (en) * | 2018-10-30 | 2019-03-15 | 北京地平线机器人技术研发有限公司 | Update method, updating device and the electronic equipment of neural network |
| CN109471074A (en) * | 2018-11-09 | 2019-03-15 | 西安电子科技大学 | Recognition Method of Radar Emitters based on singular value decomposition Yu one-dimensional CNN network |
| CN109462564A (en) * | 2018-11-16 | 2019-03-12 | 泰山学院 | Subsurface communication Modulation Mode Recognition method and system based on deep neural network |
| CN109523023A (en) * | 2018-11-16 | 2019-03-26 | 泰山学院 | A kind of deep learning network and system for subsurface communication Modulation Identification |
| CN109462564B (en) * | 2018-11-16 | 2021-08-03 | 泰山学院 | Underwater communication modulation mode identification method and system based on deep neural network |
| CN109361635A (en) * | 2018-11-23 | 2019-02-19 | 泰山学院 | Subsurface communication Modulation Mode Recognition method and system based on depth residual error network |
| CN109361635B (en) * | 2018-11-23 | 2021-06-04 | 泰山学院 | Underwater communication modulation mode identification method and system based on depth residual error network |
| CN109547374A (en) * | 2018-11-23 | 2019-03-29 | 泰山学院 | A kind of depth residual error network and system for subsurface communication Modulation Identification |
| CN109547374B (en) * | 2018-11-23 | 2021-11-23 | 泰山学院 | Depth residual error network and system for underwater communication modulation recognition |
| CN109302238A (en) * | 2018-11-29 | 2019-02-01 | 武汉邮电科学研究院有限公司 | A kind of smooth I/Q modulator parameter adjusting method and system |
| CN109657604A (en) * | 2018-12-17 | 2019-04-19 | 中国人民解放军战略支援部队信息工程大学 | Satellite width phase signals identification demodulation method and device based on Recognition with Recurrent Neural Network |
| CN109614952A (en) * | 2018-12-27 | 2019-04-12 | 成都数之联科技有限公司 | A kind of target echo detection recognition methods based on Waterfall plot |
| CN109787927A (en) * | 2019-01-03 | 2019-05-21 | 荆门博谦信息科技有限公司 | Modulation Identification method and apparatus based on deep learning |
| CN109922019A (en) * | 2019-02-26 | 2019-06-21 | 天津大学 | Intelligent communication method based on deep learning |
| CN109890043A (en) * | 2019-02-28 | 2019-06-14 | 浙江工业大学 | A kind of wireless signal noise-reduction method based on production confrontation network |
| CN109936423B (en) * | 2019-03-12 | 2021-11-30 | 中国科学院微电子研究所 | Training method, device and recognition method of fountain code recognition model |
| CN109936423A (en) * | 2019-03-12 | 2019-06-25 | 中国科学院微电子研究所 | A kind of training method, device and the recognition methods of fountain codes identification model |
| CN110266620A (en) * | 2019-07-08 | 2019-09-20 | 电子科技大学 | 3D MIMO-OFDM system channel estimation method based on convolutional neural networks |
| CN110515096A (en) * | 2019-08-20 | 2019-11-29 | 东南大学 | Satellite navigation interference signal identification device and its method based on convolutional neural networks |
| CN110659684A (en) * | 2019-09-23 | 2020-01-07 | 中国人民解放军海军航空大学 | Convolutional neural network-based STBC signal identification method |
| CN110798417A (en) * | 2019-10-24 | 2020-02-14 | 北京邮电大学 | Signal modulation identification method and device based on cyclic residual error network |
| US11909563B2 (en) | 2019-10-24 | 2024-02-20 | Beijing University Of Posts And Telecommunications | Method and apparatus for modulation recognition of signals based on cyclic residual network |
| CN110798417B (en) * | 2019-10-24 | 2020-07-31 | 北京邮电大学 | Signal modulation identification method and device based on cyclic residual error network |
| CN110889496A (en) * | 2019-12-11 | 2020-03-17 | 北京工业大学 | Human brain effect connection identification method based on confrontation generation network |
| CN111259861A (en) * | 2020-02-18 | 2020-06-09 | 西北工业大学 | Underwater acoustic communication signal modulation mode identification method under data set unbalanced condition |
| CN111464469A (en) * | 2020-03-12 | 2020-07-28 | 南京航空航天大学 | Hybrid digital modulation mode identification method based on neural network |
| CN111585922A (en) * | 2020-03-23 | 2020-08-25 | 成都奥特为科技有限公司 | Modulation mode identification method based on convolutional neural network |
| CN111585923A (en) * | 2020-03-23 | 2020-08-25 | 成都奥特为科技有限公司 | Modulation mode recognition device and system based on convolutional neural network |
| CN111585925A (en) * | 2020-04-18 | 2020-08-25 | 西北工业大学 | Robust real-time radio frequency signal modulation identification method based on deep learning |
| CN111553248A (en) * | 2020-04-24 | 2020-08-18 | 中国人民解放军海军航空大学 | STBC signal-oriented identification network STBCCNN |
| CN111628833A (en) * | 2020-06-10 | 2020-09-04 | 桂林电子科技大学 | MIMO antenna number estimation method based on convolutional neural network |
| CN111988252A (en) * | 2020-08-24 | 2020-11-24 | 成都华日通讯技术股份有限公司 | Signal modulation mode identification method based on deep learning |
| CN112364753B (en) * | 2020-11-09 | 2024-06-28 | 北京计算机技术及应用研究所 | Method for detecting abnormal signal of low-voltage power carrier |
| CN112364753A (en) * | 2020-11-09 | 2021-02-12 | 北京计算机技术及应用研究所 | Low-voltage power carrier abnormal signal detection method |
| CN112565128A (en) * | 2020-11-28 | 2021-03-26 | 西北工业大学 | Radio signal modulation recognition network based on hybrid neural network and implementation method |
| CN112731522A (en) * | 2020-12-14 | 2021-04-30 | 中国地质大学（武汉） | Intelligent recognition method, device and equipment for seismic stratum and storage medium |
| CN113726350A (en) * | 2021-08-09 | 2021-11-30 | 哈尔滨工程大学 | Deep neural network-based strong correlation self-interference cancellation method |
| CN113723556A (en) * | 2021-09-08 | 2021-11-30 | 中国人民解放军国防科技大学 | Modulation mode identification method based on entropy weighting-multi-mode domain antagonistic neural network |
| CN113723556B (en) * | 2021-09-08 | 2022-05-31 | 中国人民解放军国防科技大学 | Modulation mode identification method based on entropy weighting-multi-mode domain antagonistic neural network |
| CN113723353A (en) * | 2021-09-13 | 2021-11-30 | 上海交通大学 | Modulated signal identification method based on CBD network under random multipath interference condition |
| CN113723353B (en) * | 2021-09-13 | 2023-12-12 | 上海交通大学 | Modulation signal identification method based on CBD network under random multipath interference condition |
| CN114002334A (en) * | 2021-09-29 | 2022-02-01 | 西安交通大学 | Structural damage acoustic emission signal identification method and device and storage medium |
| CN114285701A (en) * | 2021-11-30 | 2022-04-05 | 西安电子科技大学重庆集成电路创新研究院 | Method, system, equipment and terminal for identifying transmitting power of master user |
| CN114285701B (en) * | 2021-11-30 | 2024-03-29 | 西安电子科技大学重庆集成电路创新研究院 | Method, system, equipment and terminal for identifying transmitting power of main user |
| CN114615118B (en) * | 2022-03-14 | 2023-09-22 | 中国人民解放军国防科技大学 | Modulation identification method based on multi-terminal convolution neural network |
| CN114615118A (en) * | 2022-03-14 | 2022-06-10 | 中国人民解放军国防科技大学 | Modulation identification method based on multi-terminal convolution neural network |
| CN115296758B (en) * | 2022-07-07 | 2023-08-11 | 中山大学 | Method, system, computer equipment and storage medium for identifying interference signals |
| CN115296758A (en) * | 2022-07-07 | 2022-11-04 | 中山大学 | Method, system, computer device and storage medium for identifying interference signal |
| CN114900407B (en) * | 2022-07-12 | 2022-10-14 | 南京科伊星信息科技有限公司 | Modulation mode automatic identification and countermeasure method based on data enhancement |
| CN114900407A (en) * | 2022-07-12 | 2022-08-12 | 南京科伊星信息科技有限公司 | Modulation mode automatic identification and countermeasure method based on data enhancement |
| CN115277324B (en) * | 2022-07-25 | 2023-11-10 | 电信科学技术第五研究所有限公司 | FSK signal identification method based on convolutional neural network |
| CN115277324A (en) * | 2022-07-25 | 2022-11-01 | 电信科学技术第五研究所有限公司 | FSK signal identification method based on convolutional neural network |
| CN116132235A (en) * | 2023-01-29 | 2023-05-16 | 北京林业大学 | Continuous phase modulation signal demodulation method based on deep learning |
| CN117354106A (en) * | 2023-12-06 | 2024-01-05 | 中国海洋大学 | Communication signal modulation identification method and system based on heavy parameter causal convolution network |
| CN117354106B (en) * | 2023-12-06 | 2024-03-01 | 中国海洋大学 | Communication signal modulation identification method and system based on heavy parameter causal convolution network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108616470A (en) | Modulation Signals Recognition method based on convolutional neural networks | |
| Zhang et al. | An efficient deep learning model for automatic modulation recognition based on parameter estimation and transformation | |
| CN106059972B (en) | A kind of Modulation Identification method under MIMO correlated channels based on machine learning algorithm | |
| CN104994045B (en) | A kind of digital modulation mode automatic identification platform and method based on USRP platforms | |
| CN110162799A (en) | Model training method, machine translation method and relevant apparatus and equipment | |
| CN105512676A (en) | Food recognition method at intelligent terminal | |
| CN109344884A (en) | The method and device of media information classification method, training picture classification model | |
| CN108508411A (en) | Passive radar external sort algorithm signal recognition method based on transfer learning | |
| CN109474388B (en) | Low-complexity MIMO-NOMA system signal detection method based on improved gradient projection method | |
| CN109471074A (en) | Recognition Method of Radar Emitters based on singular value decomposition Yu one-dimensional CNN network | |
| Elbir et al. | Federated learning for physical layer design | |
| CN109922019A (en) | Intelligent communication method based on deep learning | |
| CN110808932B (en) | Multi-layer sensor rapid modulation identification method based on multi-distribution test data fusion | |
| CN116192307A (en) | Distributed cooperative multi-antenna cooperative spectrum intelligent sensing method, system, equipment and medium under non-Gaussian noise | |
| CN105959004A (en) | Single-precision ADC adaptive threshold quantification method based on large-scale MIMO | |
| CN115473771A (en) | Model evolution-based environment sensing method | |
| CN101964055B (en) | Visual perception mechansim simulation natural scene type identification method | |
| CN116566777B (en) | Frequency hopping signal modulation identification method based on graph convolution neural network | |
| Yadav et al. | Application of Machine Learning Framework for Next‐Generation Wireless Networks: Challenges and Case Studies | |
| Feng et al. | FCGCN: Feature Correlation Graph Convolution Network for Few-Shot Individual Identification | |
| CN116319190A (en) | GAN-based large-scale MIMO system channel estimation method, device, equipment and medium | |
| CN113343924B (en) | Modulation signal identification method based on cyclic spectrum characteristics and generation countermeasure network | |
| CN115856811A (en) | Micro Doppler feature target classification method based on deep learning | |
| Guo et al. | Custom convolutional layer designs for CNN based automatic modulation classification solution | |
| CN104517141B (en) | Radio frequency identification network topology method based on load balance Yu particle cluster algorithm |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181002 |