CN111507293A - Signal classification method based on graph convolution model - Google Patents
Signal classification method based on graph convolution model Download PDFInfo
- Publication number
- CN111507293A CN111507293A CN202010323837.5A CN202010323837A CN111507293A CN 111507293 A CN111507293 A CN 111507293A CN 202010323837 A CN202010323837 A CN 202010323837A CN 111507293 A CN111507293 A CN 111507293A
- Authority
- CN
- China
- Prior art keywords
- signal
- network
- representing
- graph
- moment
- 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
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 7
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000015654 memory Effects 0.000 claims abstract description 3
- 239000011159 matrix material Substances 0.000 claims description 19
- 230000004913 activation Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 8
- 230000007787 long-term memory Effects 0.000 claims description 5
- 230000006403 short-term memory Effects 0.000 claims description 5
- 230000001364 causal effect Effects 0.000 claims description 3
- 238000010586 diagram Methods 0.000 claims description 2
- 239000000126 substance Substances 0.000 claims description 2
- 230000017105 transposition Effects 0.000 claims description 2
- 238000004891 communication Methods 0.000 abstract description 3
- 238000013145 classification model Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- 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
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/045—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
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Digital Transmission Methods That Use Modulated Carrier Waves (AREA)
Abstract
The invention discloses a signal classification method based on graph convolution model, which is mainly applied to classifying the modulation mode of the received signals in a communication system so as to adopt a corresponding demodulation method to ensure that the communication systems are interconnected, a receiving end receives the modulation signals with fixed time sequence length, a preprocessing method such as in-phase orthogonal component decomposition is utilized to obtain proper multidimensional data, the processed signal data is input into a long-short term memory network to extract characteristics, the extracted characteristics of the current time t are ensured to be related to the signal data of the previous time, then graph structure representation of the signals is constructed by utilizing the characteristics, and finally graph structure information of the signals and the preprocessed multidimensional data are input into a graph convolution network and a full connection network to obtain the classification result of the signals.
Description
Technical Field
The invention belongs to the field of graph science, and particularly relates to a signal classification method based on a graph convolution model.
Background
The signal can be transmitted in a wired channel as a carrier of information, but cannot be directly transmitted through a wireless channel due to reasons such as low frequency and the like, and the signal modulation technology is required to be applied to the signal of a high-frequency carrier, so that the signal can be smoothly transmitted in the wireless channel. The modulation mode is from simple to complex, the modes are various, and the receiving end needs to know the modulation mode and corresponding parameters of the received signal to successfully demodulate the signal to obtain the original signal. Nowadays, the fifth generation (5G) cellular system will gradually cover the whole country, step into commercial use, and the effect of intelligent signal receiving and transmitting is realized to the communication signal of different modulation modes of quick, efficient discernment adoption, will help improving quality of service. One important technology is an Automatic Modulation Classification (AMC), and a receiving end executes AMC to identify a modulation type without knowing system parameters in advance, so that the frequency of exchanging prior protocol information is reduced, signaling overhead is greatly reduced, and transmission efficiency is improved. In the past decades, various methods have been proposed which can be generally divided into two categories: probability-based methods and feature-based methods. Probability-based methods use maximization based on likelihood functions under the assumption of probability density functions of the input signal to solve. While the probability-based approach may provide optimal performance in a bayesian sense, it requires sophisticated knowledge of the received signal and is computationally complex. In the feature-based approach, decisions are made based on extracted received signal features, such as High Order Statistics (HOS) and power spectral density. Compared with the probability-based method, the characteristic-based method is simple to implement and has near-optimal performance.
In recent years, as the machine learning technology has continuously made major breakthrough in various fields, many researchers also use the machine learning technology as a classifier for Feature extraction, such as a support vector machine, K neighbors, etc. Deep learning does not require manual engineering to extract features, advanced features can be automatically learned, and thus the methods have attracted much attention in the task of identifying complex and Deep architectures, and the methods have been used for solving the problem of high efficiency of the classification of the features of the network by using a new and Efficient method for carrying out the classification of the features of the network based on the map of the individual nodes, such as a new and Efficient method for carrying out the classification of the features of the network by using a map of the individual nodes, such as a central-family plan, a central-based classification model, a temporal-spatial distribution Graph L, a network classification model, a Graph of the network, a network classification model, a network classification method based on the characteristics of the individual nodes, a network classification method for estimating, a network classification of the network, a network classification method for expressing the characteristics of the network, a network classification method for realizing the accurate and a high efficiency of the classification of the network.
At present, a deep learning-based graph classification model has achieved a remarkable effect, and achieves certain results on social networks and biological information data, but research on signal modulation mode classification tasks is less.
Disclosure of Invention
The invention aims to provide a signal classification method based on a graph convolution model, which constructs a graph structure representation of a signal through features extracted by a long-term and short-term memory network and classifies the graph structure of the signal by utilizing a graph convolution depth model.
In order to achieve the purpose, the invention provides the following technical scheme:
a signal classification method based on graph convolution model, the signal classification method comprising the steps of:
(1) acquiring a modulation signal with a certain time sequence length, and preprocessing the modulation signal to obtain multi-dimensional modulation signal data;
(2) extracting the characteristics of the multi-dimensional modulation signal data by using a trained long-short term memory network (L STM) to ensure that the characteristics at the current moment are related to the modulation signal data at the previous moment;
(3) constructing graph structure representation information of the signal by using the extracted features;
(4) using the graph structure representation information of the signal and the multidimensional modulation signal data as input data of a trained Graph Convolution Network (GCN), and extracting the characteristics of the input data by utilizing the graph convolution network;
(5) and classifying the characteristics of the input data by using the trained full-connection network to obtain a signal classification result.
In the step (1), the step (c),
and carrying out in-phase and quadrature component decomposition on the modulation signal to realize preprocessing of the modulation signal, wherein the decomposition result is the multi-dimensional modulation signal data.
In the step (2), the influence degree of the network state at the previous moment on the feature extraction at the current moment is calculated by storing the network state at the previous moment so as to ensure the causal relationship of the system, and specifically, the feature extraction of the multi-dimensional modulation signal data by using the trained long and short term memory network comprises the following steps:
ft＝σ(Wf·[ht-1,xt]+bf)
it＝σ(Wi·[ht-1,xt]+bi)
ot＝σ(Wo·[ht-1,xt]+bo)
ht＝ot*tanh(Ct)
wherein x istInput data representing time t of signal, ht-1An output characteristic representing the time instant of the signal t-1, [,]representing cascade, sigma (-) and tanh (-) representing activation function to ensure the nonlinearity of network, W and b representing the weight of parameters to be trained, WfAnd bfParameters, f, representing L STM forget gate, respectivelytIs a vector between 0 and 1, which represents the network state C at the last momentt-1For the relevance degree of the extracted features at the current time, 0 represents no relevance, 1 represents the maximum value of the relevance degree, itIndicating the degree of influence of the input data at the current time t on the features, WiAnd biRespectively, the parameters of the input gate are indicated,
initial value, W, representing the state of the network at the present momentCAnd bCParameters respectively representing the state of computation, CtIndicating the state of the network at the present moment, otIndicating the probability of activation of the feature at the current moment of output, WoAnd boRespectively representing the parameters of the output gates, htRepresenting the output characteristics of the signal at the current time.
In the step (3), the characteristic H ∈ R extracted according to L STMT×dGraph structure representation information for constructing signal and adjacency matrix A ∈ R thereofT×TFirstly, normalizing the characteristics of each moment, and then calculating the similarity between every two characteristics of the moment, wherein the calculation formula is as follows:
A＝H′H′T
wherein, H'TThe method is characterized in that the method represents the transposition of the H' characteristic, A represents an adjacent matrix of a signal diagram structure, nodes represent each moment of a signal, and continuous edges represent the similarity of the characteristic between the two moments.
In the step (4), the adjacent matrix A of the graph structure representation information of the signal and the multi-dimensional modulation signal data X are used as the input of the graph convolution network, and the characteristics of the signal under the graph structure representation are calculated, and the calculation formula is as follows:
H*＝ReLU(LXW)
wherein L denotes Laplace matrix L ═ I-D-1/2AD-1/2D is adjacentA matrix of values connected to the matrix A, I being an identity matrix, H*∈RT×dIndicating the resulting node signature, W ∈ Rd×dIs the weight of a trained parameter, RT×dA matrix representing dimension T × d, d being the dimension of the preprocessed modulated signal data, Rd×dA matrix representing d × d dimensions, Re L U (·) represents the activation function.
In the step (5), the characteristics of the T moments obtained by the graph convolution network are averaged, and then the confidence of the signal to each modulation mode is obtained by the calculation of the full connection layer, wherein the calculation formula is as follows:
wherein the content of the first and second substances,
representing features of the output of the graph convolutional network, j represents the index of the feature, L∈ RlRepresents the probability of the signal being predicted for each class, l represents the number of signal modulation schemes, W represents the weight of the fully-connected layer, and softmax (·) represents the activation function.
Compared with the prior art, the invention has the beneficial effects that:
the signal classification method based on the graph convolution model utilizes the long-term and short-term memory network to extract the characteristics, ensures that the extracted characteristics at the current moment t are related to the signal data at the previous moment, ensures the causality of the system, utilizes the characteristics to convert the signals into a representation mode of a graph structure, and finally inputs graph structure information of the signals and preprocessed multidimensional data into a graph convolution network and a full-connection network to obtain the classification result of the signals. The classification result has high accuracy.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is an overall flowchart of a signal classification method based on a graph convolution model provided by the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the detailed description and specific examples, while indicating the scope of the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention.
Fig. 1 is an overall flowchart of a signal classification method based on a graph convolution model provided by the invention. As shown in fig. 1, the signal classification method based on graph convolution model includes the following steps:
(1) and receiving a modulation signal with the time sequence length of T, and obtaining appropriate multidimensional data by using methods such as in-phase orthogonal component decomposition and the like.
The signals of the experimental data are equal in number under the conditions that the signal-to-noise ratios are 0db, 5db, 10db, 15db, 20db, 25db and 30db respectively, and the signals of 12 modulation modes are provided under each type of signal-to-noise ratio, the number of the signals of each modulation mode is 100, each signal is 512 moments, and the data dimensionality of each moment is 2.
(2) L STM is used for extracting the characteristics of signal data with the time sequence length T, the influence degree of the signal data on the characteristics extracted at the current moment is calculated by storing the network state at the previous moment, and the causal relationship of the system is ensured.
(3) Feature H ∈ R extracted according to L STMT×dConstructing a graph-structure representation of a signal and its adjacency matrix A ∈ RT×TFirstly, normalizing the characteristic of each moment to enable the modulus of the characteristic vector to be 1, and then calculating the similarity between every two moments.
(4) The characteristics of the signal in the graph structure representation are calculated by inputting the graph structure representation a of the signal and the signal data X into the GCN model.
(5) And averaging the characteristics of the T moments obtained through GCN, and calculating the confidence of the signal to each modulation mode through a full-connection layer.
The method comprises the steps of receiving a modulation signal with the time sequence length of T through a receiving end, obtaining proper multi-dimensional data by utilizing preprocessing methods such as in-phase orthogonal component decomposition and the like, inputting the processed signal data into a long-term and short-term memory network to extract characteristics, ensuring that the extracted characteristics of the current moment T are related to the signal data of the previous moment, constructing a graph structure representation of the signal by utilizing the characteristics, and finally inputting graph structure information of the signal and the preprocessed multi-dimensional data into a graph convolution network and a full-connection network to obtain a classification result of the signal.
The above-mentioned embodiments are intended to illustrate the technical solutions and advantages of the present invention, and it should be understood that the above-mentioned embodiments are only the most preferred embodiments of the present invention, and are not intended to limit the present invention, and any modifications, additions, equivalents, etc. made within the scope of the principles of the present invention should be included in the scope of the present invention.
Claims (6)
1. A signal classification method based on graph convolution model, characterized in that the signal classification method comprises the following steps:
(1) acquiring a modulation signal with a certain time sequence length, and preprocessing the modulation signal to obtain multi-dimensional modulation signal data;
(2) extracting the characteristics of the multi-dimensional modulation signal data by using a trained long-term and short-term memory network so as to ensure that the characteristics at the current moment are related to the modulation signal data at the previous moment;
(3) constructing graph structure representation information of the signal by using the extracted features;
(4) using graph structure representation information of the signal and the multi-dimensional modulation signal data as input data of a trained graph convolution network, and extracting features of the input data by utilizing the graph convolution network;
(5) and classifying the characteristics of the input data by using the trained full-connection network to obtain a signal classification result.
2. The method for signal classification based on graph convolution model according to claim 1, wherein, in step (1),
and carrying out in-phase and quadrature component decomposition on the modulation signal to realize preprocessing of the modulation signal, wherein the decomposition result is the multi-dimensional modulation signal data.
3. The method for classifying signals based on graph convolution model according to claim 1, wherein in the step (2), the degree of influence of the network state at the previous moment on the extracted features at the current moment is calculated by saving the network state at the previous moment so as to ensure the causal relationship of the system, and the extracting the features of the multi-dimensional modulation signal data by using the trained long-short term memory network specifically comprises:
ft＝σ(Wf·[ht-1,xt]+bf)
it＝σ(Wi·[ht-1,xt]+bi)
ot＝σ(Wo·[ht-1,xt]+bo)
ht＝ot*tanh(Ct)
wherein x istInput data representing time t of signal, ht-1An output characteristic representing the time instant of the signal t-1, [,]representing cascade, sigma (-) and tanh (-) representing activation function to ensure the nonlinearity of network, W and b representing the weight of parameters to be trained, WfAnd bfParameters, f, representing L STM forget gate, respectivelytIs a direction between 0 and 1Quantity, representing the network state C at the last momentt-1For the relevance degree of the extracted features at the current time, 0 represents no relevance, 1 represents the maximum value of the relevance degree, itIndicating the degree of influence of the input data at the current time t on the features, WiAnd biRespectively, the parameters of the input gate are indicated,
initial value, W, representing the state of the network at the present momentCAnd bCParameters respectively representing the state of computation, CtIndicating the state of the network at the present moment, otIndicating the probability of activation of the feature at the current moment of output, WoAnd boRespectively representing the parameters h of the output gatestRepresenting the output characteristics of the signal at the current time.
4. The method of signal classification based on graph convolution model according to claim 1, wherein in step (3), the features extracted according to L STM H ∈ RT×dGraph structure representation information for constructing signal and adjacency matrix A ∈ R thereofT×TFirstly, normalizing the characteristics of each moment, and then calculating the similarity between every two characteristics of the moment, wherein the calculation formula is as follows:
A＝H′H′T
wherein, H'TThe method is characterized in that the method represents the transposition of the H' characteristic, A represents an adjacent matrix of a signal diagram structure, nodes represent each moment of a signal, and continuous edges represent the similarity of the characteristic between the two moments.
5. The method for classifying a signal based on a graph convolution model according to claim 1, wherein in the step (4), the adjacency matrix a of the graph structure representation information of the signal and the multi-dimensional modulation signal data X are used as the input of the graph convolution network, and the feature of the signal represented by the graph structure is calculated by the following calculation formula:
H*＝ReLU(LXW)
wherein L denotes Laplace matrix L ═ I-D-1/2AD-1/2D is a matrix of values of the adjacency matrix A, I is an identity matrix, H*∈RT×dIndicating the resulting node signature, W ∈ Rd×dIs the weight of a trained parameter, RT×dA matrix representing dimension T × d, d being the dimension of the preprocessed modulated signal data, Rd×dA matrix representing d × d dimensions, Re L U (·) represents the activation function.
6. The method for classifying signals based on graph convolution model according to claim 1, wherein in step (5), the features of T moments obtained by the graph convolution network are averaged, and then the confidence of the signal to each modulation mode is obtained by the calculation of the full connection layer, and the calculation formula is:
wherein the content of the first and second substances,
representing features of the output of the graph convolutional network, j represents the index of the feature, L∈ RlRepresents the probability of the signal being predicted for each class, l represents the number of signal modulation schemes, W represents the weight of the fully-connected layer, and softmax (·) represents the activation function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010323837.5A CN111507293A (en) | 2020-04-22 | 2020-04-22 | Signal classification method based on graph convolution model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010323837.5A CN111507293A (en) | 2020-04-22 | 2020-04-22 | Signal classification method based on graph convolution model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111507293A true CN111507293A (en) | 2020-08-07 |
Family
ID=71874597
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010323837.5A Pending CN111507293A (en) | 2020-04-22 | 2020-04-22 | Signal classification method based on graph convolution model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111507293A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113268916A (en) * | 2021-04-07 | 2021-08-17 | 浙江工业大学 | Traffic accident prediction method based on space-time graph convolutional network |
| CN114129175A (en) * | 2021-11-19 | 2022-03-04 | 江苏科技大学 | LSTM and BP based motor imagery electroencephalogram signal classification method |
| CN117807529A (en) * | 2024-02-29 | 2024-04-02 | 南京工业大学 | Modulation mode identification method and system for output signals of signal generator |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109389055A (en) * | 2018-09-21 | 2019-02-26 | 西安电子科技大学 | Video classification methods based on mixing convolution sum attention mechanism |
| CN109657652A (en) * | 2019-01-16 | 2019-04-19 | 平安科技（深圳）有限公司 | A kind of face identification method and device |
| 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 |
| CN110086737A (en) * | 2019-03-13 | 2019-08-02 | 西安电子科技大学 | A kind of recognition methods of the modulation mode of communication signal based on figure neural network |
| CN110351580A (en) * | 2019-07-12 | 2019-10-18 | 四川长虹电器股份有限公司 | TV programme special recommendation method and system based on Non-negative Matrix Factorization |
| CN110598733A (en) * | 2019-08-05 | 2019-12-20 | 南京智谷人工智能研究院有限公司 | Multi-label distance measurement learning method based on interactive modeling |
-
2020
- 2020-04-22 CN CN202010323837.5A patent/CN111507293A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109389055A (en) * | 2018-09-21 | 2019-02-26 | 西安电子科技大学 | Video classification methods based on mixing convolution sum attention mechanism |
| 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 |
| CN109657652A (en) * | 2019-01-16 | 2019-04-19 | 平安科技（深圳）有限公司 | A kind of face identification method and device |
| CN110086737A (en) * | 2019-03-13 | 2019-08-02 | 西安电子科技大学 | A kind of recognition methods of the modulation mode of communication signal based on figure neural network |
| CN110351580A (en) * | 2019-07-12 | 2019-10-18 | 四川长虹电器股份有限公司 | TV programme special recommendation method and system based on Non-negative Matrix Factorization |
| CN110598733A (en) * | 2019-08-05 | 2019-12-20 | 南京智谷人工智能研究院有限公司 | Multi-label distance measurement learning method based on interactive modeling |
Non-Patent Citations (1)
| Title |
|---|
| YABO LIU等: "Modulation Recognition With Graph Convolutional Network", vol. 9, no. 5, pages 625 - 626 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113268916A (en) * | 2021-04-07 | 2021-08-17 | 浙江工业大学 | Traffic accident prediction method based on space-time graph convolutional network |
| CN114129175A (en) * | 2021-11-19 | 2022-03-04 | 江苏科技大学 | LSTM and BP based motor imagery electroencephalogram signal classification method |
| CN117807529A (en) * | 2024-02-29 | 2024-04-02 | 南京工业大学 | Modulation mode identification method and system for output signals of signal generator |
| CN117807529B (en) * | 2024-02-29 | 2024-05-07 | 南京工业大学 | Modulation mode identification method and system for output signals of signal generator |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111507293A (en) | Signal classification method based on graph convolution model | |
| Huang et al. | Automatic modulation classification using compressive convolutional neural network | |
| CN110598677B (en) | Space-time multi-channel deep learning system for automatic modulation recognition | |
| CN110113288B (en) | Design and demodulation method of OFDM demodulator based on machine learning | |
| CN109787696A (en) | The cognitive radio resource allocation methods of case-based reasioning and cooperation Q study | |
| Ali et al. | Algorithm for automatic recognition of PSK and QAM with unique classifier based on features and threshold levels | |
| CN113705787B (en) | Digital modulation signal identification method based on deep collaborative training | |
| CN113095162B (en) | Spectrum sensing method based on semi-supervised deep learning | |
| CN110113084B (en) | Channel prediction method of MIMO closed-loop transmission system | |
| CN113239788A (en) | Mask R-CNN-based wireless communication modulation mode identification method | |
| CN114745234B (en) | Deep learning MIMO system signal detection method and system | |
| Zhang et al. | Heterogeneous deep model fusion for automatic modulation classification | |
| CN115086123B (en) | Modulation identification method and system based on fusion of time-frequency diagram and constellation diagram | |
| CN115118591B (en) | Cluster federation learning method based on alliance game | |
| CN114337883B (en) | CNN collaborative spectrum sensing method and system for covariance matrix Cholesky decomposition | |
| CN113938234B (en) | Low-complexity sparse large-scale MIMO detection method | |
| Li et al. | Deep modulation recognition in an unknown environment | |
| Kumar et al. | 2D-FFT based modulation classification using deep convolution neural network | |
| Hussein et al. | Automatic Modulation Classification: Convolutional Deep Learning Neural Networks Approaches | |
| Zhang et al. | Modulation format recognition based on CNN in satellite communication system | |
| CN114189416B (en) | Digital modulation signal identification method based on consistency regularization | |
| Shahali et al. | Automatic modulation classification based deep learning: A review | |
| CN117354654B (en) | Multi-parameter optimizing method for mixed forming constellation based on learning rate dynamic clipping | |
| CN115496115B (en) | Continuous electromagnetic signal classification method based on vector space separation | |
| Aer et al. | Modulation Recognition of MIMO Systems Based on Dimensional Interactive Lightweight Network. |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200807 |
|
| RJ01 | Rejection of invention patent application after publication |