CN112347871B - Interference signal modulation identification method for communication carrier monitoring system - Google Patents
Interference signal modulation identification method for communication carrier monitoring system Download PDFInfo
- Publication number
- CN112347871B CN112347871B CN202011148118.0A CN202011148118A CN112347871B CN 112347871 B CN112347871 B CN 112347871B CN 202011148118 A CN202011148118 A CN 202011148118A CN 112347871 B CN112347871 B CN 112347871B
- Authority
- CN
- China
- Prior art keywords
- signal
- neural network
- modulation
- connection
- monitoring system
- 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.)
- Active
Links
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
- G06N3/045—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
Abstract
The invention discloses an interference signal modulation identification method for a communication carrier monitoring system, which comprises the following steps: step S1: off-line learning, namely constructing a cascade ResNet neural network classifier for carrying out modulation type classification and identification on the generated two-dimensional signal analysis diagram; a full-connection BP network classifier is constructed to capture the characteristic parameter information of the residual signal r; step S2: and (3) performing online learning, calculating a two-dimensional signal analysis graph and a cyclic spectrum statistical characteristic parameter of the generated r, and then respectively inputting the two-dimensional signal analysis graph and the cyclic spectrum statistical characteristic parameter into the cascade ResNet neural network and the fully-connected BP network to perform modulation type classification identification so as to finally predict a result. The invention relates to a neural network modulation recognition classifier based on the parallel connection of a cascade ResNet neural network and a full-connection BP neural network, which comprises the following steps: the cascade ResNet neural network learns and excavates the two-dimensional signal analysis chart structural feature of interference signal, and the loop spectrum parameter feature of interference signal is excavated in study of full-connection BP neural network, and the two neural networks are combined for processing and judging, so that the modulation recognition rate of the interference signal is effectively improved.
Description
Technical Field
The invention relates to the technical field of satellite communication, in particular to an interference signal modulation identification method for a communication carrier monitoring system.
Background
The satellite communication is very easily influenced by interference signals, and the satellite carrier monitoring system provides support services for interference investigation and positioning by detecting the carrier signals forwarded by the satellite in real time and analyzing the modulation type of the interference signals. Although the time-frequency domain characteristics of the useful signal and the interference signal are usually different from each other, the recombination of the two types of signals has a certain influence on the identification of the respective types, and if a single type of signal modulation identification method is simply combined, when the signal-to-noise ratio is low (for example, SNR =5 dB), the modulation identification of the interference signal is usually mixed or omitted, and it is difficult to achieve a high-precision identification effect.
In addition, the mobile satellite communication channel has fading characteristics such as time-varying, frequency-selective, multipath, etc. Because the satellite carrier monitoring system cannot acquire prior information such as code rate, frequency offset, phase offset and the like of non-cooperative interference signals existing in carrier signals, the satellite carrier monitoring system can only carry out preprocessing such as carrier frequency and frequency offset estimation, down-conversion, frequency offset compensation and down-sampling on received useful signals, and cannot effectively sample, frequency offset and phase offset compensation on the interference signals, so that the success rate of subsequent modulation identification on the interference signals is reduced.
At present, interference signal modulation and identification are mainly carried out on cyclic frequency characteristics extracted from received signals, classification characteristics are constructed by utilizing high-order moments, high-order cumulants, spectral characteristics and the like of the received signals, and classification and identification are carried out by using methods such as a decision tree, a neural network, a support vector machine and the like.
The existing method usually considers additive white gaussian noise channel, flat fading channel or rayleigh fading, such as chinese patent publication No. CN108234370a, publication date: 2018-06-29, which discloses a communication signal modulation mode identification method based on a convolutional neural network, and the invention discloses a modulation mode identification system and method based on a convolutional neural network, and solves the problems of complex feature extraction steps and low identification rate under low signal-to-noise ratio in the prior art. The simple feature in the identification system is constructed by taking the same-direction component and the orthogonal component of a baseband signal as the simple features of the signal and sending the simple features to a convolutional neural network module for identification; the identification method comprises the following implementation steps: modulating a sending signal and performing pulse forming; the sending signal is sent through an additive white Gaussian noise channel after up-conversion; the receiving end carries out pretreatment to obtain the homodromous component r (t) of the analytic signal; the simple structure is characterized in that the homodromous component r (t) and the orthogonal component of the analytic signal are constructed into a two-dimensional matrix; learning and classifying through the characteristics of a convolutional neural network; and sending the modulation mode to a demodulation end to obtain a demodulated signal. The method is not suitable for a frequency selective multipath fading channel in an actual satellite communication system, has certain requirements on prior information such as signal modulation parameters, carrier frequency, code rate and the like, and has lower identification precision under low signal-to-noise ratio.
However, this method is not suitable for the practical application scenario of satellite communication interference signal modulation identification. Therefore, the method for researching the modulation and identification of the interference signal containing unknown code rate, unknown frequency offset and unknown phase offset has practical engineering significance under the condition of the frequency selective multipath fading channel.
Disclosure of Invention
The invention provides an interference signal modulation identification method for a communication carrier monitoring system, which aims to solve the problem of modulation identification of an interference signal containing unknown frequency offset and phase offset under the condition of a multipath fading channel with low signal-to-noise ratio in an actual satellite carrier monitoring system and can effectively improve the modulation identification rate of the interference signal.
In order to achieve the purpose of the invention, the technical scheme is as follows: a method for identifying interference signal modulation for a communication carrier monitoring system, comprising: the modulation identification method comprises the following steps:
step S1: offline learning
S101: calculating a residual signal r of the carrier signal, and performing inverse Fourier transform on a time correlation function of the residual signal r to obtain a two-dimensional signal analysis graph;
s102: constructing a cascade ResNet neural network classifier for carrying out modulation type classification and identification on the generated two-dimensional signal analysis diagram;
s103: extracting characteristic parameters of the residual signal r, wherein the characteristic parameters are mainly high-order statistical characteristic parameters based on the cyclic spectrum;
s104: constructing a full-connection BP network classifier to capture the characteristic parameter information of a residual signal r;
s105: inputting carrier signals for training, adopting cross entropy as a loss function, taking a minimum loss function as a target, then training a cascade ResNet neural network classifier by using a two-dimensional signal analysis diagram, and training a full-connection BP network by using characteristic parameters extracted by a residual signal r;
step S2: on-line learning
S201: deploying the ResNet neural network classifier trained in the step S1 and the full-connection BP network to a monitoring system;
s202: calculating the Error Vector Magnitude (EVM) index of the signal, judging that an interference signal exists in the residual signal r when the EVM is larger than or equal to T%, and entering the step S204; otherwise, judging an interference-free signal;
s203: calculating a two-dimensional signal analysis graph and a cyclic spectrum statistical characteristic parameter of the generated r, and respectively inputting the two-dimensional signal analysis graph and the cyclic spectrum statistical characteristic parameter into a cascade ResNet neural network and a full-connection BP network for modulation type classification identification;
s204: and finally, adding the output results of the cascade ResNet neural network and the full-connection BP neural network, and carrying out normalization processing to obtain a fused final prediction result.
The invention has the following beneficial effects:
the invention relates to a neural network modulation recognition classifier based on parallel connection of a cascade ResNet neural network and a full-connection BP neural network, which comprises the following steps: the cascade ResNet neural network learns and excavates the two-dimensional signal analysis chart structural feature of interfering signal, and the loop spectrum parameter feature of full-connection BP neural network learning excavation interfering signal, handles the judgement through combining these two kinds of neural networks, and the effectual modulation recognition rate that improves interfering signal.
Drawings
Fig. 1 is a flowchart of the offline learning flow described in embodiment 1.
Fig. 2 is a schematic diagram of a residual unit described in embodiment 1.
FIG. 3 is a flowchart of online identification described in embodiment 1.
Fig. 4 is a comparison of success rates of interference modulation identification of different modulation identification algorithms under different signal to interference ratios.
Fig. 5 shows the success rate of the modulation identification of different interference modulation type signals by the parallel ResNet + BP network.
Detailed Description
The invention is described in detail below with reference to the drawings and the detailed description.
Example 1
The modulation method of the interference signal according to the present embodiment includes 10 types of MPSK (BPSK, QPSK, 8 PSK), MQAM (16 QAM, 64 QAM), MFSK (2 FSK, 4 FSK), AM, FM, and PM.
An interference signal modulation identification method for a communication carrier monitoring system, which can be used in the communication carrier monitoring system including but not limited to satellite communication, microwave communication and the like, the modulation identification method comprises the following steps:
step S1: offline learning
Preprocessing the satellite carrier interfered signals which are modulated, classified and marked in the early stage, and then training a neural network based on the parallel connection of a cascade ResNet network and a full-connection BP network, wherein the process is shown in figure 1:
before step S101, preprocessing the carrier signal used for training and having interference needs to be performed, where the preprocessing includes data normalization and interpolation processing, and is used to determine an optimal measurement position through a minimum point of a module value variance of a sampling point so as to improve accuracy when selecting an optimal sampling point, and finally compensate for a frequency offset and a phase offset existing in the carrier signal.
S101: calculating residual signal r of the preprocessed carrier signal according to the following calculation formula
r＝Z-Y
Wherein Z is a signal to be detected obtained by preprocessing, and Y is a generated ideal reference signal.
And performing inverse Fourier transform on the time correlation function of the residual signal r to obtain a two-dimensional signal analysis graph, wherein the two-dimensional signal analysis graph can effectively reflect the texture characteristics of signals of different modulation types and is insensitive to the influence of additive noise.
The time-dependent function of r is expressed as follows:
R(m,k)＝r(m+k)r * (m-k) (1)
in the formula, m and k are independent variables of the time correlation function.
Inverse fourier transform of the time-dependent function of the residual signal r:
in the formula, W n ＝e (-2πi)/n I is an imaginary unit, n is the number of discrete Fourier transform points, and j and k are independent variables of a two-dimensional discrete Fourier transform function.
S102: constructing a cascade ResNet neural network classifier for carrying out modulation type classification and identification on the generated two-dimensional signal analysis diagram;
the cascade ResNet neural network comprises a plurality of residual error units which are connected in series and are finally followed by a full connection layer and an output layer classifier, the output layer classifier adopts softmax, and the optimizer of the cascade ResNet neural network adopts Adam.
Each residual error unit includes an active layer, one or more convolutional layers, and a jump connection unit connected to the convolutional layers, as shown in fig. 2:
the active layer described in this embodiment: and (2) filtering the input less than or equal to 0 by using a ReLU function, thereby screening effective information characteristics from a very sparse two-dimensional signal analysis graph, and further realizing and maintaining sparsity to better mine correlation characteristics.
The one or more convolutional layers: consisting of one or more convolution kernels of dimensions 3 x 3. By selecting a convolution kernel with a proper dimensionality, a good detail information aggregation effect can be obtained without introducing a pooling layer, so that the problems of inaccurate information feature extraction, increased training errors and the like caused by introducing the pooling layer are solved.
The jump connection unit: the jump connection unit is used for directly connecting the input and the output of the residual error unit, improves the liquidity of input signal image information, ensures that a ResNet neural network extracts features without obstruction, and solves the problems of gradient disappearance and deep network degradation.
S103: extracting characteristic parameters of the residual signal r, wherein the characteristic parameters are mainly high-order statistical characteristic parameters based on a cyclic spectrum, and the characteristic parameters comprise: the high-order cyclic spectrum characteristic parameter, the high-order cyclic moment characteristic parameter and the high-order cyclic cumulant characteristic parameter. The cyclic spectrum can inhibit interference of stationary noise, frequency offset and phase offset and strengthen reflected characteristics.
S104: and constructing a full-connection BP network classifier to capture the characteristic parameter information of the residual signal r. The full-connection BP network classifier comprises a 6-layer hidden layer, an input layer, an output layer and an activation layer;
wherein the neuron number of the input layer is the number of the extracted characteristic parameters of the circulation spectrum,
the neuron number of the output layer is the modulation type number (namely 10 types)
The activation function between the input layer and the hidden layer adopts a tansig function;
the activation function between the hidden layer and the output layer adopts a linear purelin function;
the neural network optimization algorithm of the full-connection BP network classifier adopts an L-M algorithm.
S105: inputting carrier signal data Z for training, adopting cross entropy as a loss function, aiming at minimizing the loss function, training a cascade ResNet neural network by using a two-dimensional signal analysis diagram of a residual signal r, and training a full-connection BP neural network by using characteristic parameters extracted from the residual signal r.
Step S2: online identification
S201: the ResNet neural network classifier trained in step S1 and the fully-connected BP network are deployed to a satellite carrier monitoring system for performing interference modulation identification on a carrier signal received in real time, and the specific flow is shown in fig. 3.
In order to improve the accuracy when selecting the optimal sampling point, after step S201 and before step S202, the received carrier signal needs to be preprocessed, where the preprocessing includes data normalization and interpolation processing, and is used to determine an optimal measurement position through a minimum point of a module value variance of the sampling point, and finally compensate for frequency offset and phase offset.
S202: calculating a signal Error Vector Magnitude (EVM) index, which can be calculated by equation (3):
when the EVM is larger than or equal to 10%, judging that an interference signal exists in the residual signal r, and entering step S204; otherwise, judging the non-interference signal.
S203: and (3) calculating to generate a two-dimensional signal analysis graph of r by using the formula (1) and the formula (2), calculating a cyclic spectrum statistical characteristic parameter, and then respectively inputting the cyclic spectrum statistical characteristic parameter into the cascade ResNet neural network and the full-connection BP network to perform modulation type classification and identification.
S204: and finally, adding the output results of the cascade ResNet neural network and the full-connection BP neural network, and carrying out normalization processing to obtain a fused final prediction result.
In order to further verify the interference signal modulation identification method of the present invention, the embodiment further performs a related experiment, as shown in fig. 4, to compare the modulation identification success rates of the interference signals under different signal to interference ratios by using different modulation identification algorithms. The system comprises a TC (temperature coefficient), a SVM (support vector machine), a BP (back propagation) and a ResNet, wherein the TC, the SVM, the BP and the ResNet are respectively a classic tree-shaped recognition classifier based on signal statistical characteristics, a support vector machine recognition classifier, a full-connected BP neural network classifier and a ResNet classifier, and the CLDNN is a classic deep learning recognition classifier. As can be seen from fig. 4, the proposed parallel ResNet + BP network identification classifier has a higher success rate of interference modulation identification. The reason is that the parallel ResNet + BP network not only learns the cycle spectrum statistical characteristics of the signals through the BP network, but also learns the two-dimensional analysis characteristics of the signals in a time-frequency transform domain through the ResNet.
As shown in fig. 5, the success ratio comparison of modulation identification of interference signals of different modulation types is performed by using parallel ResNet + BP networks under different signal-to-interference ratios. It can be known from fig. 5 that the parallel ResNet + BP network can stably detect different interference modulation types.
It should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (7)
1. A method for identifying interference signal modulation for a communication carrier monitoring system, comprising: the modulation identification method comprises the following steps:
step S1: offline learning
S101: calculating a residual signal r of the carrier signal, and performing inverse Fourier transform on a time correlation function of the residual signal r to obtain a two-dimensional signal analysis graph;
s102: constructing a cascade ResNet neural network classifier for carrying out modulation type classification and identification on the generated two-dimensional signal analysis diagram;
s103: extracting characteristic parameters of the residual signal r, wherein the characteristic parameters are mainly high-order statistical characteristic parameters based on the cyclic spectrum;
s104: constructing a full-connection BP network classifier to capture the characteristic parameter information of a residual signal r;
s105: inputting carrier signals for training, adopting cross entropy as a loss function, taking a minimum loss function as a target, then training a cascade ResNet neural network classifier by using a two-dimensional signal analysis diagram, and training a full-connection BP network by using characteristic parameters extracted by a residual signal r;
step S2: on-line learning
S201: deploying the ResNet neural network classifier trained in the step S1 and the full-connection BP network to a monitoring system;
s202: calculating the Error Vector Magnitude (EVM) index of the signal, judging that an interference signal exists in the residual signal r when the EVM is larger than or equal to T%, and entering the step S204; otherwise, judging an interference-free signal;
s203: calculating a two-dimensional signal analysis graph and a cyclic spectrum statistical characteristic parameter of the generated r, and then respectively inputting the two-dimensional signal analysis graph and the cyclic spectrum statistical characteristic parameter into a cascade ResNet neural network and a full-connection BP network for modulation type classification identification;
s204: finally, adding the output results of the cascade ResNet neural network and the full-connection BP neural network, and carrying out normalization processing to obtain a fused final prediction result;
step S101, the formula for calculating the residual signal is as follows:
r＝Z-Y
wherein Z is a signal to be detected obtained by preprocessing, and Y is a generated ideal reference signal;
step S101, performing inverse fourier transform on the time-dependent function of the residual signal r to obtain a two-dimensional signal analysis chart, which is specifically as follows:
the time-related function expression of the residual signal r is as follows:
R(m,k)＝r(m+k)r * (m-k) (1)
in the formula, m and k are independent variables of a time correlation function;
inverse fourier transform of the time-dependent function of the residual signal r:
in the formula, W n ＝e (-2πi)/n I is an imaginary part unit, n is the number of discrete Fourier transform points, and j and k are independent variables of a two-dimensional discrete Fourier transform function;
step S102, the cascade ResNet neural network classifier comprises a plurality of residual error units which are connected in series and are finally followed by a full connection layer and an output layer classifier;
each residual unit comprises an active layer, one or more convolutional layers and a jump connection unit connected behind the convolutional layers;
the active layer is used for filtering out the input less than or equal to 0;
the jump connection unit: for connecting the input and output of the residual unit.
2. The interfering signal modulation identifying method for a communication carrier monitoring system according to claim 1, characterized in that: before step S101, the interfering carrier signal used for training needs to be preprocessed.
3. The interfering signal modulation identifying method for a communication carrier monitoring system according to claim 2, characterized in that: the pretreatment comprises the following specific steps: and normalizing and interpolating the carrier signal data for training to improve the precision when selecting the optimal sampling point, determining the optimal measurement position through the minimum value point of the module value variance of the sampling point, and finally compensating the frequency offset and the phase offset.
4. The interfering signal modulation identifying method for a communication carrier monitoring system according to claim 1, characterized in that: the high-order statistical characteristic parameters comprise high-order cyclic spectrum characteristic parameters, high-order cyclic moment characteristic parameters and high-order cyclic cumulant characteristic parameters.
5. The jamming signal modulation identification method for a communication carrier monitoring system according to claim 4, characterized in that: the full-connection BP network classifier comprises a hidden layer, an input layer, an output layer and an activation layer which are 6 layers; wherein the activation function between the input layer and the hidden layer adopts a tansig function; the activation function between the hidden layer and the output layer adopts a linear purelin function; the neural network optimization algorithm adopts an L-M algorithm;
the number of neurons of the input layer is the number of the extracted characteristic parameters of the circulation spectrum
The neuron number of the output layer is a modulation type number.
6. The jamming signal modulation identification method for a communication carrier monitoring system according to claim 5, characterized in that: step S202, the error vector magnitude EVM index has the following calculation formula
Wherein, Z is a signal to be detected obtained by preprocessing, and Y is a generated ideal reference signal.
7. The jamming signal modulation identification method for a communication carrier monitoring system according to claim 6, characterized in that: after step S201, after step S202, the received carrier signal needs to be preprocessed;
the pretreatment comprises the following specific steps: and normalizing and interpolating the carrier signal data for training to improve the precision when selecting the optimal sampling point, determining the optimal measurement position through the minimum value point of the module value variance of the sampling point, and finally compensating the frequency offset and the phase offset.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011148118.0A CN112347871B (en) | 2020-10-23 | 2020-10-23 | Interference signal modulation identification method for communication carrier monitoring system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011148118.0A CN112347871B (en) | 2020-10-23 | 2020-10-23 | Interference signal modulation identification method for communication carrier monitoring system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112347871A CN112347871A (en) | 2021-02-09 |
| CN112347871B true CN112347871B (en) | 2023-04-07 |
Family
ID=74360025
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011148118.0A Active CN112347871B (en) | 2020-10-23 | 2020-10-23 | Interference signal modulation identification method for communication carrier monitoring system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112347871B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113205012A (en) * | 2021-04-19 | 2021-08-03 | 云南电网有限责任公司玉溪供电局 | Signal identification method for circuit breaker fault detection |
| CN114553339B (en) * | 2022-03-13 | 2023-08-11 | 中铁第一勘察设计院集团有限公司 | Single-channel mixed communication signal interference detection method based on high-order cyclic accumulation |
| CN114785649B (en) * | 2022-04-21 | 2023-11-14 | 北京鹏鹄物宇科技发展有限公司 | Satellite communication signal identification method based on multiport neural network |
| CN114826850B (en) * | 2022-04-21 | 2022-11-25 | 中国人民解放军国防科技大学 | Modulation identification method, device and equipment based on time-frequency diagram and deep learning |
| CN114866172B (en) * | 2022-07-05 | 2022-09-20 | 中国人民解放军国防科技大学 | Interference identification method and device based on inverse residual deep neural network |
| CN116482720B (en) * | 2023-06-26 | 2023-09-05 | 山东科技大学 | GNSS deception jamming detection method based on machine learning theory |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2815794B1 (en) * | 2000-10-24 | 2003-02-21 | Telediffusion De France Tdf | DEVICES FOR IDENTIFYING A INTERFERING SIGNAL OF A RADIOELECTRIC SIGNAL WITH A SUBSTANTIALLY CONSTANT MODULE |
| US10419253B2 (en) * | 2016-12-21 | 2019-09-17 | Zte Corporation | Blind classification of modulation scheme of an interfering signal |
| CN107124381B (en) * | 2017-03-27 | 2020-05-22 | 华南理工大学 | Automatic identification method for digital communication signal modulation mode |
| CN108234370B (en) * | 2017-12-22 | 2020-12-15 | 西安电子科技大学 | Communication signal modulation mode identification method based on convolutional neural network |
| CN111050315B (en) * | 2019-11-27 | 2021-04-13 | 北京邮电大学 | Wireless transmitter identification method based on multi-core two-way network |
| CN111092836B (en) * | 2019-12-13 | 2022-05-17 | 中国人民解放军空军工程大学 | Signal modulation mode identification method and device |
| CN111259798A (en) * | 2020-01-16 | 2020-06-09 | 西安电子科技大学 | Modulation signal identification method based on deep learning |
-
2020
- 2020-10-23 CN CN202011148118.0A patent/CN112347871B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN112347871A (en) | 2021-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112347871B (en) | Interference signal modulation identification method for communication carrier monitoring system | |
| CN108234370B (en) | Communication signal modulation mode identification method based on convolutional neural network | |
| CN110855591B (en) | QAM and PSK signal intra-class modulation classification method based on convolutional neural network structure | |
| Güner et al. | Automatic digital modulation classification using extreme learning machine with local binary pattern histogram features | |
| CN109657604B (en) | Satellite amplitude-phase signal identification and demodulation method and device based on cyclic neural network | |
| Fehske et al. | A new approach to signal classification using spectral correlation and neural networks | |
| CN109543643B (en) | Carrier signal detection method based on one-dimensional full convolution neural network | |
| CN112818891A (en) | Intelligent identification method for communication interference signal type | |
| CN106357574A (en) | BPSK (Binary Phase Shift Keying)/QPSK (Quadrature Phase Shift Keying) signal modulation blind identification method based on order statistic | |
| CN116866129A (en) | Wireless communication signal detection method | |
| WO2021088465A1 (en) | Fast modulation recognition method using multilayer perceptron, and employing data fusion of multiple distribution tests | |
| CN112613443A (en) | Robustness communication radiation source intelligent identification method based on deep learning | |
| Ali et al. | Automatic modulation recognition of DVB-S2X standard-specific with an APSK-based neural network classifier | |
| CN116628566A (en) | Communication signal modulation classification method based on aggregated residual transformation network | |
| CN114422311A (en) | Signal modulation identification method and system combining deep neural network and expert prior characteristics | |
| Ghasemzadeh et al. | GGCNN: an efficiency-maximizing gated graph convolutional neural network architecture for automatic modulation identification | |
| Bouchenak et al. | A semi-supervised modulation identification in MIMO systems: A deep learning strategy | |
| Ali et al. | Modulation format identification using supervised learning and high-dimensional features | |
| Fontes et al. | Automatic modulation classification using information theoretic similarity measures | |
| Hazza et al. | Robustness of digitally modulated signal features against variation in HF noise model | |
| CN113242201B (en) | Wireless signal enhanced demodulation method and system based on generation classification network | |
| CN116304553A (en) | Interference type identification method based on deep learning under multi-modulation system | |
| Jariwala et al. | Automatic modulation classification: a novel convolutional neural network based approach | |
| CN115913849A (en) | Electromagnetic signal identification method based on one-dimensional complex value residual error network | |
| Huang et al. | Automatic recognition of communication signal modulation based on the multiple-parallel complex convolutional neural 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |