CN117118789A - MIMO-OFDM signal blind modulation identification method and system based on CLDNN neural network - Google Patents

MIMO-OFDM signal blind modulation identification method and system based on CLDNN neural network Download PDF

Info

Publication number
CN117118789A
CN117118789A CN202311274302.3A CN202311274302A CN117118789A CN 117118789 A CN117118789 A CN 117118789A CN 202311274302 A CN202311274302 A CN 202311274302A CN 117118789 A CN117118789 A CN 117118789A
Authority
CN
China
Prior art keywords
signal
neural network
mimo
cldnn
signals
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202311274302.3A
Other languages
Chinese (zh)
Other versions
CN117118789B (en
Inventor
刘洋
冯仁旭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangnan University
Original Assignee
Jiangnan University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiangnan University filed Critical Jiangnan University
Priority to CN202311274302.3A priority Critical patent/CN117118789B/en
Publication of CN117118789A publication Critical patent/CN117118789A/en
Application granted granted Critical
Publication of CN117118789B publication Critical patent/CN117118789B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • 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
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
    • H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
    • H04B7/0413—MIMO systems
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L27/00—Modulated-carrier systems
    • H04L27/26—Systems using multi-frequency codes
    • H04L27/2601—Multicarrier modulation systems
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Radio Transmission System (AREA)

Abstract

The application provides a MIMO-OFDM signal blind modulation recognition method and system based on a CLDNN neural network, and relates to the technical field of wireless communication; estimating the number of transmitting antennas of the MIMO system by using a minimum description length criterion; performing PCA whitening treatment on the received signal data, and recovering a source signal by using a JADE algorithm; generating a training data set; constructing a CLDNN convolutional neural network model, completing training of the convolutional neural network, and obtaining a trained CLDNN neural network model; and obtaining various modulation mode probabilities from the multi-antenna receiving signals through a trained CLDNN neural network model, and obtaining the final modulation mode type according to a confidence decision theory. The application realizes the modulation recognition of the subcarrier modulation signals of the MIMO-OFDM system under low signal to noise ratio. The method is high in accuracy and good in effectiveness.

Description

MIMO-OFDM signal blind modulation identification method and system based on CLDNN neural network
Technical Field
The application relates to the technical field of wireless communication, in particular to a MIMO-OFDM signal blind modulation identification method and system based on a CLDNN neural network.
Background
In non-cooperative communication, the channel environment is complex, the signal is greatly interfered, and the accuracy of signal identification influences the quality of subsequent processing of the signal, so that signal modulation identification (AMR) has very important research significance, and the arrival of the 5G age, the Multiple Input Multiple Output (MIMO) and orthogonal frequency division multiplexing (MIMO-OFDM) technology has been widely applied to modern wireless communication systems. Whereas in the last decade, many AMC algorithms were introduced into single-input single-output single-carrier (SISO) communication systems. The modulation mode of the MIMO system is more and more complex, so that the modulation recognition task is more difficult, and the research on the modulation recognition of the MIMO system has important significance.
Conventional signal modulation recognition methods can be basically classified into two types: an identification method based on maximum likelihood theory and an identification method based on expert experience feature extraction. The method based on the maximum likelihood ratio theory is guided by Bayesian theory, likelihood functions are designed through priori information of signals to obtain likelihood ratio values of received signals, and then signals with the maximum likelihood ratio values are output as classification results. The method converts the problem of modulation classification into the multiple composite hypothesis test problem by deducing the discrimination threshold of the signal, and has high recognition accuracy. However, because it is difficult to obtain prior information of non-cooperative signals, research in recent years is mainly focused on methods based on feature extraction, but based on feature extraction, obvious features which can distinguish different modulation signals, such as "high-order accumulation", instantaneous features, joint features and the like, are selected manually, and then a classifier which is matched with the features best is selected to realize classification, such as classification methods like decision trees and the like. However, in the face of huge information quantity, the method for manually selecting the features is heavy in task, and the defects of part of information loss, classifier performance and the like of feature extraction result in poor generalization capability of the method, and the method is not suitable for identifying a large number of signals in space.
In recent years, with the rapid development of the artificial intelligence field, AMR can extract features and complete classification by means of a large number of strong characterizations of neural networks, has low requirements on expert knowledge, and has become a research hotspot in various fields at present. In addition to the fact that the O' Shea et al first verify that the convolutional neural network (Convolutional Neural Network, abbreviated as CNN) uses IQ data as a data storage format, perform effective training and verification on the model, and disclose an RML2016 data set, since the modulated signals are time-series signal data, there is a strong time-series correlation between the data, a learner uses a Long Short-Term Memory (abbreviated as LSTM) to acquire time-series related information in the signal data to identify a modulation mode of the signals. In addition, the improved CNN network such as Residual network (ResNet), CNN and deep neural network DNN are combined to improve the accuracy of AMR in the deep learning field. The AMR task is completed through other angles by inputting a cyclic spectrogram, a vector graph, a time-frequency graph and other characterization graphs of the original data into an algorithm in the CNN.
The above algorithms are all aimed at single carrier signals, further research is needed for subcarrier modulation identification of the MIMO-OFDM system, and a plurality of communication modes exist nowadays, so that a plurality of modulation identification modes exist in one communication system. Therefore, the technical problem to be solved by the technicians in the field is to solve the identification of the non-cooperative modulation mode in the MIMO-OFDM system.
Disclosure of Invention
Therefore, the embodiment of the application provides a MIMO-OFDM signal blind modulation identification method and system based on a CLDNN neural network, which are used for solving the problem that the modulation identification of the traditional single-input single-output modulation mode in the prior art cannot meet the identification of a MIMO signal system.
In order to solve the above problems, an embodiment of the present application provides a method for identifying blind modulation of MIMO-OFDM signals based on CLDNN neural network, the method comprising:
s1: based on the design characteristics of the MIMO system, designing an MIMO-OFDM signal emission model;
s2: estimating the number of transmitting antennas of the MIMO system by utilizing a minimum description length criterion based on the MIMO-OFDM signal transmitting model;
s3: performing PCA whitening treatment on the received signal data, recovering a source signal by using a JADE algorithm, calculating a cyclic spectrum of the recovered signal, and extracting a cross section with the cyclic spectrum frequency of zero;
s4: preprocessing the recovered signal, generating a training data set based on the preprocessed signal and a section with zero cyclic spectrum frequency, and dividing the training data set into a training sample set and a test sample set;
s5: constructing a CLDNN convolutional neural network model, inputting a training sample set into the CLDNN convolutional neural network model for training, and completing training of the convolutional neural network when the maximum iteration number set by the neural network is reached, so as to obtain a trained CLDNN neural network model;
s6: and obtaining various modulation mode probabilities of the multi-antenna receiving signals through a trained CLDNN neural network model, estimating the number of transmitting antennas of the MIMO system according to a minimum description length criterion, selecting the number of proper receiving antennas, deducing the number of signal types based on different noise influence degrees received by the receiving signals of different antennas and the number of transmitting antennas, and obtaining the final modulation mode type according to a confidence decision theory.
Preferably, in step S1, the method for designing the MIMO-OFDM signal transmission model based on the design features of the MIMO system specifically includes:
s11: in the MIMO-OFDM system, the number of the antennas of a transmitting end and the number of the antennas of a receiving end are N t ，N r And satisfy N t ＜N r The method comprises the steps of carrying out a first treatment on the surface of the Based on a flat fading channel model theory, carrying out channel modeling on the transmitting signals, establishing a flat fading channel matrix, and constructing a channel state, wherein the model of the transmitting signals is expressed as follows:
in the method, in the process of the application,represents the nth t The first symbol, T, transmitted by the kth subchannel on the root antenna s Represents one frame of OFDM signal time, T g And T u Respectively represents the cyclic prefix time and the useful time in the frame signal, and the three satisfy T s ＝T g +T u G (t) is a matrix pulse function, and N is a constant;
s12: the transmitting signal is converted into N through serial-parallel conversion, constellation map mapping and MIMO coding t The OFDM signals;
s13: each antenna receiving signal of the multi-antenna receiving end is superposition of each transmission channel, nr receiving signals are counted, the transmitting signals reach the receiving end after being transmitted by the channels, the receiving signals are transmitting signals plus noise, and the receiving signals are expressed as:
wherein,respectively represent the nth r Gaussian white noise on kth subchannel on root antenna, +.>Representing a flat fading channel.
Preferably, in step S2, based on the MIMO-OFDM signal transmission model, the method for estimating the number of transmission antennas written by MIMO using the minimum description length criterion specifically includes:
s21: constructing a covariance matrix of a received signal, and carrying out eigenvalue decomposition on the covariance matrix: r is R (k) ＝U∑U H Where U is a unitary matrix of eigenvectors, Σ=diag (λ) 1 ,λ 2 ,... a diagonal matrix of eigenvalues of the covariance matrix, the eigenvalues are arranged in descending order;
s22: covariance matrix spectral decomposition according to the principle of spectral representation, the covariance matrix R is represented by decomposed eigenvectors and eigenvalues (k) Will covariance matrix R (k) Using a parameter mode to represent, under the condition that the length N of the received data sample is large enough, observing a data vector joint probability distribution function according to a central theorem;
s23: the degree of freedom K of parameters of the objective function is solved, and the degree of freedom K is the number of parameters which can be adjusted freely and is expressed as follows:
wherein N is r The number of receiving antennas is represented, and k represents the number of parameters which can be freely adjusted;
s24: solving a minimum description length criterion objective function, and estimating the number of transmitting antennas of the MIMO system by using a blind estimation algorithm, wherein the estimation criterion of the blind estimation algorithm is as follows:
in the method, in the process of the application,representing the estimated number of transmit antennas, N r For the number of reception antennas, AIC (k) represents a red-pool information amount criterion, argmin () represents a minimum value.
Preferably, in step S3, the method for performing PCA whitening processing on the received signal data specifically includes:
signal X e R n×m Where n is the data dimension and m is the number of samples, zero-equalizing each row of signal X to obtain
Calculating covariance matrix of the zero-averaged data:
performing eigenvalue decomposition on the covariance matrix: Σ=uΛu T ；
Rotating the data: x is X rotate ＝U T X；
The data on each dimension is divided by the standard deviation of that dimension:
wherein lambda is i The eigenvalues are represented, U represents a unitary matrix composed of eigenvectors, and Λ represents an eigenvector.
Preferably, in step S3, the method for recovering the source signal by using the JADE algorithm is as follows:
firstly, a fourth-order cumulant matrix C of a whitening signal is obtained; then singular value decomposition is carried out on the matrix C, and the front N with the maximum modulus is taken t The characteristic value phi i And its corresponding feature matrix U i Writing it as a set of matrices; finally, the matrix set is subjected to joint approximate diagonalization to obtain a separation matrix X, and the restored transmission signal isWhere q (k) represents the whitened signal.
Preferably, in step S3, the method of calculating the cyclic spectrum of the recovered signal is:
the autocorrelation function is first calculated for signal s (t):
R S (t,τ)＝E[s(t+τ)s * (t)]
where τ represents the time delay, s * (t) is conjugated with s (t);
then to the autocorrelation function R s And (t, r) performing Fourier transformation to obtain a cyclic autocorrelation function:
wherein f is the signal frequency;
when the noise is Gaussian white noise, the signal isThe autocorrelation function is:
in sigma n Representing noise variance, n (t) representing noise;
therefore, the gaussian white noise signal cycle spectrum exists as follows:
where α represents the cycle frequency.
Preferably, in step S4, the restoring signal is preprocessed, a training data set is generated based on the preprocessed signal and a section with zero cyclic spectrum frequency, and the method for dividing the training data set into a training sample set and a test sample set specifically includes:
s41: storing the recovery signal in an I/Q data type;
s42: obtaining a three-dimensional circulation spectrogram according to a circulation spectrogram calculation formula;
s43: adopting 6 kinds of signals, namely { BPSK, QPSK,8PSK,16QAM,32QAM,128QAM } total, of binary phase shift keying modulation signals, quadrature phase shift keying modulation signals, eight phase shift keying modulation signals, hexadecimal quadrature amplitude modulation signals and 128 binary quadrature amplitude modulation signals under the MIMO-OFDM system, selecting a signal to noise ratio of-10 dB to 10dB, sampling once every 2dB, and constructing 1024 x 3 data with the I/Q sequence and the characteristics of a circular spectrum section as input to generate a training data set;
s44: dividing a training set and a test set, randomly extracting 80% of samples from each modulation type of the data sample set respectively, combining the samples into the training sample set, and combining the rest 20% of samples into the test sample set.
Preferably, in step S5, the method for constructing the CLDNN convolutional neural network model specifically includes:
setting CLDNN classification neural network parameters and maximum iteration times;
constructing a CLDNN convolutional neural network model, wherein the structure of the CLDNN convolutional neural network comprises 4 CNN layers, 2 LSTM layers, 3 DNN layers, an average pooling layer and an activation layer which are softmax;
a loss function of the model is set, an optimization algorithm is set as cross entropy, the optimization algorithm selects an error back propagation algorithm, and an activation function is set as a modified linear unit activation function.
The embodiment of the application also provides a MIMO-OFDM signal blind modulation recognition system based on the CLDNN neural network, which is used for realizing the MIMO-OFDM signal blind modulation recognition method based on the CLDNN neural network, and comprises the following steps:
the MIMO-OFDM signal transmission model design module is used for designing a MIMO-OFDM signal transmission model based on the design characteristics of the MIMO system;
the transmitting antenna number estimation module is used for estimating the transmitting antenna number of the MIMO system by utilizing the minimum description length criterion based on the MIMO-OFDM signal transmitting model;
the data processing module is used for carrying out PCA whitening processing on the received signal data, recovering a source signal by using a JADE algorithm, calculating a cyclic spectrum of the recovered signal, and extracting a section with zero cyclic spectrum frequency;
the training data set generation module is used for preprocessing the recovery signal, generating a training data set based on the preprocessed signal and a section with zero cyclic spectrum frequency, and dividing the training data set into a training sample set and a test sample set;
the neural network training module is used for constructing a CLDNN convolutional neural network model, inputting a training sample set into the CLDNN convolutional neural network model for training, and completing training of the convolutional neural network when the maximum iteration number set by the neural network is reached, so as to obtain a trained CLDNN neural network model;
the debugging and identifying module is used for acquiring various modulation mode probabilities of the multi-antenna receiving signals through the trained CLDNN neural network model, estimating the number of transmitting antennas of the MIMO system according to the minimum description length criterion, selecting the proper number of receiving antennas, deducing the number of signal types based on the different noise influence degrees received by the receiving signals of different antennas and the number of transmitting antennas, and acquiring the final modulation mode type according to the confidence decision theory.
The embodiment of the application also provides an electronic device, which comprises a processor, a memory and a bus system, wherein the processor and the memory are connected through the bus system, the memory is used for storing instructions, and the processor is used for executing the instructions stored by the memory so as to realize the MIMO-OFDM signal blind modulation identification method based on the CLDNN neural network.
From the above technical scheme, the application has the following advantages:
the application provides a MIMO-OFDM signal blind modulation recognition method and system based on a CLDNN neural network, which are implemented by designing an MIMO-OFDM signal transmission model; estimating the number of transmitting antennas of the MIMO system by using a minimum description length criterion, and selecting the number of suitable receiving antennas; PCA whitening processing is carried out on the received signal data, so that redundancy of the input data is reduced, and a JADE algorithm is utilized to recover a source signal; generating a training data set based on the I/Q sequence and the cross section with the cyclic spectrum frequency of zero; constructing a CLDNN convolutional neural network model, completing training of the convolutional neural network, and obtaining a trained CLDNN neural network model; and obtaining various modulation mode probabilities from the multi-antenna receiving signals through a trained CLDNN neural network model, and obtaining the final modulation mode type according to a confidence decision theory. The method realizes the modulation identification of the subcarrier modulation signals of the MIMO-OFDM system under the low signal to noise ratio, has higher accuracy and has good effectiveness.
Drawings
For a clearer description of embodiments of the application or of solutions in the prior art, reference will be made to the accompanying drawings, which are intended to be used in the examples, for a clearer understanding of the characteristics and advantages of the application, by way of illustration and not to be interpreted as limiting the application in any way, and from which, without any inventive effort, a person skilled in the art can obtain other figures. Wherein:
fig. 1 is a flowchart of a CLDNN neural network-based MIMO-OFDM signal blind modulation recognition method provided in accordance with an embodiment;
FIG. 2 is a schematic diagram of simulation experiment results;
fig. 3 is a block diagram of a CLDNN neural network-based MIMO-OFDM signal blind modulation recognition system according to an embodiment.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Example 1
As shown in fig. 1, an embodiment of the present application provides a CLDNN neural network-based MIMO-OFDM signal blind modulation identification method, where the method includes:
s1: based on the design characteristics of the MIMO system, designing an MIMO-OFDM signal emission model;
s2: estimating the number of transmitting antennas of the MIMO system by utilizing a minimum description length criterion based on the MIMO-OFDM signal transmitting model;
s3: performing PCA whitening treatment on the received signal data, recovering a source signal by using a JADE algorithm, calculating a cyclic spectrum of the recovered signal, and extracting a cross section with the cyclic spectrum frequency of zero;
s4: preprocessing the recovered signal, generating a training data set based on the preprocessed signal and a section with zero cyclic spectrum frequency, and dividing the training data set into a training sample set and a test sample set;
s5: constructing a CLDNN convolutional neural network model, inputting a training sample set into the CLDNN convolutional neural network model for training, and completing training of the convolutional neural network when the maximum iteration number set by the neural network is reached, so as to obtain a trained CLDNN neural network model;
s6: and obtaining various modulation mode probabilities of the multi-antenna receiving signals through a trained CLDNN neural network model, estimating the number of transmitting antennas of the MIMO system according to a minimum description length criterion, selecting the number of proper receiving antennas, deducing the number of signal types based on different noise influence degrees received by the receiving signals of different antennas and the number of transmitting antennas, and obtaining the final modulation mode type according to a confidence decision theory.
The MIMO-OFDM signal transmitting model is designed; estimating the number of transmitting antennas of the MIMO system by using a minimum description length criterion, and selecting the number of suitable receiving antennas; PCA whitening processing is carried out on the received signal data, so that redundancy of the input data is reduced, and a JADE algorithm is utilized to recover a source signal; generating a training data set based on the I/Q sequence and the cross section with the cyclic spectrum frequency of zero; constructing a CLDNN convolutional neural network model, completing training of the convolutional neural network, and obtaining a trained CLDNN neural network model; and obtaining various modulation mode probabilities from the multi-antenna receiving signals through a trained CLDNN neural network model, and obtaining the final modulation mode type according to a confidence decision theory. The method realizes the modulation identification of the subcarrier modulation signals of the MIMO-OFDM system under the low signal to noise ratio, has higher accuracy and has good effectiveness.
In this embodiment, in step S1, a MIMO-OFDM signal transmission model is designed based on the design features of the MIMO system, and specifically includes the following steps:
s11: in the MIMO-OFDM system, the number of the antennas of a transmitting end and the number of the antennas of a receiving end are N t ，N r And satisfy N t ＜N r The method comprises the steps of carrying out a first treatment on the surface of the Based on the theory of flat fading channel model, carrying out channel modeling on the transmitting signal, establishing a flat fading channel matrix (the application assumes that the MIMO channel is flat fading and time-invariant channel, and can also be other types of channel.) and constructing a channel state, wherein the model of the transmitting signal is expressed as:
in the method, in the process of the application,represents the nth t The first symbol, T, transmitted by the kth subchannel on the root antenna s Represents one frame of OFDM signal time, T g And T u Respectively represents the cyclic prefix time and the useful time in the frame signal, and the three satisfy T s ＝T g +T u G (t) is a matrix pulse function and N is a constant.
S12: the transmitting signal is converted into N through serial-parallel conversion, constellation map mapping and MIMO coding t And OFDM signals.
S13: each antenna receiving signal of the multi-antenna receiving end is superposition of each transmission channel, nr receiving signals are counted, the transmitting signals reach the receiving end after being transmitted by the channels, the receiving signals are transmitting signals plus noise, and the receiving signals are expressed as:
wherein,respectively represent the nth r Gaussian white noise on kth subchannel on root antenna, +.>Representing a flat fading channel.
In this embodiment, in step S2, based on the MIMO-OFDM signal transmission model, the number of transmitting antennas of the MIMO system is estimated using the minimum description length criterion, and specifically includes the following steps:
s21: constructing a covariance matrix of a received signal, and carrying out eigenvalue decomposition on the covariance matrix: r is R (k) ＝UΣU H Where U is a unitary matrix composed of eigenvectors, Σ=diag (λ 1 ,λ 2 ,... a diagonal matrix of eigenvalues of the covariance matrix, the eigenvalues are arranged in descending order.
S22: covariance matrix spectral decomposition is represented by decomposed eigenvectors and eigenvalues according to the principle of spectral representationCovariance matrix R (k) Will covariance matrix R (k) Expressed in parametric mode, the data vector joint probability distribution function is observed according to the central theorem, assuming that the received data sample length N is sufficiently large.
S23: the degree of freedom K of parameters of the objective function is solved, and the degree of freedom K is the number of parameters which can be adjusted freely and is expressed as follows:
wherein N is r The number of receiving antennas is represented, and k represents the number of parameters that can be freely adjusted.
S24: solving a minimum description length criterion objective function, and estimating the number of transmitting antennas of the MIMO system by using a blind estimation algorithm, wherein the estimation criterion of the blind estimation algorithm is as follows:
in the method, in the process of the application,representing the estimated number of transmit antennas, N r For the number of reception antennas, AIC (k) represents a red-pool information amount criterion, argmin () represents a minimum value.
In this embodiment, in step S3, PCA whitening is performed on the received signal data, the source signal is recovered by using the JADE algorithm, the cyclic spectrum of the recovered signal is calculated, and a cross section with the cyclic spectrum frequency of zero is extracted.
Specifically, PCA whitening of received signal data, which is a linear transformation for decorrelating the source signal, is an important preprocessing procedure for reducing redundancy of the input data such that the whitened input data has the following properties: the correlation between the features is eliminated; the variance of all features is 1.
The method for performing PCA whitening processing on the received signal data specifically comprises the following steps:
signal X e R n×m Where n is the data dimension and m is the number of samples, zero-equalizing each row of the signal X (i.e., subtracting the mean value of the row, specifically: averaging the features of all samples in this dimension, then subtracting the mean value from all values of the row), to obtain
Calculating covariance matrix of the zero-averaged data:it is noted that here the covariance matrix is calculated using the idea of maximum likelihood estimation, so the coefficient is +.>This is a progressive unbiased estimate of the covariance matrix, the true unbiased estimate being +.>
Performing eigenvalue decomposition on the covariance matrix: Σ=uΛu T 。
Rotate the data (projection onto the principal component axis, decorrelation achieved): x is X rotate ＝U T X。
The data on each dimension is divided by the standard deviation of that dimension: wherein lambda is i The eigenvalues are represented, U represents a unitary matrix composed of eigenvectors, and Λ represents an eigenvector.
The method for recovering the source signal by using the JADE algorithm comprises the following steps:
firstly, a fourth-order cumulant matrix C of a whitening signal is obtained; then singular value decomposition is carried out on the matrix C, and the maximum modulus is takenFront N t The characteristic value phi i And its corresponding feature matrix U i Writing it as a set of matrices; finally, the matrix set is subjected to joint approximate diagonalization to obtain a separation matrix X, and the restored transmission signal isWhere q (k) represents the whitened signal.
The method for calculating the cyclic spectrum of the recovered signal comprises the following steps:
the autocorrelation function is first calculated for signal s (t):
R S (t,τ)＝E[s(t+τ)s * (t)]
where τ represents the time delay, s * (t) is conjugated with s (t).
Then to the autocorrelation function R s And (t, r) performing Fourier transformation to obtain a cyclic autocorrelation function:
where f is the signal frequency.
When the noise is Gaussian white noise, the signal isThe autocorrelation function is:
in sigma n Representing the noise variance, n (t) represents the noise.
Therefore, the gaussian white noise signal cycle spectrum exists as follows:
where α represents the cycle frequency.
In this embodiment, in step S4, the recovery signal is preprocessed, a training data set is generated based on the preprocessed signal and a section with zero cyclic spectrum frequency, and the training data set is divided into a training sample set and a test sample set, which specifically includes the following steps:
s41: the recovery signal is stored in an I/Q data type. The I/Q sequence can be utilized to nondestructively express the characteristics of signal phase, amplitude and the like, so that the method has stronger characteristic expression capability.
S42: and obtaining a three-dimensional circulation spectrogram according to the calculation formula of the circulation spectrogram.
Because the calculated circulation spectrogram contains a large amount of data and redundant information, if the neural network is directly used for training, the training time is too long. Therefore, data dimension reduction is generally required before the neural network training is performed, so that the data can be used more easily, noise is removed to a certain extent, and meanwhile, calculation overhead is reduced. The application adopts the section with zero spectrum frequency as the full characteristic vector as the characteristic.
S43: adopting 6 kinds of signals, namely { BPSK, QPSK,8PSK,16QAM,32QAM,128QAM } total, of binary phase shift keying modulation signals, quadrature phase shift keying modulation signals, eight phase shift keying modulation signals, hexadecimal quadrature amplitude modulation signals and 128 binary quadrature amplitude modulation signals under the MIMO-OFDM system, selecting a signal to noise ratio of-10 dB to 10dB, sampling once every 2dB, and constructing 1024 x 3 dimension data with an I/Q sequence and a cyclic spectrum tangential surface characteristic (a section with zero cyclic spectrum frequency) as input to generate a training data set;
s44: dividing a training set and a test set, randomly extracting 80% of samples from each modulation type of the data sample set respectively, combining the samples into the training sample set, and combining the rest 20% of samples into the test sample set.
In the embodiment, in step S5, a CLDNN convolutional neural network model is constructed, and a training sample set is input into the CLDNN convolutional neural network model for training, and when the maximum iteration number set by the neural network is reached, training of the convolutional neural network is completed, so as to obtain a trained CLDNN neural network model.
Specifically, the method for constructing the CLDNN convolutional neural network model specifically comprises the following steps:
and setting the parameters of the CLDNN classification neural network and the maximum iteration times. The maximum iteration number of the method is set to 100 steps; the learning rate was set to 0.001 and the batch size was set to 128.
Constructing a CLDNN convolutional neural network model, wherein the structure of the CLDNN classifying neural network comprises 4 CNN layers, 2 LSTM layers, 3 DNN layers, an average pooling layer and an activation layer which are softmax; further, the CLDNN classification neural network is composed of CNN layer 1, pooling layer 1, CNN layer 2, pooling layer 2, CNN layer 3, pooling layer 3, CNN layer 4, pooling layer 4, LSTM layer 1, LSTM layer 2, DNN layer 1, DNN layer 2, and classification layer. Wherein the parameters are set as follows, and the number of the convolution layer cores is set to 64, 128, 256 and 256 respectively; the LSTM layer is respectively set to 128 cores and 128 cores; cores are set to 128, 32; each pooling layer is average pooling; softmax is the type of activation output.
A loss function of the model is set, an optimization algorithm is set as cross entropy, the optimization algorithm selects an error back propagation algorithm, and an activation function is set as a modified linear unit activation function.
In addition, after the trained CLDNN neural network model is obtained, the test sample set is input into the trained CLDNN neural network model, and the recognition result is obtained. And comparing the identification result with the real category of the test set, and counting the identification accuracy.
In the embodiment, in step S6, the receiving signals of multiple antennas are obtained through a trained CLDNN neural network model to obtain various modulation mode probabilities, the number of transmitting antennas of the MIMO system is estimated according to the minimum description length criterion, and the number of suitable receiving antennas is selected. Based on the weight, the product and the summation preset strategy, the feature information of each mode can be kept complete. The number of signal types is deduced based on the different degrees of influence of noise received by the received signals of different antennas and the number of transmitting antennas. Because the noise influence degrees received by the receiving signals of different antennas are different, the recovery effect of each path of signal is different, the probability of the signal contributes to the final result differently, and the final modulation mode type is obtained according to the confidence decision theory.
The effect of the application can be further illustrated by the following simulations:
1. simulation conditions:
the simulation experiment of the application completes the generation of the radio signal and the simulation training of the CLDNN network decision classifier on a Pytorch=2.0.1 operation platform under an Intel (R) I7-13700CPU 3.5GHz,RTXA5000,Ubuntu21.04LTS system.
2. Simulation experiment contents:
the 6 three-dimensional cyclic spectrum images used in the simulation experiment are divided into a BPSK signal three-dimensional cyclic spectrum, a QPSK signal three-dimensional cyclic spectrum, an 8PSK signal three-dimensional cyclic spectrum, a 16QAM three-dimensional cyclic spectrum, a 32QAM signal three-dimensional cyclic spectrum and a 128QAM three-dimensional cyclic spectrum according to a modulation mode.
Simulation experiments are carried out by adopting the MIMO-OFDM signal blind modulation identification method based on the CLDNN neural network. And will not be described in detail herein.
3. Simulation experiment results:
the simulation experiment result of the present application is shown in fig. 2. The horizontal axis in fig. 2 represents the number of iterations, and the vertical axis corresponds to the loss function value for each iteration. In the training process of the CLDNN convolutional neural network model, the loss function value of each training result is counted, and the smaller the loss function value is, the better the training effect of the model is represented. As can be seen from fig. 2, the loss function value decreases with the increase of the iteration number and finally converges to be stable, which indicates that the training effect of the simulation experiment increases with the increase of the training number. And inputting the test sample into a trained CLDNN convolutional neural network model, and obtaining the recognition accuracy of the simulation experiment to be 95%. The simulation experiment can show that the method can complete different types of modulation recognition tasks aiming at the modulation recognition of the radio signal, and is effective and feasible.
Example two
As shown in fig. 3, the present application provides a CLDNN neural network-based MIMO-OFDM signal blind modulation recognition system, which includes:
the MIMO-OFDM signal transmission model design module 10 is used for designing a MIMO-OFDM signal transmission model based on the design characteristics of the MIMO system;
a transmitting antenna number estimating module 20, configured to estimate the number of transmitting antennas of the MIMO system based on the MIMO-OFDM signal transmitting model by using a minimum description length criterion;
the data processing module 30 is configured to perform PCA whitening on the received signal data, recover the source signal using a JADE algorithm, calculate a cyclic spectrum of the recovered signal, and extract a cross section with a cyclic spectrum frequency of zero;
a training data set generating module 40, configured to pre-process the recovered signal, generate a training data set based on the pre-processed signal and a section with a cyclic spectrum frequency of zero, and divide the training data set into a training sample set and a test sample set;
the neural network training module 50 is configured to construct a CLDNN convolutional neural network model, input a training sample set to the CLDNN convolutional neural network model for training, and complete training of the convolutional neural network when the maximum iteration number set by the neural network is reached, thereby obtaining a trained CLDNN neural network model;
the debug identifying module 60 is configured to obtain probabilities of various modulation modes from the multi-antenna received signal through the trained CLDNN neural network model, estimate the number of transmitting antennas of the MIMO system according to the minimum description length criterion, select a suitable number of receiving antennas, infer the number of signal types based on different noise influence degrees received by the receiving signals of different antennas and the number of transmitting antennas, and obtain the final modulation mode type according to the confidence decision theory.
The embodiment of the CLDNN-based MIMO-OFDM signal blind modulation recognition system is applicable to implementing the foregoing CLDNN-based MIMO-OFDM signal blind modulation recognition method, so that the specific implementation of the CLDNN-based MIMO-OFDM signal blind modulation recognition system can be seen in the foregoing embodiment of the CLDNN-based MIMO-OFDM signal blind modulation recognition method, for example, the MIMO-OFDM signal transmission model design module 10, the transmitting antenna number estimation module 20, the data processing module 30, the training data set generation module 40, the neural network training module 50, and the debug recognition module 60 are respectively used to implement steps S1, S2, S3, S4, S5, and S6 in the foregoing CLDNN-based MIMO-OFDM signal blind modulation recognition method, which will not be repeated herein for avoiding redundancy.
Example III
The embodiment of the application also provides an electronic device, which comprises a processor, a memory and a bus system, wherein the processor and the memory are connected through the bus system, the memory is used for storing instructions, and the processor is used for executing the instructions stored by the memory so as to realize the MIMO-OFDM signal blind modulation identification method based on the CLDNN neural network.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations and modifications of the present application will be apparent to those of ordinary skill in the art in light of the foregoing description. It is not necessary here nor is it exhaustive of all embodiments. And obvious variations or modifications thereof are contemplated as falling within the scope of the present application.

Claims (10)

1. The MIMO-OFDM signal blind modulation identification method based on the CLDNN neural network is characterized by comprising the following steps of:
s1: based on the design characteristics of the MIMO system, designing an MIMO-OFDM signal emission model;
s2: estimating the number of transmitting antennas of the MIMO system by utilizing a minimum description length criterion based on the MIMO-OFDM signal transmitting model;
s3: performing PCA whitening treatment on the received signal data, recovering a source signal by using a JADE algorithm, calculating a cyclic spectrum of the recovered signal, and extracting a cross section with the cyclic spectrum frequency of zero;
s4: preprocessing the recovered signal, generating a training data set based on the preprocessed signal and a section with zero cyclic spectrum frequency, and dividing the training data set into a training sample set and a test sample set;
s5: constructing a CLDNN convolutional neural network model, inputting a training sample set into the CLDNN convolutional neural network model for training, and completing training of the convolutional neural network when the maximum iteration number set by the neural network is reached, so as to obtain a trained CLDNN neural network model;
s6: and obtaining various modulation mode probabilities of the multi-antenna receiving signals through a trained CLDNN neural network model, estimating the number of transmitting antennas of the MIMO system according to a minimum description length criterion, selecting the number of proper receiving antennas, deducing the number of signal types based on different noise influence degrees received by the receiving signals of different antennas and the number of transmitting antennas, and obtaining the final modulation mode type according to a confidence decision theory.
2. The method for identifying blind modulation of MIMO-OFDM signals based on CLDNN neural network according to claim 1, wherein in step S1, the method for designing a MIMO-OFDM signal transmission model based on design features of a MIMO system specifically comprises:
s11: in the MIMO-OFDM system, the number of the antennas of a transmitting end and the number of the antennas of a receiving end are N t ，N r And satisfy N t ＜N r The method comprises the steps of carrying out a first treatment on the surface of the Based on a flat fading channel model theory, carrying out channel modeling on the transmitting signals, establishing a flat fading channel matrix, and constructing a channel state, wherein the model of the transmitting signals is expressed as follows:
in the method, in the process of the application,represents the nth t The first symbol, T, transmitted by the kth subchannel on the root antenna s Represents one frame of OFDM signal time, T g And T u Respectively represents the cyclic prefix time and the useful time in the frame signal, and the three satisfy T s ＝T g +T u G (t) is a matrix pulse function, and N is a constant;
s12: hair brushThe radio signal is converted by serial-parallel conversion, and the constellation map mapping and MIMO coding are changed into N t The OFDM signals;
s13: each antenna receiving signal of the multi-antenna receiving end is superposition of each transmission channel, nr receiving signals are counted, the transmitting signals reach the receiving end after being transmitted by the channels, the receiving signals are transmitting signals plus noise, and the receiving signals are expressed as:
wherein,respectively represent the nth r Gaussian white noise on kth subchannel on root antenna, +.>Representing a flat fading channel.
3. The blind modulation recognition method of MIMO-OFDM signal based on CLDNN neural network according to claim 1, wherein in step S2, based on the MIMO-OFDM signal transmission model, the method for estimating the number of transmission antennas of MIMO writing using the minimum description length criterion specifically comprises:
s21: constructing a covariance matrix of a received signal, and carrying out eigenvalue decomposition on the covariance matrix: r is R (k) ＝U∑U H Where U is a unitary matrix of eigenvectors, Σ=diag (λ) 1 ,λ 2 ,... a diagonal matrix of eigenvalues of the covariance matrix, the eigenvalues are arranged in descending order;
s22: covariance matrix spectral decomposition according to the principle of spectral representation, the covariance matrix R is represented by decomposed eigenvectors and eigenvalues (k) Will covariance matrix R (k) Using a parameter mode to represent, under the condition that the length N of the received data sample is large enough, observing a data vector joint probability distribution function according to a central theorem;
s23: the degree of freedom K of parameters of the objective function is solved, and the degree of freedom K is the number of parameters which can be adjusted freely and is expressed as follows:
wherein N is r The number of receiving antennas is represented, and k represents the number of parameters which can be freely adjusted;
s24: solving a minimum description length criterion objective function, and estimating the number of transmitting antennas of the MIMO system by using a blind estimation algorithm, wherein the estimation criterion of the blind estimation algorithm is as follows:
in the method, in the process of the application,representing the estimated number of transmit antennas, N r For the number of reception antennas, AIC (k) represents a red-pool information amount criterion, argmin () represents a minimum value.
4. The CLDNN neural network-based MIMO-OFDM signal blind modulation recognition method according to claim 1, wherein in step S3, the method for performing PCA whitening processing on the received signal data specifically comprises:
signal X e R n×m Where n is the data dimension and m is the number of samples, zero-equalizing each row of signal X to obtain
Calculating covariance matrix of the zero-averaged data:
performing eigenvalue decomposition on the covariance matrix: sigma (sigma)＝UΛU T ；
Rotating the data: x is X rotate ＝U T X；
The data on each dimension is divided by the standard deviation of that dimension:
wherein lambda is i The eigenvalues are represented, U represents a unitary matrix composed of eigenvectors, and Λ represents an eigenvector.
5. The CLDNN neural network-based MIMO-OFDM signal blind modulation recognition method according to claim 1, wherein in step S3, the method for recovering the source signal by using the JADE algorithm is as follows:
firstly, a fourth-order cumulant matrix C of a whitening signal is obtained; then singular value decomposition is carried out on the matrix C, and the front N with the maximum modulus is taken t The characteristic value phi i And its corresponding feature matrix U i Writing it as a set of matrices; finally, the matrix set is subjected to joint approximate diagonalization to obtain a separation matrix X, and the restored transmission signal isWhere q (k) represents the whitened signal.
6. The CLDNN neural network-based MIMO-OFDM signal blind modulation recognition method according to claim 1, wherein in step S3, the method for calculating the cyclic spectrum of the recovered signal is as follows:
the autocorrelation function is first calculated for signal s (t):
R S (t,τ)＝E[s(t+τ)s * (t)]
where τ represents the time delay, s * (t) is conjugated with s (t);
then to the autocorrelation function R s (t, r) performing Fourier transform to obtain a cycleLoop autocorrelation function:
wherein f is the signal frequency;
when the noise is Gaussian white noise, the signal isThe autocorrelation function is:
where σ represents the noise variance and n (t) represents the noise;
therefore, the gaussian white noise signal cycle spectrum exists as follows:
where α represents the cycle frequency.
7. The method for identifying blind modulation of MIMO-OFDM signals based on CLDNN neural network according to claim 1, wherein in step S4, the recovery signal is preprocessed, a training data set is generated based on the preprocessed signal and a section with zero cyclic spectrum frequency, and the training data set is divided into a training sample set and a test sample set, which specifically includes:
s41: storing the recovery signal in an I/Q data type;
s42: obtaining a three-dimensional circulation spectrogram according to a circulation spectrogram calculation formula;
s43: adopting 6 kinds of signals, namely { BPSK, QPSK,8PSK,16QAM,32QAM,128QAM } total, of binary phase shift keying modulation signals, quadrature phase shift keying modulation signals, eight phase shift keying modulation signals, hexadecimal quadrature amplitude modulation signals and 128 binary quadrature amplitude modulation signals under the MIMO-OFDM system, selecting a signal to noise ratio of-10 dB to 10dB, sampling once every 2dB, and constructing 1024 x 3 data with the I/Q sequence and the characteristics of a circular spectrum section as input to generate a training data set;
s44: dividing a training set and a test set, randomly extracting 80% of samples from each modulation type of the data sample set respectively, combining the samples into the training sample set, and combining the rest 20% of samples into the test sample set.
8. The method for identifying blind modulation of MIMO-OFDM signals based on CLDNN neural network according to claim 1, wherein in step S5, the method for constructing CLDNN convolutional neural network model specifically comprises:
setting CLDNN classification neural network parameters and maximum iteration times;
constructing a CLDNN convolutional neural network model, wherein the structure of the CLDNN convolutional neural network comprises 4 CNN layers, 2 LSTM layers, 3 DNN layers, an average pooling layer and an activation layer which are softmax;
a loss function of the model is set, an optimization algorithm is set as cross entropy, the optimization algorithm selects an error back propagation algorithm, and an activation function is set as a modified linear unit activation function.
9. A CLDNN neural network-based MIMO-OFDM signal blind modulation recognition system, which is configured to implement the CLDNN neural network-based MIMO-OFDM signal blind modulation recognition method according to any one of claims 1 to 8, the system comprising:
the MIMO-OFDM signal transmission model design module is used for designing a MIMO-OFDM signal transmission model based on the design characteristics of the MIMO system;
the transmitting antenna number estimation module is used for estimating the transmitting antenna number of the MIMO system by utilizing the minimum description length criterion based on the MIMO-OFDM signal transmitting model;
the data processing module is used for carrying out PCA whitening processing on the received signal data, recovering a source signal by using a JADE algorithm, calculating a cyclic spectrum of the recovered signal, and extracting a section with zero cyclic spectrum frequency;
the training data set generation module is used for preprocessing the recovery signal, generating a training data set based on the preprocessed signal and a section with zero cyclic spectrum frequency, and dividing the training data set into a training sample set and a test sample set;
the neural network training module is used for constructing a CLDNN convolutional neural network model, inputting a training sample set into the CLDNN convolutional neural network model for training, and completing training of the convolutional neural network when the maximum iteration number set by the neural network is reached, so as to obtain a trained CLDNN neural network model;
the debugging and identifying module is used for acquiring various modulation mode probabilities of the multi-antenna receiving signals through the trained CLDNN neural network model, estimating the number of transmitting antennas of the MIMO system according to the minimum description length criterion, selecting the proper number of receiving antennas, deducing the number of signal types based on the different noise influence degrees received by the receiving signals of different antennas and the number of transmitting antennas, and acquiring the final modulation mode type according to the confidence decision theory.
10. An electronic device, characterized in that the electronic device comprises a processor, a memory and a bus system, the processor and the memory are connected through the bus system, the memory is used for storing instructions, and the processor is used for executing the instructions stored by the memory so as to realize the MIMO-OFDM signal blind modulation identification method based on the CLDNN neural network according to any one of claims 1 to 8.
CN202311274302.3A 2023-09-28 2023-09-28 MIMO-OFDM signal blind modulation identification method and system based on CLDNN neural network Active CN117118789B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311274302.3A CN117118789B (en) 2023-09-28 2023-09-28 MIMO-OFDM signal blind modulation identification method and system based on CLDNN neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311274302.3A CN117118789B (en) 2023-09-28 2023-09-28 MIMO-OFDM signal blind modulation identification method and system based on CLDNN neural network

Publications (2)

Publication Number Publication Date
CN117118789A true CN117118789A (en) 2023-11-24
CN117118789B CN117118789B (en) 2024-07-19

Family

ID=88796630

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311274302.3A Active CN117118789B (en) 2023-09-28 2023-09-28 MIMO-OFDM signal blind modulation identification method and system based on CLDNN neural network

Country Status (1)

Country Link
CN (1) CN117118789B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106713190A (en) * 2017-01-05 2017-05-24 西安电子科技大学 MIMO (Multiple Input Multiple Output) transmitting antenna number blind estimation algorithm based on random matrix theory and feature threshold estimation
CN107948114A (en) * 2017-11-15 2018-04-20 桂林电子科技大学 A kind of MIMO blind source signal separations system and signal separating method
CN114764577A (en) * 2022-04-29 2022-07-19 天津大学 Lightweight modulation recognition model based on deep neural network and method thereof
CN116800575A (en) * 2022-03-18 2023-09-22 重庆邮电大学 MIMO-OFDM signal subcarrier blind modulation identification

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106713190A (en) * 2017-01-05 2017-05-24 西安电子科技大学 MIMO (Multiple Input Multiple Output) transmitting antenna number blind estimation algorithm based on random matrix theory and feature threshold estimation
CN107948114A (en) * 2017-11-15 2018-04-20 桂林电子科技大学 A kind of MIMO blind source signal separations system and signal separating method
CN116800575A (en) * 2022-03-18 2023-09-22 重庆邮电大学 MIMO-OFDM signal subcarrier blind modulation identification
CN114764577A (en) * 2022-04-29 2022-07-19 天津大学 Lightweight modulation recognition model based on deep neural network and method thereof

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
刘镇: "基于深度学习的调制方式识别算法研究", 中国优秀硕士学位论文全文数据库》, 16 September 2022 (2022-09-16), pages 6 - 30 *
张和发等: "一种适用于微弱信号盲提取的白化方法", 《电子学报》, 15 June 2011 (2011-06-15), pages 1 - 4 *

Also Published As

Publication number Publication date
CN117118789B (en) 2024-07-19

Similar Documents

Publication Publication Date Title
Ali et al. Automatic modulation classification of digital modulation signals with stacked autoencoders
Ozpoyraz et al. Deep learning-aided 6G wireless networks: A comprehensive survey of revolutionary PHY architectures
CN112637093B (en) Signal detection method based on model-driven deep learning
CN109802905B (en) CNN convolutional neural network-based digital signal automatic modulation identification method
CN108234370B (en) Communication signal modulation mode identification method based on convolutional neural network
CN112418014B (en) Modulated signal identification method based on wavelet transformation and convolution long-term and short-term memory neural network
CN113014524B (en) Digital signal modulation identification method based on deep learning
Abdel‐Moneim et al. A survey of traditional and advanced automatic modulation classification techniques, challenges, and some novel trends
Smith et al. A communication channel density estimating generative adversarial network
Ali et al. Automatic modulation classification using different neural network and PCA combinations
Zhang et al. Pilot-assisted MIMO-V-OFDM systems: Compressed sensing and deep learning approaches
CN114745233B (en) Joint channel estimation method and device based on pilot frequency design
Zheng et al. Deep learning based trainable approximate message passing for massive MIMO detection
Zhang et al. Efficient residual shrinkage CNN denoiser design for intelligent signal processing: Modulation recognition, detection, and decoding
Zhang et al. High-order convolutional attention networks for automatic modulation classification in communication
Ali et al. Modulation format identification using supervised learning and high-dimensional features
CN114124168A (en) MIMO-NOMA system signal detection method and system based on deep learning
CN108566227B (en) Multi-user detection method
Baumgartner et al. Neural network based data estimation for unique word OFDM
CN117118789B (en) MIMO-OFDM signal blind modulation identification method and system based on CLDNN neural network
CN107395540B (en) Modulation signal identification method based on LMD approximate entropy, high accumulation product quantity and SVM
CN116094885A (en) Signal modulation mode identification method based on One2ThreeNet
CN109523023A (en) A kind of deep learning network and system for subsurface communication Modulation Identification
CN113242201B (en) Wireless signal enhanced demodulation method and system based on generation classification network
Yıldırım et al. Deep receiver design for multi-carrier waveforms using cnns

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