CN114866162A - Signal data enhancement method and system and identification method and system of communication radiation source - Google Patents
Signal data enhancement method and system and identification method and system of communication radiation source Download PDFInfo
- Publication number
- CN114866162A CN114866162A CN202210807165.4A CN202210807165A CN114866162A CN 114866162 A CN114866162 A CN 114866162A CN 202210807165 A CN202210807165 A CN 202210807165A CN 114866162 A CN114866162 A CN 114866162A
- Authority
- CN
- China
- Prior art keywords
- signal
- radiation source
- data enhancement
- communication radiation
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/16—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Medical Informatics (AREA)
- Mathematical Physics (AREA)
- Electromagnetism (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
The invention provides a signal data enhancement method for communication radiation source identification, which comprises the following steps: generating more than 0 and less than 2 piNA rotation angle; acquiring a sample signal; rotating the phase of each sample signal separatelyNA rotation angle ofNA replica signal. The invention also provides a signal data enhancement system for identifying the communication radiation source, and a method and a system for identifying the communication radiation source. The invention can realize the enhancement of sample signal data while keeping the individual fingerprint characteristics of the radiation source, expand the number of training samples and improve the quality of the training samplesAnd (4) the individual identification accuracy of the communication radiation source under the condition of limited training data.
Description
Technical Field
The invention belongs to the technical field of radiation source identification, and particularly relates to a signal data enhancement method and system and a communication radiation source identification method and system.
Background
Communication radiation source individual identification refers to distinguishing the radiation source individuals emitting signals by analyzing received radio frequency signals. Due to inevitable precision errors of devices inside the radiation source equipment in the production process, even radio frequency signals with the same content emitted by the radiation source individuals of the same model have certain slight differences. This subtle difference is also known as a "radiation source fingerprint" or "radio frequency fingerprint". The individual identification of the communication radiation source has wide application in the fields of wireless network security, frequency spectrum management and control, military information reconnaissance and the like.
In the communication radiation source individual identification, the method based on deep learning is the method with higher accuracy at present. The method uses a signal sample of training data to train a deep neural network, and inputs a test sample into the neural network in a test stage to obtain a recognition result. Because the quantity of parameters of the neural network is large, the deep learning method usually needs training data of a certain scale for training. Due to the randomness of the initial phase of the communication signal, when the number of training samples is small, the initial phase of the training samples cannot cover all possible values, so that overfitting is easily caused, namely the identification accuracy of the neural network on the test sample is low.
Disclosure of Invention
One of the objectives of the present invention is to provide a signal data enhancement method for communication radiation source identification, which can enhance the sample signal data while maintaining the fingerprint characteristics of the individual radiation sources.
It is a further object of the present invention to provide a signal data enhancement system for communicating radiation source identification.
It is a further object of the present invention to provide a method for identifying a communication radiation source.
The fourth object of the present invention is to provide an identification system for a communication radiation source.
In order to achieve one of the purposes, the invention adopts the following technical scheme:
a signal data enhancement method for communication radiation source identification, the signal data enhancement method comprising the steps of:
generating more than 0 and less than 2 piNA rotation angle;
acquiring a sample signal;
rotating the phase of each sample signal separatelyNA rotation angle ofNA replica signal.
Further, the generation is greater than 0 and less than 2 piNThe rotation angle further comprises:
according to a uniform distribution to generate a value greater than 0 and less than 2 piNAn angle of rotation.
Further, the replica signal is:
x ′ =x*e iθ ；
wherein the content of the first and second substances,x ′ is a replica signal;xis a sample signal;θis the angle of rotation,θ~U(0，2π)。
further, the rotation angle is a random rotation angle; or at equally spaced rotational angles.
In order to achieve the second purpose, the invention adopts the following technical scheme:
a signal data enhancement system for communication radiation source identification, the signal data enhancement system comprising:
a rotation angle generation module for generating a value greater than 0 and less than 2 piNA rotation angle;
the acquisition module is used for acquiring a sample signal;
a rotation module for rotating the phase of each sample signal respectivelyNA rotation angle ofNA replica signal.
Further, the replica signal is:
x ′ =x*e iθ ；
wherein the content of the first and second substances,x ′ is a replica signal;xis a sample signal;θis the angle of rotation,θ~U(0，2π)。
further, the rotation angle is a random rotation angle; or at equally spaced rotational angles.
In order to achieve the third purpose, the invention adopts the following technical scheme:
a method of identifying a source of communication radiation, the method comprising the steps of:
the method comprises the steps of firstly, obtaining a first sample signal of each communication radiation source individual and a second sample signal of a communication radiation source individual to be identified;
step two, according to the signal data enhancement method, data enhancement is respectively carried out on each first sample signal and each second sample signal to obtain a first copy signal set and a second copy signal set;
inputting the first copy signal set into a neural network for training;
inputting the second copy signal set into the trained neural network to perform communication radiation source individual identification, and obtaining the probability that each second copy signal belongs to each communication radiation source individual;
and fifthly, calculating the average probability values of all the second replica signals respectively belonging to all the communication radiation source individuals, and selecting the communication radiation source individual corresponding to the maximum average probability value as a final identification result.
In order to achieve the fourth purpose, the invention adopts the following technical scheme:
an identification system for a communication radiation source, the identification system comprising:
the acquisition module is used for acquiring a first sample signal of each communication radiation source individual and a second sample signal of the communication radiation source individual to be identified;
a data enhancement module, configured to perform data enhancement on each of the first sample signal and the second sample signal respectively according to the signal data enhancement method described above, so as to obtain a first replica signal set and a second replica signal set;
the training module is used for inputting the first copy signal set into a neural network for training;
the communication radiation source individual identification module is used for inputting the second copy signal set into the trained neural network to carry out communication radiation source individual identification, and the probability that each second copy signal belongs to each communication radiation source individual is obtained;
and the selecting module is used for calculating the average probability values of all the second copy signals respectively belonging to all the communication radiation source individuals and selecting the communication radiation source individual corresponding to the maximum average probability value as the final identification result.
The invention has the beneficial effects that:
according to the invention, through phase rotation of the sample signal, while the fingerprint characteristics of the individual radiation source are maintained, the enhancement of sample signal data is realized, the number of training samples is expanded, overfitting caused by less sample signals during deep nerve training is avoided, the identification accuracy of a neural network on a test sample is improved, the identification result of a plurality of phase copies of the same sample signal is integrated, and the identification accuracy of the individual communication radiation source is improved.
Drawings
FIG. 1 is a schematic flow chart of a signal data enhancement method for individual identification of a communication radiation source according to the present invention;
FIG. 2 is a flow chart of the individual identification method of the communication radiation source according to the present invention;
FIG. 3 is a schematic diagram of a neural network architecture;
FIG. 4 is a graph illustrating recognition accuracy versus number of copies using data enhancement.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings.
Data enhancement of communication radiation source individual signals is to ensure that individual characteristics of radiation sources do not change, and the nonlinear characteristic of a power amplifier is an important reason for the signal to have individual radiation source difference. Recording sample signalxBelong to the firstkA radiation source, corresponding toThe desired baseband signal issThen, the nonlinear characteristic of the power amplifier can be represented by taylor series as:
wherein the content of the first and second substances,γ k [m]for the sample signal to belong tokThe power amplification non-linear characteristic of each radiation source,m=1,2,…,
，Min order of the non-linearity of the power amplifier,s * [l]is as followslAn ideal baseband signals[l]The complex number of the conjugate of (a),l=1,2,…,L，Lis the sample signal length. Since the band pass filters of the individual radiation sources filter the even-order signals, the taylor series of the above formula only contains the odd-order.
It is explained that phase rotating the signal samples does not change the power amplifier non-linear characteristics in the signal samples. Sampling the sample signalxIs rotated by any angleθObtaining:
wherein the content of the first and second substances,s ’ [l]=s[l]e iθ ，s * [l]is composed ofs[l]Conjugation of (1).
As can be seen from the above equation, the sample signal is rotatedxEquivalent to the phase rotation of an ideal baseband signal in the same power amplifier nonlinear parameterγ k [m]The following was obtained. Therefore, sample signal data enhancement by phase rotation can maintain individual characteristics of the radiation source while increasing the number of training samples. In addition, as can be seen from the above equation, since the degrees of contribution to data enhancement by rotating different phase angles are the same, the phase rotation angleθIs chosen uniformly between 0 and 2 pi.
The present embodiment provides a signal data enhancement method for communication radiation source identification, and referring to fig. 1, the signal data enhancement method includes the following steps:
s11, generating more than 0 and less than 2 piNA rotation angle;
for sample signalxIs randomly rotated to select a rotation angleθ(i.e. the phase of the rotation),θobeying a uniform distribution between 0 and 2 pi, i.e.θ~U(0, 2 π). Therefore, the rotation angle in the present embodimentθCan generate more than 0 and less than 2 pi according to uniform distributionNAn angle of rotation.
And S12, acquiring a sample signal.
For quadrature sampled signals, each sample signalxFor complex phasors, the real and imaginary parts being respectively quadrature sampled I and Q signals, i.e.
x[l]=x I [l]+i*x Q [l]；
Wherein the content of the first and second substances,x I [l]is an I-path signal of the sample signal,x Q [l]the Q path signal, which is a sample signal, is in units of imaginary numbers,l=1,2,…,L，Lis the sample signal length.
S13, rotating the phase of each sample signalNA rotation angle ofNA replica signal.
The replica signal of this embodiment is:
x ′ =x*e iθ ；
wherein the content of the first and second substances,x ′ is a replica signal (i.e., an enhanced sample signal);xis a sample signal;θis the rotation angle (i.e. the rotation phase),θ~U(0，2π)。
the rotation angle of the present embodiment is a random rotation angle, or a rotation angle at equal intervals.
This embodiment is achieved by generating the voltage greater than 0 and less than 2 piNA rotation angle and phase-rotating the sample signal to obtain a plurality of replica signals,
in the communication radiation source individual identification, the method based on deep learning is the method with higher accuracy at present. The method uses a signal sample of training data to train a deep neural network, and inputs a test sample into the neural network in a test stage to obtain a recognition result. Because the quantity of parameters of the neural network is large, the deep learning method usually needs training data of a certain scale for training. Due to the randomness of the initial phase of the communication signal, when the number of training samples is small, the initial phase of the training samples cannot cover all possible values, so that overfitting is easily caused, namely the identification accuracy of the neural network on the test sample is low.
In the embodiment, through the phase rotation of the sample signal, the individual fingerprint characteristics of the radiation source are maintained, meanwhile, the enhancement of the sample signal data is realized, the overfitting caused by less sample signals in deep neural training is avoided, and the identification accuracy of the neural network to the test sample is improved.
Another embodiment provides a signal data enhancement system for communicating radiation source identification, the signal data enhancement system comprising:
a rotation angle generation module for generating a value greater than 0 and less than 2 piNAn angle of rotation. The rotation angle is a random rotation angle; or at equally spaced rotational angles.
The acquisition module is used for acquiring a sample signal;
a rotation module for rotating the phase of each sample signal respectivelyNA rotation angle ofNA replica signal.
A further embodiment provides an identification method of a communication radiation source, and referring to fig. 2, the identification method includes the following steps:
and S21, acquiring a first sample signal of each individual communication radiation source and a second sample signal of the individual communication radiation source to be identified.
Usually, a batch method is adopted to optimize the neural network, and random selection is carried out each timeBA sample signal, denoted asX={（x j ,c j ）}，j=1，2，…，B，BIn order to count the number of the sample signals,x j is as followsjThe number of the samples of the signal is,c j is as followsjA label corresponding to the sample signalBAnd the sample signals and the corresponding labels thereof are used for training the neural network. The sample signal in this embodiment includes a first sample signal and a second sample signal.
S22, performing data enhancement on each of the first sample signal and the second sample signal respectively according to the signal data enhancement method provided in the foregoing embodiment, to obtain a first replica signal set and a second replica signal set.
The phase rotation angle for data enhancement in the training process in this embodiment is chosen randomly between 0 and 2 pi, so that during the test phase, for each sample signalxSelected at equal intervals between 0 and 2 piNThe angle performs data enhancement. The selected sample signalxIs rotated by 0, 2 pi-N、4π/N、…,2(N-1)π/NTo obtainNA replica signal, i.e.
Wherein the content of the first and second substances,x n is a sample signalxTo (1) anThe number of the replica signals is such that,n=1，2，…，N，Nas a sample signalxThe enhanced replica signal number of (2). According to the requirements of identification accuracy and reasoning speed, selecting proper oneN，NThe larger the identification accuracy rate is, the slower the inference speed is.
And S23, inputting the first copy signal set into a neural network for training.
Data enhancement is performed on line in the network training process, the same sample signal can participate in training for multiple times, and the phase of each rotation is different.
Inputting the sample signal (replica signal) after data enhancement into a neural network for training to obtain a corresponding confidence coefficient vector, such as inputting the sample signal into the neural networkx j Is rotated by random angle to obtainx j ′ Will bex j ′ Inputting into neural network to obtain corresponding confidence coefficient vectory j ={y j [k]}，y j To (1) akAn elementy j [k]，y j [k]Is a sample signalx j Belong to the firstkThe probability of an individual source of radiation,k=1，2，3，…，K，Kthe following conditions are satisfied for the number of radiation source individuals:
0≤y j [k]≤1；
from the confidence and sample truth labels, cross-entropy losses are calculated, e.g.x j Belong to the firstkIndividual radiation source, thenx j Has a cross entropy loss off j =- klog y j [k]The smaller the cross-entropy loss is,y j [k]the greater the probability of (c). Averaging the cross entropy loss of all samples within a batch
Namely:
updating neural network minimization through back propagation algorithm
The probability that the neural network prediction is correct is maximized. The neural network may employ a convolutional neural network or a cyclic neural network until the number of training iterations is reached.
And S24, inputting the second copy signal set into the trained neural network for individual identification of the communication radiation source, and obtaining the probability that each second copy signal belongs to each individual communication radiation source.
Sampling the sample signalxIs/are as followsNReplica signalx ′n Respectively transportEntering a neural network to obtain a confidence vector corresponding to each replica signaly n ={y n [k]}, confidence vectory n To (1) akAn elementy n [k]As replica signalsx n Belong to the firstkProbability of individual sources.
And S25, calculating the average probability values of all the second copy signals respectively belonging to each communication radiation source individual, and selecting the communication radiation source individual corresponding to the maximum average probability value as a final identification result.
Due to replica signalsx ′n Is the same sample signalxRotating different replica signals obtained by different phases, and averaging the confidence coefficients to obtain an average confidence coefficient, namely:
wherein the content of the first and second substances,y[k]is composed ofNAll the replica signals belong tokThe average confidence of each individual radiation source;y n [k]is as followsnA replica signalx ′n Belong to the firstkProbability of individual sources.
Selecting an average confidence vectory=={y[k]The label corresponding to the maximum average confidence in thekAs sample signalxThe identification result of (a) is:
the following describes the identification process based on data collected by an actual device.
The embodiment totally comprises 8 signal samples generated by communication radiation sources, wherein the sampling rate is 50MHz, the signal carrier frequency is 500MHz, the modulation mode is QPSK, and the modulation rate is 1 MBaud/s. The number of training data and test data samples of each individual radiation source is 500, each signal sample comprises two I/Q channels, each channel comprises 8192 sampling points, namelyN J =500*8，K=8，L=8192。
The neural network is adopted for individual identification of the radiation source, the neural network of the embodiment comprises 9 convolution modules and 1 full connection layer, and the network structure is shown in fig. 3. Each convolution module includes a convolution layer, a batch normalization layer, and a linear rectification layer. Each convolution layer is composed of a plurality of convolution units, the convolution layer performs convolution operation on parameters of each convolution unit and input data to obtain output, and the parameters of the convolution units are obtained through optimization of a back propagation algorithm. The convolution operation can be expressed as:
wherein the content of the first and second substances,hare convolution unit parameters.
The convolution operation aims to extract different levels of input features, the first layer of convolution layer can only extract some low-level features such as edges, lines, corners and other levels, and more layers of networks can iteratively extract more complex features from the low-level features. The batch normalization layer performs normalization processing on the output of the convolution layer so as to facilitate optimization of a subsequent network layer. Linear rectifying layers, i.e. ramp functions in mathematicsr(x)=max(0,x) Which enhances the non-linear behavior of the network as an excitation function. Finally, a fully-connected layer is used to map the features extracted by the convolution module to the required dimensions, which can be expressed asfc(x)=xW+bIn this case
，
，
. After passing through the full connection layer, for each sample signalxObtaining an output vector z, the vector comprisingK=8 elements, nokZ [ for individual element ]k]Showing that z is subjected to softmax function to obtain the sampleThe signalxCorresponding confidence vectory：
Wherein the content of the first and second substances,y[k]representing a sample signalxBelong to the firstkProbability of individual sources.
The training process of the neural network is as follows: firstly, standardizing all data, namely subtracting a sample mean value from sample data and dividing the sample mean value by a sample standard deviation for each sample signal to normalize numerical values of all samples to the same value interval; next, 64 samples are randomly selected from the training data at a time, i.e.B=64, data dimension of the batch is [64, 2, 8192]Rotating the phase of each sample signal by random angles respectively, inputting the data enhanced samples into a neural network, and outputting the data with dimensionality of [64,8 ] by the network]Calculating a loss function according to the network output and the sample real label, wherein the loss function adopts cross entropy loss, namely, for each sample signalxIf its true tag iskThen its corresponding loss is-logy[k]And averaging the loss values of 64 samples to obtain a final loss function value. Updating the network weight through a back propagation algorithm to minimize a loss function value; the sample selection, data enhancement, and network update processes are repeated until all training data are traversed 500 times.
The testing process of the neural network is as follows: sequentially selecting a sample from the test data; rotating the phase of the sampleNAt a fixed angle, obtaining the sample signalNA copy; will be provided withNThe replica signal is input into a neural network and outputNAveraging the confidence vectors to obtain the confidence vector of the sample signal; obtaining a recognition result according to the confidence coefficient vector of the sample signal, wherein the recognition result is obtained in an experimentNThe comparison was performed with the settings 1, 2, 3, 4, and 5 in this order. To be provided withNFor example, =2, for one test sample signalxThe phase of the two signals is respectively rotated by 0 degree and 180 degrees to obtain two replica signalsx ′1 Andx ′2 will bex ′1 Andx ′2 respectively input into the neural network to obtain output confidence vectors respectivelyy 1 Andy 2 then, the two confidence vectors are averaged, i.e.:
y[k]=（y 2 [k]+y 2 [k]）/2；
gety[k]The maximum value in (4) corresponds to the individual radiation source as the identification result.
When data enhancement is not used, the recognition accuracy of the neural network on the test data is 78.64%; the recognition accuracy enhanced using the data is shown in fig. 4. The data enhancement method based on phase rotation can effectively improve the accuracy of individual identification of the communication radiation source, and the number of data enhancement copies in the testNThe larger the identification accuracy and the longer the test time.
Yet another embodiment provides an identification system of a communication radiation source, the identification system comprising:
the acquisition module is used for acquiring a first sample signal of each communication radiation source individual and a second sample signal of the communication radiation source individual to be identified;
a data enhancement module, configured to perform data enhancement on each of the first sample signal and the second sample signal respectively to obtain a first replica signal set and a second replica signal set;
the training module is used for inputting the first copy signal set into a neural network for training;
the communication radiation source individual identification module is used for inputting the second copy signal set into the trained neural network to carry out communication radiation source individual identification, and the probability that each second copy signal belongs to each communication radiation source individual is obtained;
and the selecting module is used for calculating the average probability values of all the second copy signals respectively belonging to all the communication radiation source individuals and selecting the communication radiation source individual corresponding to the maximum average probability value as a final identification result.
Although the embodiments of the present invention have been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the spirit and scope of the embodiments of the present invention.
Claims (9)
1. A signal data enhancement method for communication radiation source identification, the signal data enhancement method comprising the steps of:
generating more than 0 and less than 2 piNA rotation angle;
acquiring a sample signal;
rotating the phase of each sample signal separatelyNA rotation angle ofNThe replica signal.
2. The signal data enhancement method of claim 1, wherein the generating is greater than 0 and less than 2 piNThe rotation angles further include:
according to a uniform distribution to generate a value greater than 0 and less than 2 piNAn angle of rotation.
3. The signal data enhancement method of claim 2, wherein the replica signal is:
x ′ =x*e iθ ；
wherein the content of the first and second substances,x ′ is a replica signal;xis a sample signal;θis the angle of rotation,θ~U(0，2π)。
4. a signal data enhancement method according to any one of claims 1 to 3, characterized in that the rotation angle is a random rotation angle; or at equally spaced rotational angles.
5. A signal data enhancement system for communicating radiation source identification, the signal data enhancement system comprising:
a rotation angle generation module for generating a value greater than 0 and less than 2 piNA rotation angle;
the acquisition module is used for acquiring a sample signal;
a rotation module for rotating the phase of each sample signal respectivelyNA rotation angle ofNA replica signal.
6. The signal data enhancement system of claim 5, wherein the replica signal is:
x ′ =x*e iθ ；
wherein the content of the first and second substances,x ′ is a replica signal;xis a sample signal;θis the angle of rotation,θ~U(0，2π)。
7. a signal data enhancement system according to claim 5 or 6 wherein the rotation angle is a random rotation angle; or at equally spaced rotational angles.
8. A method for identifying a communication radiation source, the method comprising the steps of:
the method comprises the steps of firstly, obtaining a first sample signal of each communication radiation source individual and a second sample signal of a communication radiation source individual to be identified;
step two, according to the signal data enhancement method of any one of claims 1 to 4, performing data enhancement on each of the first sample signal and the second sample signal respectively to obtain a first replica signal set and a second replica signal set;
inputting the first copy signal set into a neural network for training;
inputting the second copy signal set into the trained neural network to perform communication radiation source individual identification, and obtaining the probability that each second copy signal belongs to each communication radiation source individual;
and fifthly, calculating the average probability values of all the second replica signals respectively belonging to all the communication radiation source individuals, and selecting the communication radiation source individual corresponding to the maximum average probability value as a final identification result.
9. An identification system for a communication radiation source, the identification system comprising:
the acquisition module is used for acquiring a first sample signal of each communication radiation source individual and a second sample signal of the communication radiation source individual to be identified;
a data enhancement module, configured to perform data enhancement on each of the first sample signal and the second sample signal respectively according to the signal data enhancement method of any one of claims 1 to 4, so as to obtain a first replica signal set and a second replica signal set;
the training module is used for inputting the first copy signal set into a neural network for training;
the communication radiation source individual identification module is used for inputting the second copy signal set into the trained neural network to carry out communication radiation source individual identification, and the probability that each second copy signal belongs to each communication radiation source individual is obtained;
and the selecting module is used for calculating the average probability values of all the second copy signals respectively belonging to all the communication radiation source individuals and selecting the communication radiation source individual corresponding to the maximum average probability value as the final identification result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210807165.4A CN114866162B (en) | 2022-07-11 | 2022-07-11 | Signal data enhancement method and system and communication radiation source identification method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210807165.4A CN114866162B (en) | 2022-07-11 | 2022-07-11 | Signal data enhancement method and system and communication radiation source identification method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114866162A true CN114866162A (en) | 2022-08-05 |
| CN114866162B CN114866162B (en) | 2023-09-26 |
Family
ID=82627076
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210807165.4A Active CN114866162B (en) | 2022-07-11 | 2022-07-11 | Signal data enhancement method and system and communication radiation source identification method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114866162B (en) |
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008032407A1 (en) * | 2006-09-15 | 2008-03-20 | Fujitsu Limited | Apparatus and method for transmitting signals by use of multicarrier system |
| CN109633588A (en) * | 2018-12-29 | 2019-04-16 | 杭州电子科技大学 | Recognition Method of Radar Emitters based on depth convolutional neural networks |
| US10291335B1 (en) * | 2017-12-04 | 2019-05-14 | Industrial Technology Research Institute | Beamforming calibration system and method |
| EP3651067A1 (en) * | 2018-11-12 | 2020-05-13 | ISHIDA CO., Ltd. | Training data generation method, training data generation apparatus, and product identification apparatus |
| CN111563227A (en) * | 2020-04-30 | 2020-08-21 | 中国人民解放军军事科学院评估论证研究中心 | Fingerprint characteristic parameter extraction method for radiation source signal and radiation source identification |
| CN111693997A (en) * | 2019-03-12 | 2020-09-22 | 半导体元件工业有限责任公司 | High resolution multiple input multiple output radar system |
| CN112115821A (en) * | 2020-09-04 | 2020-12-22 | 西北工业大学 | Multi-signal intelligent modulation mode identification method based on wavelet approximate coefficient entropy |
| CN112348060A (en) * | 2020-10-26 | 2021-02-09 | 上海眼控科技股份有限公司 | Classification vector generation method and device, computer equipment and storage medium |
| CN112398552A (en) * | 2020-11-03 | 2021-02-23 | 上海电机学院 | Communication radiation source individual identification method based on differential complex deep neural network |
| CN112734709A (en) * | 2020-12-31 | 2021-04-30 | 山西三友和智慧信息技术股份有限公司 | Melanoma detection method based on attention mechanism and transfer learning |
| CN112990432A (en) * | 2021-03-04 | 2021-06-18 | 北京金山云网络技术有限公司 | Target recognition model training method and device and electronic equipment |
| CN113536919A (en) * | 2021-06-10 | 2021-10-22 | 重庆邮电大学 | Signal modulation recognition algorithm based on data enhancement and convolutional neural network |
| CN113705446A (en) * | 2021-08-27 | 2021-11-26 | 电子科技大学 | Open set identification method for individual radiation source |
| CN113763573A (en) * | 2021-09-17 | 2021-12-07 | 北京京航计算通讯研究所 | Three-dimensional object digital marking method and device |
| WO2022002059A1 (en) * | 2020-06-30 | 2022-01-06 | 北京灵汐科技有限公司 | Initial neural network training method and apparatus, image recognition method and apparatus, device, and medium |
| CN114519679A (en) * | 2022-02-21 | 2022-05-20 | 安徽大学 | Intelligent SAR target image data enhancement method |
-
2022
- 2022-07-11 CN CN202210807165.4A patent/CN114866162B/en active Active
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008032407A1 (en) * | 2006-09-15 | 2008-03-20 | Fujitsu Limited | Apparatus and method for transmitting signals by use of multicarrier system |
| US10291335B1 (en) * | 2017-12-04 | 2019-05-14 | Industrial Technology Research Institute | Beamforming calibration system and method |
| EP3651067A1 (en) * | 2018-11-12 | 2020-05-13 | ISHIDA CO., Ltd. | Training data generation method, training data generation apparatus, and product identification apparatus |
| CN109633588A (en) * | 2018-12-29 | 2019-04-16 | 杭州电子科技大学 | Recognition Method of Radar Emitters based on depth convolutional neural networks |
| CN111693997A (en) * | 2019-03-12 | 2020-09-22 | 半导体元件工业有限责任公司 | High resolution multiple input multiple output radar system |
| CN111563227A (en) * | 2020-04-30 | 2020-08-21 | 中国人民解放军军事科学院评估论证研究中心 | Fingerprint characteristic parameter extraction method for radiation source signal and radiation source identification |
| WO2022002059A1 (en) * | 2020-06-30 | 2022-01-06 | 北京灵汐科技有限公司 | Initial neural network training method and apparatus, image recognition method and apparatus, device, and medium |
| CN112115821A (en) * | 2020-09-04 | 2020-12-22 | 西北工业大学 | Multi-signal intelligent modulation mode identification method based on wavelet approximate coefficient entropy |
| CN112348060A (en) * | 2020-10-26 | 2021-02-09 | 上海眼控科技股份有限公司 | Classification vector generation method and device, computer equipment and storage medium |
| CN112398552A (en) * | 2020-11-03 | 2021-02-23 | 上海电机学院 | Communication radiation source individual identification method based on differential complex deep neural network |
| CN112734709A (en) * | 2020-12-31 | 2021-04-30 | 山西三友和智慧信息技术股份有限公司 | Melanoma detection method based on attention mechanism and transfer learning |
| CN112990432A (en) * | 2021-03-04 | 2021-06-18 | 北京金山云网络技术有限公司 | Target recognition model training method and device and electronic equipment |
| CN113536919A (en) * | 2021-06-10 | 2021-10-22 | 重庆邮电大学 | Signal modulation recognition algorithm based on data enhancement and convolutional neural network |
| CN113705446A (en) * | 2021-08-27 | 2021-11-26 | 电子科技大学 | Open set identification method for individual radiation source |
| CN113763573A (en) * | 2021-09-17 | 2021-12-07 | 北京京航计算通讯研究所 | Three-dimensional object digital marking method and device |
| CN114519679A (en) * | 2022-02-21 | 2022-05-20 | 安徽大学 | Intelligent SAR target image data enhancement method |
Non-Patent Citations (5)
| Title |
|---|
| （瑞士）翁贝托·米凯卢奇著: "《深度学习进阶 卷积神经网络和对象检测》", 30 July 2020, 机械工业出版社, pages: 201 * |
| XUEBAO WANG: "Radar emitter recognition based on the short time fourier transform and convolutional neural networks", 《2017 10TH INTERNATIONAL CONGRESS ON IMAGE AND SIGNAL PROCESSING, BIOMEDICAL ENGINEERING AND INFORMATICS (CISP-BMEI)》 * |
| XUEBAO WANG: "Radar emitter recognition based on the short time fourier transform and convolutional neural networks", 《2017 10TH INTERNATIONAL CONGRESS ON IMAGE AND SIGNAL PROCESSING, BIOMEDICAL ENGINEERING AND INFORMATICS (CISP-BMEI)》, 26 February 2018 (2018-02-26) * |
| 普运伟: "基于卷积神经网络和模糊函数主脊坐标变换的雷达辐射源信号识别", 《兵工学报》 * |
| 普运伟: "基于卷积神经网络和模糊函数主脊坐标变换的雷达辐射源信号识别", 《兵工学报》, vol. 42, no. 8, 30 August 2021 (2021-08-30) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114866162B (en) | 2023-09-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | An efficient deep learning model for automatic modulation recognition based on parameter estimation and transformation | |
| Meng et al. | Automatic modulation classification: A deep learning enabled approach | |
| Li et al. | Robust automated VHF modulation recognition based on deep convolutional neural networks | |
| CN110113288B (en) | Design and demodulation method of OFDM demodulator based on machine learning | |
| CN111428817B (en) | Defending method for radio signal identification against attack | |
| Daldal et al. | Deep long short-term memory networks-based automatic recognition of six different digital modulation types under varying noise conditions | |
| CN109617845A (en) | A kind of design and demodulation method of the wireless communication demodulator based on deep learning | |
| Wang et al. | Convolutional neural network applied to specific emitter identification based on pulse waveform images | |
| Zhang et al. | A deep learning approach for modulation recognition | |
| Yunhao et al. | ConvLSTMAE: A spatiotemporal parallel autoencoders for automatic modulation classification | |
| Zhu et al. | Y-shaped net-based signal detection for OFDM-IM systems | |
| Zhang et al. | High-order convolutional attention networks for automatic modulation classification in communication | |
| Zhao et al. | The identification of secular variation in IoT based on transfer learning | |
| Yang et al. | Deep learning based RFF recognition with differential constellation trace figure towards closed and open set | |
| CN114866162B (en) | Signal data enhancement method and system and communication radiation source identification method and system | |
| CN116738251A (en) | Radio frequency fingerprint identification training data generation method based on generation countermeasure network | |
| CN116471154A (en) | Modulation signal identification method based on multi-domain mixed attention | |
| Yin et al. | Short-time modulation classification of complex wireless communication signal based on deep neural network | |
| Zhang et al. | Towards the automatic modulation classification with adaptive wavelet network | |
| CN113361569A (en) | Electromagnetic signal modulation type identification model construction method based on time-frequency fusion information | |
| Wang et al. | Aircarft Signal Feature Extraction and Recognition Based on Deep Learning | |
| Liang et al. | Self-training based adversarial domain adaptation for radio signal recognition | |
| Rong et al. | RFFsNet-SEI: A multidimensional balanced-RFFs deep neural network framework for specific emitter identification | |
| Motwani et al. | Analysis of various neural network architectures for automatic modulation techniques | |
| Ye et al. | Cross-session Specific Emitter Identification using Adversarial Domain Adaptation with Wasserstein distance |
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 |