CN115062642A - Signal radiation source identification method, device, equipment and storage medium - Google Patents
Signal radiation source identification method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN115062642A CN115062642A CN202210521459.0A CN202210521459A CN115062642A CN 115062642 A CN115062642 A CN 115062642A CN 202210521459 A CN202210521459 A CN 202210521459A CN 115062642 A CN115062642 A CN 115062642A
- Authority
- CN
- China
- Prior art keywords
- complex
- radiation source
- signal
- radiation
- individual
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The application provides a signal radiation source identification method, a device, equipment and a storage medium, wherein the method comprises the following steps: extracting individual complex characteristics of a radiation source from the acquired complex radiation signals, wherein the complex radiation signals are formed by combining real part signals and imaginary part signals; and identifying a signal radiation source irradiating the complex radiation signal based on the individual complex characteristics of the radiation sources. The signal source identification method makes full use of the signal characteristic information contained in the real part signal and the imaginary part signal of the complex signal, so that the extracted individual complex characteristic of the radiation source is more accurate, and the more accurate signal radiation source identification effect can be obtained.
Description
Technical Field
The present application relates to the field of computer technologies, and in particular, to a method, an apparatus, a device, and a storage medium for identifying a signal radiation source.
Background
In an actual electromagnetic environment, the number of signal radiation sources is huge, the signal modulation pattern is complex and changeable, and the electromagnetic signals of the whole electromagnetic environment are dense and complex, so that the difficulty in identifying the signal radiation sources is increased. Therefore, there is a need to develop an effective signal radiation source identification scheme to achieve effective identification of the signal radiation source.
Disclosure of Invention
Based on the above technical current situation, the present application provides a method, an apparatus, a device and a storage medium for identifying a signal radiation source, which can realize effective identification of the signal radiation source.
A first aspect of the present application provides a signal radiation source identification method, including:
extracting individual complex characteristics of a radiation source from the acquired complex radiation signals, wherein the complex radiation signals are formed by combining real part signals and imaginary part signals;
and identifying a signal radiation source irradiating the complex radiation signal based on the individual complex characteristics of the radiation sources.
Optionally, extracting the individual complex features of the radiation source from the acquired complex radiation signals includes:
inputting the acquired complex radiation signals into a pre-trained complex feature extraction model to obtain individual complex features of the radiation source;
the complex feature extraction model is obtained by extracting complex features of the signal radiation source individuals from complex radiation signals radiated by the signal radiation source individuals and training.
Optionally, the complex feature extraction model is obtained by performing operator adjustment on a preset contrast predictive coding network according to a complex operation rule.
Optionally, the operator adjustment is performed on the preset contrast predictive coding network according to a complex operation rule, including:
based on a complex operation rule, constructing a complex operation operator by using a preset real operation operator in a contrast predictive coding network;
and replacing the real number arithmetic operator in the contrast predictive coding network with a corresponding complex number arithmetic operator.
Optionally, identifying a signal radiation source that radiates the complex radiation signal based on the individual complex features of the radiation source includes:
inputting the individual complex features of the radiation source into a pre-trained complex feature classification model, and identifying a signal radiation source for radiating the complex radiation signal;
the complex characteristic classification model is obtained by taking the individual complex characteristics of the radiation source as input and performing signal radiation source classification and identification training.
Optionally, the complex feature classification model is obtained by performing operator adjustment on a preset depth residual error network according to a complex operation rule.
Optionally, the operator adjustment is performed on the preset depth residual error network according to a complex number operation rule, including:
constructing a complex operator by utilizing a real operator in a preset depth residual error network based on a complex operation rule;
and replacing the real number arithmetic operator in the depth residual error network with a corresponding complex number arithmetic operator.
Optionally, identifying a signal radiation source that radiates the complex radiation signal based on the individual complex features of the radiation source includes:
performing modular operation on the individual complex features of the radiation source;
inputting the modular operation result of the individual complex features of the radiation source into a depth residual error network trained in advance, and identifying a signal radiation source for radiating the complex radiation signal;
the deep residual error network is obtained by performing individual classification, identification and training of the radiation source based on input individual characteristics of the radiation source.
Optionally, the method for extracting the individual complex features of the radiation source from the acquired complex radiation signals, and identifying the signal radiation source radiating the complex radiation signals based on the individual complex features of the radiation source includes:
inputting the acquired complex radiation signals into a pre-trained signal radiation source identification model so that the signal radiation source identification model extracts individual complex characteristics of a radiation source from the acquired complex radiation signals, and identifying the signal radiation source irradiating the complex radiation signals based on the individual complex characteristics of the radiation source;
the signal radiation source identification model is constructed by a complex contrast prediction coding network and a complex depth residual error network, the complex contrast prediction coding network is constructed based on a complex arithmetic operator, and the complex depth residual error network is constructed based on the complex arithmetic operator.
A second aspect of the present application provides a signal radiation source identification apparatus, the apparatus comprising:
the characteristic extraction unit is used for extracting and obtaining the individual complex characteristics of the radiation source from the acquired complex radiation signals, wherein the complex radiation signals are formed by combining real part signals and imaginary part signals;
and the characteristic classification unit is used for identifying a signal radiation source for radiating the complex radiation signal based on the individual complex characteristics of the radiation source.
A third aspect of the present application provides a signal radiation source identification device, the device comprising:
a memory and a processor;
wherein the memory is connected with the processor and used for storing programs;
the processor is used for realizing the signal radiation source identification method by running the program in the memory.
A fourth aspect of the present application provides a storage medium having a computer program stored thereon, where the computer program, when executed by a processor, implements the signal radiation source identification method described above.
The signal radiation source identification method can process the complex radiation signals and extract individual complex characteristics of the radiation source from the complex radiation signals. The characteristic extraction mode fully utilizes the signal characteristic information contained in the real part signal and the imaginary part signal of the complex signal, so that the extracted individual complex characteristic of the radiation source is more accurate. The signal radiation source for radiating the complex radiation signal is identified based on the individual complex characteristics of the radiation source, so that a more accurate signal radiation source identification effect can be obtained.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the embodiments or the prior art descriptions will be briefly described below, it is obvious that the drawings in the following description are only the embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a diagram of a contrast predictive coding network framework provided in an embodiment of the present application;
fig. 2 is a schematic flowchart of a signal radiation source identification method according to an embodiment of the present application;
FIG. 3 is a schematic diagram illustrating a complex convolution process according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of a complex contrast predictive coding network according to an embodiment of the present application;
fig. 5 is a schematic structural diagram of a complex depth residual error network according to an embodiment of the present disclosure;
FIG. 6 is a schematic diagram of a signal radiation source identification process provided in an embodiment of the present application;
fig. 7 is a schematic structural diagram of a signal radiation source identification device according to an embodiment of the present application;
fig. 8 is a schematic structural diagram of a signal radiation source identification device according to an embodiment of the present application.
Detailed Description
The technical scheme of the embodiment of the application is suitable for application scenes of signal radiation source identification, such as identification of signal radiation sources in electromagnetic environments and the like. By adopting the technical scheme of the embodiment of the application, the signal radiation source for radiating the signal can be effectively identified on the basis of the captured radiation signal.
Currently, in the field of Specific Emitter Identification (SEI), the most commonly used individual Identification scheme of the radiation source is based on CPC (proportional Predictive Coding, contrast Predictive Coding) + ResNet (deep Residual Network).
CPC is an unsupervised characterization learning technology based on a deep neural network, and the scheme utilizes the CPC network to automatically extract the individual fingerprint characteristics of a radiation source. In the CPC model training phase, the parameter updating and optimization of the network are completed based on the comparison learning and the prediction coding. Predictive coding, i.e. using the GRU network, inputs Z based on the current time t And the state output S at the previous moment t-1 To obtain the output O of the current time t And using O t Predicting the Embellding vectors at n moments in the future to obtain
To
Then the prediction result and the real Embedding vector Z t+1 To Z t+n Similarity scoring is performed to judge whether the prediction effect is good or not, and as shown in fig. 1, the prediction result at the t + k moment
The calculation is made by the following formula:
wherein Wk is a linear layer, which is optimized by updating parameters.
By the predictive coding mode, namely, the Embedding at the current time and the past time is utilized to predict the Embedding at the future time, so that a background feature which stably exists in the whole stage in the signal is effectively extracted.
Contrast learning in the training phase, each radiation source sample x [ i ] in the same batch is utilized](i ═ 1,2 …, M) output O at the current time t (i) To predict all radiation source samples in the batch at n time points in the future
The training criterion is X [ i ] per sample]By O of the current sample t Predicting the future n moments of the current sample
With similarity score increasing and predicting other samples at n future times
(M ≠ 1,2 … M, and M ≠ i) the similarity score gets lower and lower as shown by:
here, f (x) is used to represent similarity score calculation, such as calculating cosine similarity.
Inputting a radiation source signal X for a contrast prediction coding model which is trained t Then, the coding layer containing 5 one-dimensional convolutional layers is sent to generate an Embedding vector Z t . Then the Embedding vector Z t Sending the data into a GRU layer to generate a feature vector O with context information t The feature vector is the individual feature of the radiation source.
Z t ＝f 1d_cnn (X t )
O t ＝f GRU (Z t )
After the radiation source individual features are extracted, for a back-end classification module, the scheme takes a deep Residual error Network (ResNet) as a radiation source individual recognition back-end classifier. The network can increase the depth of a network model without the problem of network degradation through residual learning, so that the individual identification accuracy of the radiation source can be effectively improved.
The radiation source individual identification scheme based on CPC + ResNet is not good in identification effect on a signal radiation source in practical application.
After intensive research, the inventor of the present invention finds that the signals radiated by the signal radiation source are complex signals including real part signals and imaginary part signals, and the real part signals and the imaginary part signals of the complex signals are combined to include rich characterization information, such as phase and the like. However, conventional radiation source individual identification schemes, including the above radiation source individual identification scheme based on CPC + ResNet, all use only the real part signal of the radiation signal to perform calculation, which will lose a large amount of information, thereby resulting in inaccurate grasping of signal characteristics, and further directly affecting the identification effect of the signal radiation source.
In view of the above problems, the inventors of the present application have further studied and proposed the identification scheme of the signal radiation source provided in the embodiments of the present application, and through experimental comparison, the identification scheme of the signal radiation source provided in the embodiments of the present application has a higher identification accuracy of the signal radiation source.
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Exemplary method
The embodiment of the present application first provides a method for identifying a signal radiation source, which is shown in fig. 2 and includes:
s201, extracting and obtaining individual complex characteristics of the radiation source from the acquired complex radiation signals.
The complex radiation signal refers to a wireless signal captured from an electromagnetic signal environment by a signal acquisition device such as a radar or an antenna. In the embodiment of the application, the signal radiated by the signal radiation source is captured from a dense and complex electromagnetic signal environment, and the captured signal is a complex radiation signal because the signal radiated by the signal radiation source is a complex signal. The complex radiation signal is formed by combining a real part signal and an imaginary part signal.
Due to differences in the type, use, carrier, operating parameters, etc. of the signal radiation sources, the signals radiated by the individual signal radiation sources exhibit different and unique signal characteristics. Based on the characteristics of the acquired complex radiation signals, the type, the application, the carrier and the like of the signal radiation source individual radiating the complex radiation signals can be deduced reversely, so that the aim of identifying the signal radiation source is fulfilled.
Based on the technical principle, the embodiment of the application performs feature extraction on the acquired complex radiation signal to obtain the signal feature of the complex radiation signal, and the signal feature can be used as the feature of the radiation source individual corresponding to the complex radiation signal.
Meanwhile, the signal feature is a signal feature obtained by performing feature extraction on the complex radiation signal, and includes complex information, so that the signal feature can be regarded as a complex signal feature, and accordingly, the complex signal feature can be regarded as an individual complex feature of the radiation source corresponding to the complex radiation signal.
By way of example, by encoding the complex radiation signal, the encoding characteristic of the complex radiation signal can be obtained, and the encoding characteristic can be used as the individual complex characteristic of the radiation source corresponding to the complex radiation signal.
S202, identifying a signal radiation source for radiating the complex radiation signal based on the individual complex characteristics of the radiation source.
As mentioned above, the complex signals radiated by different signal radiation sources have different and unique signal characteristics, and accordingly, the signal characteristics of the complex radiation source signals can be used as unique individual radiation source characteristics of the signal radiation source radiating the signals.
Therefore, after the complex radiation signal is subjected to feature extraction to obtain the individual complex feature of the radiation source corresponding to the complex radiation signal by performing step S201, the signal radiation source radiating the complex radiation signal can be identified according to the individual complex feature of the radiation source.
For example, a variety of different types, different uses, different carriers, different operating parameters of the signal radiation source, and its radiated complex radiation signal may be pre-collected. Then, the characteristic extraction is carried out on the complex radiation signal radiated by each signal radiation source to obtain the individual complex characteristic of the signal radiation source.
On this basis, the individual complex features of the radiation source extracted in step S201 are compared with the individual complex features of each predetermined signal source, and the individual complex features that are the same as the extracted individual complex features of the radiation source or have a similarity greater than a set similarity threshold are determined, where the signal source corresponding to the individual complex features is a signal radiation source for radiating the complex radiation signals.
The signal radiation source for identifying the complex radiation signal includes, but is not limited to, identifying the type, application, carrier, wavelength band, threat level, reliability, etc. of the signal radiation source for radiating the complex radiation signal.
The signal radiation source identification method provided by the embodiment of the application can process the complex radiation signals and extract the individual complex features of the radiation source from the complex radiation signals. The characteristic extraction mode fully utilizes the signal characteristic information contained in the real part signal and the imaginary part signal of the complex signal, so that the extracted individual complex characteristic of the radiation source is more accurate. The signal radiation source for radiating the complex radiation signal is identified based on the individual complex characteristics of the radiation source, so that a more accurate signal radiation source identification effect can be obtained.
As an optional implementation manner, in the embodiment of the present application, a complex feature extraction model is trained in advance, and is used for processing a complex radiation signal, and extracting a complex feature of an individual signal radiation source corresponding to the complex radiation signal.
A large number of complex radiation signals radiated by the signal radiation source individuals are collected in advance, and each complex radiation signal and the corresponding signal radiation source individual are recorded respectively.
Respectively inputting the collected complex radiation signals into the complex feature extraction model to obtain the individual complex features of the signal radiation source output by the model, then carrying out individual identification of the signal radiation source based on the individual complex features of the signal radiation source output by the model, comparing the identification result with the signal radiation source individual corresponding to the input complex radiation signals, calculating the loss function of the identification result, and carrying out parameter correction on the complex feature extraction model based on the loss function.
After the training of the complex feature extraction model is completed by repeating the above process, any complex radiation signal can be input into the complex feature extraction model to obtain the individual complex features of the radiation source corresponding to the input complex radiation signal.
As a preferred implementation manner, in the embodiments of the present application, the complex feature extraction model is constructed based on a Comparative Predictive Coding (CPC).
As mentioned above, the contrast-predictive coding network is an unsupervised characterization learning technique based on a deep neural network, which can be used to automatically extract the "fingerprint" features of the signal radiation source individuals. However, the conventional contrast predictive coding network is based on real number operation, that is, each operator is an operator for a real number, and is suitable for operating on a real number and cannot be suitable for processing a complex radiation signal.
Therefore, according to the complex operation rule, the embodiment of the application performs operator adjustment on the contrast prediction coding network, so that the included operators are suitable for processing the complex radiation signals, and the individual complex features of the radiation source can be extracted and obtained.
Specifically, firstly, based on a complex operation rule, a preset real number operator in a contrast predictive coding network is utilized to construct a complex number operator; and then replacing the real number arithmetic operator in the comparison predictive coding network with a corresponding complex number arithmetic operator, thereby realizing the operator adjustment.
Complex numbers are usually expressed as z ═ a + ib (a, b are real numbers), where a is the real part, b is the imaginary part, and i is the imaginary unit. And the operation between complex numbers is also essentially composed of real number operation, so that the real number network operator can be used for building the complex number network operator according to the complex number operation rule.
For the convolutional neural network of the contrast prediction coding network, the complex convolutional neural network construction method is as follows:
for complex input h ═ x + iy (x and y are real vectors), assuming that the complex filter matrix is W ═ a + iB (a and B are real matrices), the complex convolutional network output obtained according to the complex algorithm is:
W*h＝(A*x–B*y)+i(B*x+A*y)
if the real and imaginary parts of the convolution operation are represented in matrix form, the above equation can be expressed as:
wherein R (W × h) is the real part of the W × h convolution operation, and I (W × h) is the imaginary part.
Taking the two-dimensional complex convolution Conv2d as an example, the complex input is denoted as h and the real part is h R Imaginary part of h I The convolution kernel is denoted as W and the real part is W R Imaginary part of W I . From the above analysis, it can be seen that the convolution result can be divided into four parts: h is R And W R Performing convolution, h I And W I Performing convolution, h R And W I Performing convolution, h I And W R And performing convolution. The four parts of convolution are carried out according to a real number mode to obtain four new outputs, wherein h R *W R -h I *W I Is the real part of the complex output, h R *W I +h I *W R For the imaginary part of the complex output, the real part and the imaginary part are combined into a new complex number, that is, the complex of the complex convolutionAnd (6) outputting the number. As shown in fig. 3, the operation process of the two-dimensional complex convolution network is composed of a real convolution network.
Similarly, for other complex network operators, such as the complex linear layer of the network, the convolution operation in the complex convolution network can be realized by the real linear layer by only converting the convolution operation into matrix multiplication. And other implementation principles such as the complex activation function ReLU, the complex batch normalization layer and the complex pooling layer can be implemented by corresponding real number networks.
Referring to the above processing, after the real number arithmetic operator in the network is constructed to obtain the corresponding Complex number arithmetic operator, the real number arithmetic operator in the contrast Predictive Coding network is replaced with the corresponding Complex number arithmetic operator, so that the original contrast Predictive Coding network is improved to a Complex contrast Predictive Coding network (CCPC), and the Complex contrast Predictive Coding network can input Complex signals and perform arithmetic processing on the Complex signals.
Referring to the above embodiments, in the embodiments of the present application, a complex feature classification model is further constructed in advance, and is used for classifying complex features of radiation source individuals, so as to identify signal radiation source individuals.
As a preferred implementation manner, in the embodiment of the present application, a deep Residual Network (ResNet) is used as a basis to construct the complex feature classification model.
As mentioned above, the deep residual network is also based on real number operation, that is, each operator is an operator for a real number, and is suitable for classifying and identifying real number signal features, but not suitable for processing complex number features.
According to the method and the device, operator adjustment is carried out on the deep residual error network according to a complex number operation rule, so that the contained operators are suitable for processing the complex number characteristics of the radiation source individuals, and classification and identification of the signal radiation source individuals can be realized.
The operator adjustment scheme of the depth residual error network is the same as the operator adjustment scheme of the contrast predictive coding network, and a complex operator is constructed by utilizing a real number operator in a preset depth residual error network based on a complex operation rule; and replacing the real number arithmetic operator in the depth residual error network with a corresponding complex number arithmetic operator.
Specifically, the specific process of constructing a complex operator by using a real operator in a deep residual error network may refer to the process of constructing a complex operator by using a real operator in a contrast predictive coding network described in the foregoing embodiment. Namely, according to a Complex operation rule, a Complex Network operator is constructed by using a real Network operator in the depth Residual Network, and then the constructed Complex Network operator is used for replacing the corresponding real Network operator in the depth Residual Network, so that a Complex depth Residual Network (CResNet) is obtained.
As an optional training method, when the complex feature classification model is trained, a large number of complex feature vectors of radiation source individuals are collected in advance, and each radiation source individual complex feature vector and a signal radiation source individual corresponding to the radiation source individual complex feature vector are recorded respectively.
Respectively inputting the collected radiation source individual complex feature vectors into the complex feature classification model to obtain a signal radiation source individual classification result output by the model, then comparing the signal radiation source individual classification result output by the model with the signal radiation source individual corresponding to the input radiation source individual complex feature vectors, calculating a loss function of the classification result, and performing parameter correction on the complex feature classification model based on the loss function.
After the training of the complex feature classification model is completed by repeating the above process, for any individual complex feature of the radiation source, the individual complex feature of the radiation source can be input into the complex feature classification model, and the individual classification and identification result of the radiation source corresponding to the input individual complex feature of the radiation source is obtained.
In addition, the complex feature extraction model and the complex feature classification model may be subjected to joint training, that is, a complex radiation signal sample is input into the complex feature extraction model to obtain a complex feature of a radiation source individual, and then the complex feature of the radiation source individual is input into the complex feature classification model to obtain a signal radiation source identification result. And comparing the signal radiation source identification result with a signal radiation source corresponding to the input complex radiation signal, calculating a loss function, and then performing reverse parameter correction on the complex feature classification model and the complex feature extraction model according to the loss function.
And repeating the training process until the obtained signal radiation source identification result meets the preset requirement.
As another more preferred embodiment, in the embodiment of the present application, the end-to-end signal radiation source identification model is formed by combining the complex feature extraction model obtained by the training of the complex contrast prediction coding network CCPC and the complex feature classification model obtained by the training of the complex depth residual error network cressnet.
Specifically, based on the above construction scheme of the complex arithmetic operator, the complex contrast predictive coding network CCPC and the complex depth residual error network CresNet are constructed based on the constructed network operators such as the complex convolution, the complex Relu, and the complex linear layer.
As shown in fig. 4, the CCPC network is composed of 5 complex encoding blocks (encoder blocks) and 1 complex GRU layer, where an encoder block is composed of a complex one-dimensional convolution layer + a complex batch normalization layer + a complex ReLU active layer, a complex signal input passes through the 5 encoder blocks to generate a compressed complex Embedding vector, and then passes through the complex GRU layers to generate an electromagnetic signal complex individual feature with context information, that is, a radiation source individual complex feature.
As shown in fig. 5, each complex residual block is formed by linearly connecting a plurality of two-dimensional convolution layers + a plurality of two-dimensional batch normalization layers + a plurality of ReLU active layers + a plurality of two-dimensional convolution layers + a plurality of two-dimensional batch normalization layers, and then a plurality of complex residual blocks are built to obtain a CResNet network. The input of the CResNet is CCPC complex number characteristic, and the recognition result is output, wherein the last layer of the CResNet is a real number full connection layer, so before that, the modulus value of the complex number input is taken, and then the complex number input is sent to the real number full connection layer for classification recognition.
On the basis, as shown in fig. 6, for the received complex radiation signals, the complex radiation signals are directly input into a complex contrast prediction coding network CCPC to obtain individual complex characteristics of the radiation source (CCPC real part characteristics + CCPC imaginary part characteristics), and then the individual complex characteristics of the radiation source are input into a complex depth residual error network creson to perform individual classification of the radiation source, so as to obtain a signal radiation source identification result.
The training process of the end-to-end signal radiation source identification model may refer to the joint training process of the complex feature extraction model and the complex feature classification model.
Based on the signal radiation source identification model, when a complex radiation signal is obtained, the complex radiation signal is directly input into the signal radiation source identification model, the signal radiation source identification model can automatically extract and obtain the individual complex characteristics of the radiation source from the complex radiation signal, and based on the individual complex characteristics of the radiation source, the signal radiation source which radiates the complex radiation signal is identified, and a signal radiation source identification result is output. The application of the signal radiation source identification model greatly improves the identification efficiency of the signal radiation source.
As another optional implementation, after the complex characteristic of the individual radiation source is extracted from the complex radiation source signal, a modulus operation may be performed on the complex characteristic of the individual radiation source, and then a modulus result of the complex characteristic of the individual radiation source is input into a depth residual error network trained in advance, and the signal radiation source that radiates the complex radiation signal is identified by the depth residual error network.
The depth residual error network is the depth residual error network based on real number operation introduced in the above embodiment, and in the training process, the depth residual error network is used for classifying the input individual characteristics of the radiation source, so as to realize individual identification of the radiation source.
In this embodiment, the individual characteristics of the radiation source, on which the depth residual error network classifies the individual radiation sources, are obtained by modulo extraction of the individual complex characteristics of the radiation source, and include more characteristic information than the individual real characteristics of the radiation source, so that a more accurate individual radiation source identification effect can be obtained.
In order to verify the identification performance of the signal radiation source identification method provided by the embodiment of the application, a comparison experiment is also carried out in the embodiment of the application, and the performance of the signal radiation source identification method provided by the embodiment of the application is evaluated by identifying and comparing the signal radiation source identification scheme provided by the embodiment of the application with other signal source individual identification schemes.
The CPU used in the experiment was the Intel Xeon series, the GPU was Tesla T4, and the detailed configuration is shown in Table 1.
Table 1 configuration of the experimental platform
The experimental data are carried on carriers at different empty positions by 10 communication radiation source devices of the same type, communication radiation source intermediate frequency data collected by the same receiver are divided into IQ two paths, the sampling rate is 785KHz, and IQ two paths of signals are sequentially and alternately stored in a 32-bit floating point number format. The training data is 2000 data segments collected by each individual of 10 radiation source individuals, and each segment has about 20000 sampling points; the test data is consistent with the training data acquisition means and the storage mode, and each individual has 500 test data, and the total number is 5000.
In order to verify the effect of the scheme, a VMD + SVM scheme and a real number neural network scheme which are superior to the current effect in the SEI field are respectively set up and used as baseline systems for comparison and analysis. The VMD + SVM scheme firstly carries out 20-order modal decomposition on radiation source data, then splices the mode spectrums and the center frequencies of all the modes into 60-dimensional feature vectors, and sends the 60-dimensional feature vectors into the SVM for training and recognition; the ResNet scheme directly sends radiation source data to a residual error network for end-to-end classification training; according to the CPC + ResNet scheme, firstly, radiation source data are sent into a CPC network for front-end feature extraction, and then are sent into a ResNet network for rear-end classification and identification; according to the CCPC + ResNet scheme, radiation source data are sent to a CCPC network, complex features are extracted, then a modulus value is obtained for the complex features, and the complex features are sent to a real number residual error network for classification and identification; CCPC + CResNet is the scheme proposed by the embodiment of the application, firstly, a plurality of radiation source data are sent to a CCPC network to extract a plurality of characteristics, and the plurality of characteristics are sent to the CResNet network to be identified, and the experimental results of five schemes are shown in Table 2.
TABLE 2 record of experimental results
As can be seen from the experimental results in table 2, the scheme of extracting the individual complex features of the radiation source and classifying the individual complex features of the radiation source, that is, the CCPC + CResNet scheme relatively improves the recognition accuracy by more than 17% compared with the conventional scheme of VMD + SVM; compared with a corresponding real number neural network scheme CPC + ResNet, the identification accuracy is relatively improved by more than 13%. By combining five groups of experimental results, it can be found that the CCPC complex features (i.e., the radiation source individual complex features) provided by the embodiment of the present application can effectively represent the "fingerprint" features of the individual attributes of the radiation source, and the back-end classifier cresint can effectively identify the radiation source individual based on the features.
Exemplary devices
Correspondingly, the embodiment of the present application further provides a signal radiation source identification device, as shown in fig. 7, the device includes:
a feature extraction unit 100, configured to extract complex features of an individual radiation source from acquired complex radiation signals, where the complex radiation signals are formed by combining real part signals and imaginary part signals;
a feature classification unit 110, configured to identify a signal radiation source that radiates the complex radiation signal based on the individual complex features of the radiation sources.
As an alternative embodiment, extracting the individual complex features of the radiation source from the acquired complex radiation signals includes:
inputting the acquired complex radiation signals into a pre-trained complex feature extraction model to obtain individual complex features of the radiation source;
the complex characteristic extraction model is obtained by extracting complex characteristics of the signal radiation source individuals from complex radiation signals radiated by the signal radiation source individuals and training.
As an optional implementation manner, the complex feature extraction model is obtained by performing operator adjustment on a preset contrast predictive coding network according to a complex operation rule.
As an optional implementation manner, performing operator adjustment on the preset contrast predictive coding network according to a complex operation rule, including:
based on a complex operation rule, constructing a complex operation operator by using a preset real operation operator in a contrast predictive coding network;
and replacing the real number arithmetic operator in the contrast predictive coding network with a corresponding complex number arithmetic operator.
As an optional implementation, identifying a signal radiation source that radiates the complex radiation signal based on the individual complex features of the radiation source includes:
inputting the individual complex features of the radiation source into a pre-trained complex feature classification model, and identifying a signal radiation source for radiating the complex radiation signal;
the complex characteristic classification model is obtained by taking the individual complex characteristics of the radiation source as input and performing signal radiation source classification and identification training.
As an optional implementation manner, the complex feature classification model is obtained by performing operator adjustment on a preset depth residual error network according to a complex operation rule.
As an optional implementation manner, performing operator adjustment on the preset depth residual error network according to a complex operation rule, including:
constructing a complex operator by utilizing a real operator in a preset depth residual error network based on a complex operation rule;
and replacing the real number arithmetic operator in the depth residual error network with a corresponding complex number arithmetic operator.
As an optional implementation, identifying a signal radiation source that radiates the complex radiation signal based on the individual complex features of the radiation source includes:
performing modular operation on the individual complex features of the radiation source;
inputting the modular operation result of the individual complex features of the radiation source into a depth residual error network trained in advance, and identifying a signal radiation source for radiating the complex radiation signal;
the deep residual error network is obtained by performing individual classification, identification and training of the radiation source based on input individual characteristics of the radiation source.
As an optional implementation manner, extracting complex features of an individual radiation source from the acquired complex radiation signal, and identifying a signal radiation source radiating the complex radiation signal based on the complex features of the individual radiation source comprises:
inputting the acquired complex radiation signals into a pre-trained signal radiation source identification model so that the signal radiation source identification model extracts individual complex characteristics of a radiation source from the acquired complex radiation signals, and identifying the signal radiation source irradiating the complex radiation signals based on the individual complex characteristics of the radiation source;
the signal radiation source identification model is constructed by a complex contrast prediction coding network and a complex depth residual error network, the complex contrast prediction coding network is constructed based on a complex arithmetic operator, and the complex depth residual error network is constructed based on the complex arithmetic operator.
The signal radiation source identification device provided by the embodiment belongs to the same application concept as the signal radiation source identification method provided by the embodiment of the present application, can execute the signal radiation source identification method provided by any embodiment of the present application, and has corresponding functional modules and beneficial effects for executing the signal radiation source identification method. For details of the signal radiation source identification method provided in the foregoing embodiments of the present application, reference may be made to specific processing contents of the signal radiation source identification method provided in the foregoing embodiments, and details are not described herein again.
Exemplary electronic device
Another embodiment of the present application further provides a signal radiation source identification device, as shown in fig. 8, including:
a memory 200 and a processor 210;
wherein, the memory 200 is connected to the processor 210 for storing programs;
the processor 210 is configured to implement the method for identifying a signal radiation source disclosed in any of the above embodiments by running the program stored in the memory 200.
Specifically, the signal radiation source identification device may further include: a bus, a communication interface 220, an input device 230, and an output device 240.
The processor 210, the memory 200, the communication interface 220, the input device 230, and the output device 240 are connected to each other through a bus. Wherein:
a bus may include a path that transfers information between components of a computer system.
The processor 210 may be a general-purpose processor, such as a general-purpose Central Processing Unit (CPU), microprocessor, etc., an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs in accordance with the present invention. But may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components.
The processor 210 may include a main processor and may also include a baseband chip, modem, and the like.
The memory 200 stores programs for executing the technical solution of the present invention, and may also store an operating system and other key services. In particular, the program may include program code including computer operating instructions. More specifically, memory 200 may include a read-only memory (ROM), other types of static storage devices that may store static information and instructions, a Random Access Memory (RAM), other types of dynamic storage devices that may store information and instructions, a disk storage, a flash, and so forth.
The input device 230 may include a means for receiving data and information input by a user, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor, among others.
The processor 210 executes the program stored in the memory 200 and invokes other devices, which can be used to implement the steps of any one of the signal radiation source identification methods provided by the above-mentioned embodiments of the present application.
Exemplary computer program product and storage Medium
In addition to the above-described methods and apparatus, embodiments of the present application may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the signal radiation source identification method described in the "exemplary methods" section above of this specification.
The computer program product may be written with program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, an embodiment of the present application may also be a storage medium having stored thereon a computer program, which is executed by a processor to perform the steps in the signal radiation source identification method described in the above-mentioned "exemplary method" section of this specification.
While, for purposes of simplicity of explanation, the foregoing method embodiments have been described as a series of acts or combination of acts, it will be appreciated by those skilled in the art that the present application is not limited by the order of acts or acts described, as some steps may occur in other orders or concurrently with other steps in accordance with the application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The steps in the method of each embodiment of the present application may be sequentially adjusted, combined, and deleted according to actual needs, and technical features described in each embodiment may be replaced or combined.
The modules and sub-modules in the device and the terminal of the embodiment of the application can be combined, divided and deleted according to actual needs.
In the several embodiments provided in the present application, it should be understood that the disclosed terminal, apparatus and method may be implemented in other manners. For example, the above-described terminal embodiments are merely illustrative, and for example, the division of a module or a sub-module is only one logical division, and there may be other divisions when the terminal is actually implemented, for example, a plurality of sub-modules or modules may be combined or integrated into another module, or some features may be omitted or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or modules, and may be in an electrical, mechanical or other form.
The modules or sub-modules described as separate components may or may not be physically separate, and the components described as modules or sub-modules may or may not be physical modules or sub-modules, may be located in one place, or may be distributed on a plurality of network modules or sub-modules. Some or all of the modules or sub-modules can be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, each functional module or sub-module in the embodiments of the present application may be integrated into one processing module, or each module or sub-module may exist alone physically, or two or more modules or sub-modules may be integrated into one module. The integrated modules or sub-modules may be implemented in the form of hardware, or may be implemented in the form of software functional modules or sub-modules.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software unit executed by a processor, or in a combination of the two. The software cells may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (12)
1. A signal radiation source identification method, comprising:
extracting individual complex characteristics of a radiation source from the acquired complex radiation signals, wherein the complex radiation signals are formed by combining real part signals and imaginary part signals;
and identifying a signal radiation source irradiating the complex radiation signal based on the individual complex characteristics of the radiation sources.
2. The method of claim 1, wherein extracting individual complex features of the radiation source from the acquired complex radiation signals comprises:
inputting the acquired complex radiation signals into a pre-trained complex feature extraction model to obtain individual complex features of the radiation source;
the complex feature extraction model is obtained by extracting complex features of the signal radiation source individuals from complex radiation signals radiated by the signal radiation source individuals and training.
3. The method of claim 2, wherein the complex feature extraction model is obtained by performing operator adjustment on a predetermined contrast predictive coding network according to a complex operation rule.
4. The method of claim 3, wherein performing operator adjustment on the predetermined contrast predictive coding network according to a complex operation rule comprises:
based on a complex operation rule, constructing a complex operation operator by using a preset real operation operator in a contrast predictive coding network;
and replacing the real number arithmetic operator in the contrast predictive coding network with a corresponding complex number arithmetic operator.
5. The method of claim 1, wherein identifying a signal radiation source that radiates the complex radiation signal based on the individual complex signature of the radiation source comprises:
inputting the individual complex features of the radiation source into a pre-trained complex feature classification model, and identifying a signal radiation source irradiating the complex radiation signal;
the complex characteristic classification model is obtained by taking the complex characteristic of the radiation source individual as input and performing signal radiation source classification and identification training.
6. The method of claim 5, wherein the complex feature classification model is obtained by performing operator adjustment on a preset depth residual error network according to a complex operation rule.
7. The method of claim 6, wherein performing operator adjustment on the preset depth residual error network according to a complex operation rule comprises:
constructing a complex operator by utilizing a real operator in a preset depth residual error network based on a complex operation rule;
and replacing the real number arithmetic operator in the depth residual error network with a corresponding complex number arithmetic operator.
8. The method of claim 1, wherein identifying a signal radiation source that radiates the complex radiation signal based on the individual complex signature of the radiation source comprises:
performing modular operation on the individual complex features of the radiation source;
inputting the modular operation result of the individual complex characteristics of the radiation source into a depth residual error network trained in advance, and identifying a signal radiation source for radiating the complex radiation signal;
the deep residual error network is obtained by performing individual classification, identification and training of the radiation source based on input individual characteristics of the radiation source.
9. The method of claim 1, wherein extracting individual complex features of the radiation source from the acquired complex radiation signal, and identifying the signal radiation source radiating the complex radiation signal based on the individual complex features of the radiation source comprises:
inputting the acquired complex radiation signals into a pre-trained signal radiation source identification model so that the signal radiation source identification model extracts individual complex characteristics of a radiation source from the acquired complex radiation signals, and identifying the signal radiation source irradiating the complex radiation signals based on the individual complex characteristics of the radiation source;
the signal radiation source identification model is constructed by a complex contrast prediction coding network and a complex depth residual error network, the complex contrast prediction coding network is constructed based on a complex arithmetic operator, and the complex depth residual error network is constructed based on the complex arithmetic operator.
10. A signal radiation source identification device, comprising:
the characteristic extraction unit is used for extracting and obtaining the individual complex characteristics of the radiation source from the acquired complex radiation signals, wherein the complex radiation signals are formed by combining real part signals and imaginary part signals;
and the characteristic classification unit is used for identifying a signal radiation source for radiating the complex radiation signal based on the individual complex characteristics of the radiation source.
11. A signal radiation source identification device, comprising:
a memory and a processor;
wherein the memory is connected with the processor and used for storing programs;
the processor is configured to implement the signal radiation source identification method according to any one of claims 1 to 9 by executing a program in the memory.
12. A storage medium, characterized in that the storage medium has stored thereon a computer program which, when being executed by a processor, implements the signal radiation source identification method according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210521459.0A CN115062642A (en) | 2022-05-13 | 2022-05-13 | Signal radiation source identification method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210521459.0A CN115062642A (en) | 2022-05-13 | 2022-05-13 | Signal radiation source identification method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115062642A true CN115062642A (en) | 2022-09-16 |
Family
ID=83198192
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210521459.0A Pending CN115062642A (en) | 2022-05-13 | 2022-05-13 | Signal radiation source identification method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115062642A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115236606A (en) * | 2022-09-23 | 2022-10-25 | 中国人民解放军战略支援部队航天工程大学 | Radar signal feature extraction method and complex number field convolution network architecture |
| CN115809426A (en) * | 2023-02-03 | 2023-03-17 | 西安睿奥电磁环境科技有限公司 | Radiation source individual identification method and system |
-
2022
- 2022-05-13 CN CN202210521459.0A patent/CN115062642A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115236606A (en) * | 2022-09-23 | 2022-10-25 | 中国人民解放军战略支援部队航天工程大学 | Radar signal feature extraction method and complex number field convolution network architecture |
| CN115809426A (en) * | 2023-02-03 | 2023-03-17 | 西安睿奥电磁环境科技有限公司 | Radiation source individual identification method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shanthamallu et al. | A brief survey of machine learning methods and their sensor and IoT applications | |
| CN113707235B (en) | Drug micromolecule property prediction method, device and equipment based on self-supervision learning | |
| Wei et al. | Road structure refined CNN for road extraction in aerial image | |
| CN103678702B (en) | Video duplicate removal method and device | |
| CN115062642A (en) | Signal radiation source identification method, device, equipment and storage medium | |
| US20140181171A1 (en) | Method and system for fast tensor-vector multiplication | |
| CN110752028A (en) | Image processing method, device, equipment and storage medium | |
| CN101853392A (en) | Remote sensing hyperspectral image band selection method based on conditional mutual information | |
| CN115953665B (en) | Target detection method, device, equipment and storage medium | |
| CN110059288B (en) | System and method for obtaining an optimal mother wavelet for facilitating a machine learning task | |
| CN111126396A (en) | Image recognition method and device, computer equipment and storage medium | |
| CN111553215A (en) | Personnel association method and device, and graph convolution network training method and device | |
| CN112395941B (en) | Radio signal radiation source individual identification method, computer device and storage medium | |
| CN110929806A (en) | Picture processing method and device based on artificial intelligence and electronic equipment | |
| Nie et al. | Online multiview deep forest for remote sensing image classification via data fusion | |
| Concolato et al. | Data science: A new paradigm in the age of big-data science and analytics | |
| CN111222399B (en) | Method and device for identifying object identification information in image and storage medium | |
| CN106503112A (en) | Video retrieval method and device | |
| Ha et al. | Selective deep convolutional neural network for low cost distorted image classification | |
| CN113591892A (en) | Training data processing method and device | |
| US11886445B2 (en) | Classification engineering using regional locality-sensitive hashing (LSH) searches | |
| US10726090B1 (en) | Per-user accuracy measure for social network based geocoding algorithms | |
| CN107607723A (en) | A kind of protein-protein interaction assay method based on accidental projection Ensemble classifier | |
| Fushimi et al. | Accelerating Greedy K-Medoids Clustering Algorithm with Distance by Pivot Generation | |
| Fan et al. | Robust visual tracking via bag of superpixels |
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 |