CN114679359B - Multicarrier waveform identification method based on convolutional neural network - Google Patents
Multicarrier waveform identification method based on convolutional neural network Download PDFInfo
- Publication number
- CN114679359B CN114679359B CN202210368127.3A CN202210368127A CN114679359B CN 114679359 B CN114679359 B CN 114679359B CN 202210368127 A CN202210368127 A CN 202210368127A CN 114679359 B CN114679359 B CN 114679359B
- Authority
- CN
- China
- Prior art keywords
- waveform
- neural network
- ofdm
- convolutional neural
- waveforms
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 71
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000009466 transformation Effects 0.000 claims abstract description 13
- 238000004891 communication Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 4
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 13
- 238000005516 engineering process Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 6
- 239000000969 carrier Substances 0.000 description 5
- 238000010606 normalization Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 125000004122 cyclic group Chemical group 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000007493 shaping process Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000003595 spectral effect Effects 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 239000000654 additive Substances 0.000 description 2
- 230000000996 additive effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 108010076504 Protein Sorting Signals Proteins 0.000 description 1
- 238000010420 art technique Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/0012—Modulated-carrier systems arrangements for identifying the type of modulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/26—Systems using multi-frequency codes
- H04L27/2601—Multicarrier modulation systems
- H04L27/2602—Signal structure
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/26—Systems using multi-frequency codes
- H04L27/2601—Multicarrier modulation systems
- H04L27/2647—Arrangements specific to the receiver only
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Noise Elimination (AREA)
Abstract
Aiming at the limitation of the prior art, the invention provides a multi-carrier wave form identification method based on a convolutional neural network, and aiming at six novel multi-carrier wave forms of OFDM, W-OFDM, F-OFDM, UFMC, FBMC and GFDM, a convolutional neural network model is applied, and the characteristics of the wave forms are extracted by combining FSST, DFT and wavelet transformation, so that the identification accuracy is further improved, and the scheme still has higher identification accuracy at the time of low signal to noise ratio.
Description
Technical Field
The invention relates to the technical field of waveform identification in wireless communication, in particular to a multicarrier waveform identification method based on a convolutional neural network.
Background
In the past, OFDM technology is widely used in the conventional long term evolution system (Long Term Evolution, LTE), and three application scenarios of 5G enhance Mobile BroadBand (eMBB), massive Machine-Type-Communications (emtc), ultra-Reliable Low-latency Communications (ullc) and transmission waveforms have raised higher requirements. In recent years, in order to meet the higher demands of communication systems, various new multi-carrier waveforms have been proposed, such as: W-OFDM, F-OFDM, FBMC, UFMC, GFDM, etc. However, there is no multi-carrier waveform available for all application scenarios, so coexistence of multiple multi-carrier waveforms is an inevitable trend in future communication scenarios. Therefore, the receiver needs to detect and identify the received multi-carrier waveforms, demodulate and further analyze and process different multi-carrier waveforms according to the identification result.
At present, the single carrier signal and OFDM waveform signal identification are studied more, wherein the related signal characteristics are mainly divided into time domain characteristics and transformation domain characteristics, and mainly divided into the following steps: instantaneous information characteristics (e.g., amplitude, frequency, phase, etc.), signal higher order cumulants, time-frequency analysis (e.g., wavelet transform, short-time fourier transform, fuzzy function, etc.), constellation characteristics, spectral analysis characteristics (e.g., spectral correlation, higher order spectra, etc.). However, there are few related studies on feature extraction and identification of the 5G new multi-carrier waveform.
Chinese invention patent publication date 2020-11-24: in a novel multi-carrier recognition method based on a back propagation neural network, firstly, three baseband multi-carrier signals with various amplitudes are generated, and the multi-carrier signals are sampled; carrying out power normalization processing and linear normalization processing on the signal sequence obtained by sampling; constructing and training a back propagation neural network; and carrying out the two-step normalization processing on the signal to be identified, inputting the signal to be identified into a network, and judging the type of the multi-carrier signal according to the output value. It alleges that by the method, the identification of three novel multi-carrier signals can be quickly and accurately realized with low complexity; and the normalization processing and correct recognition of the non-standardized amplitude signals under the low signal-to-noise ratio can be realized, and the generalization capability of the recognition network is improved. However, the prior art has certain limitations.
Disclosure of Invention
Aiming at the limitation of the prior art, the invention provides a multicarrier waveform identification method based on a convolutional neural network, which adopts the following technical scheme:
a multi-carrier waveform identification method based on convolutional neural network comprises the following steps:
s1, acquiring a received signal of a waveform to be identified;
s2, performing two classification on the received signal according to a first type waveform and a second type waveform by using a preset first convolutional neural network model; wherein the first type of waveform comprises three waveforms of OFDM, W-OFDM and F-OFDM; the second type of waveform comprises three waveforms of FBMC, GFDM and UFMC;
s3, performing discrete Fourier transform and Haar wavelet transform on the waveform signals divided into the first type of waveforms in the step S2, and inputting a preset second convolution neural network model to perform three classification to obtain classification results of the received signals on three waveforms of OFDM, W-OFDM and F-OFDM;
and S4, carrying out Fourier transform on the waveform signals divided into the second type of waveforms in the step S2, inputting a preset third convolutional neural network model for three classification, and obtaining classification results of the received signals on three waveforms of FBMC, GFDM and UFMC.
Compared with the prior art, the invention provides a CNN-based multi-carrier waveform identification scheme aiming at six novel multi-carrier waveforms of OFDM, W-OFDM, F-OFDM, UFMC, FBMC and GFDM, and the characteristics of the waveforms are extracted by combining FSST, DFT and wavelet transformation by applying a convolutional neural network model, so that the identification accuracy is further improved, and the scheme still has higher identification accuracy at low signal-to-noise ratio
As a preferred embodiment, the normalization processing is performed on the received signal before the step S2 is performed.
As a preferable scheme, the first convolutional neural network model, the second convolutional neural network model and the third convolutional neural network model are respectively composed of two convolutional layers and two full-connection layers in sequence.
Further, 128 filters with the size of 1×3 are adopted in the first convolution layer, and a linear rectifying unit is used as an active layer.
Further, 64 filters of size 1×3 and an active layer ReLU are used in the second convolution layer.
Further, the first fully-connected layer has 128 neurons.
Further, the second full connection layer is output via a SoftMax.
The invention also includes the following:
the multi-carrier waveform recognition system based on the convolutional neural network is characterized by comprising a signal receiving module, a first classifying module, a second classifying module and a third classifying module; the first classification module is respectively connected with the signal receiving module, the second classification module and the third classification module; wherein:
the signal receiving module is used for obtaining a received signal of the waveform to be identified;
the first classification module is used for performing two classifications on the received signal according to a first type waveform and a second type waveform by using a preset first convolutional neural network model; wherein the first type of waveform comprises three waveforms of OFDM, W-OFDM and F-OFDM; the second type of waveform comprises three waveforms of FBMC, GFDM and UFMC;
the second classification module is used for performing discrete Fourier transform and Haar wavelet transform on the waveform signals divided into the first type of waveforms in the step S2, inputting a preset second convolution neural network model for three classification, and obtaining classification results of the received signals on three waveforms of OFDM, W-OFDM and F-OFDM;
the third classification module is configured to perform fourier transform on the waveform signal divided into the second type waveform in the step S2, input a preset third convolutional neural network model to perform three classification, and obtain classification results of the received signal on three waveforms of FBMC, GFDM and UFMC.
A storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a convolutional neural network-based multicarrier waveform identification method as described above.
A communication device comprising a storage medium, a processor and a computer program stored in the storage medium and executable by the processor, which when executed by the processor performs the steps of a convolutional neural network based multi-carrier waveform identification method as described above.
Drawings
Fig. 1 is an OFDM modulation block diagram;
FIG. 2 is a block diagram of F-OFDM\UFMC modulation;
FIG. 3 is a block diagram of FBMC modulation;
fig. 4 is a GFDM modulation block diagram;
fig. 5 is a schematic step diagram of a multi-carrier waveform recognition method based on convolutional neural network according to an embodiment of the present invention;
fig. 6 is a schematic diagram of a CNN structure used in an embodiment of the present invention;
FIG. 7 is a graph showing the recognition accuracy of each waveform in a simulation experiment according to an embodiment of the present invention;
FIG. 8 is a graph showing the average recognition accuracy of waveforms in a simulation experiment according to an embodiment of the present invention;
fig. 9 is a schematic diagram of a multicarrier waveform identification system based on convolutional neural network according to an embodiment of the present invention.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the present patent;
it should be understood that the described embodiments are merely some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the embodiments of the present application, are within the scope of the embodiments of the present application.
The terminology used in the embodiments of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the embodiments of the application. As used in this application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any or all possible combinations of one or more of the associated listed items.
When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present application as detailed in the accompanying claims. In the description of this application, it should be understood that the terms "first," "second," "third," and the like are used merely to distinguish between similar objects and are not necessarily used to describe a particular order or sequence, nor should they be construed to indicate or imply relative importance. The specific meaning of the terms in this application will be understood by those of ordinary skill in the art as the case may be.
Furthermore, in the description of the present application, unless otherwise indicated, "a plurality" means two or more. "and/or", describes an association relationship of an association object, and indicates that there may be three relationships, for example, a and/or B, and may indicate: a exists alone, A and B exist together, and B exists alone. The character "/" generally indicates that the context-dependent object is an "or" relationship. The invention is further illustrated in the following figures and examples.
In order to solve the limitations of the prior art, the present embodiment provides a technical solution, and the technical solution of the present invention is further described below with reference to the drawings and the embodiments.
Example 1
This embodiment will be described in principle with reference to fig. 1, where a modulation block diagram of ofdm is shown. OFDM is the basis for most multicarrier waveforms, and W-OFDM, F-OFDM, UFMC, FBMC, GFDM and other multicarrier waveforms have been developed from OFDM. According to FIG. 1, N c The modulation symbols are mapped to N by an inverse Fourier transform (Inverse Fourier Transform, IFT) c On the sub-carriers, it can be expressed in particular as
Wherein x is 1 (k) The subscript "1" of (c) represents the first of six multi-carrier waveforms, k=1, 2, … represents a time sequence,
s (m, n) represents transmission data on an nth subcarrier of an mth OFDM signal in imaginary units; p is p 1 (k) Is a prototype filter, typically a rectangular window; />
Is the length of OFDM symbol, where N c N is the number of subcarriers cp The cyclic prefix is to reduce inter-symbol interference, which is the length of the cyclic prefix. The OFDM waveform has slower frequency spectrum attenuation and larger out-of-band leakage, and the main reason is that the OFDM is usedCausing signal discontinuities at symbol boundaries. To improve the spectral performance of OFDM, a non-rectangular pulse p with smooth edges may be used in equation (1) 1 (k) Instead of rectangular windows, the waveform thus generated is called W-OFDM.
If x (k) is a baseband signal, the received signal at the receiving end can be expressed as
Wherein l=0, …, L ch -1, wherein L ch Is the number of multipath, h l (. Cndot.) and eta l Channel parameters and delay spread of the first multipath respectively; n (k) is gaussian additive white noise (Additive White Gaussian Noise, AWGN), and Δf is the carrier frequency offset.
The modulation block diagram for reference to figure 2,F-OFDM is shown. To reduce the higher out-of-band leakage of OFDM, each sub-band of the F-OFDM waveform will pass through a band-limited filter. N of F-OFDM symbol c Sub-carriers are divided into R 1 Each sub-band having Q 1 Subcarriers, i.e. N c ＝R 1 ×Q 1 . Then F-OFDM can be expressed as
Where s (m, r, q) is transmission data on the q-th subcarrier of the m-th symbol on the r-th subband. P is p 2 (k) Is of length L 1 Is a filter of the sine (r) of the filter,
is the length of the F-OFDM symbol.
Referring to fig. 3, a modulation block diagram of fbmc is shown. To further reduce out-of-band leakage, achieving higher spectral efficiency, FBMC implements a prototype filter for K consecutive multi-carrier symbols simultaneously in the time domain, which improves its frequency-location characteristics, but between adjacent sub-carriersInterference is introduced. To overcome this drawback, FBMC employs offset quadrature amplitude modulation (Offset Quadrature Amplitude Modulation, OQAM) technique, which staggers the real and imaginary parts of the OQAM symbols by N c The offset of/2 transmits data separately, the FBMC symbol can be expressed as
Wherein s is R (m, n) and s I (m, n) represent real and imaginary parts, p, respectively, of a complex-valued QAM signal 3 (k) Is its prototype filter.
Referring to fig. 2, a modulation block diagram of ufmc is shown. In order to avoid inter-symbol interference, UFMC is filtered for each sub-band, unlike F-OFDM, where a shorter prototype filter is used for UFMC. N of UFMC waveform c Sub-carriers are divided into R 2 A number of sub-bands, each sub-band containing Q 2 The UFMC waveform signal may be represented as
Where s (m, r, q) represents data transmitted by an mth symbol on a qth subcarrier on an mth subband; p is p 4 (k) Is of length L 2 Is a Chebyshev filter;
is the symbol length of the UFMC. />
Referring to fig. 4, a modulation block diagram of the gfdm is shown. In order to reduce the filter length in FBMC, GFDM replaces the linear pulse shaping filter in FBMC with a cyclic pulse shaping filter, which ensures that the length of the waveform symbols before and after filtering remains unchanged. One GFDM symbol period contains B symbol blocks, each block being formed by a MN c Data symbols of complex value, where M is the number of sub-symbols, N c The number of subcarriers per sub-symbol. Thus, GFDM symbols can be expressed as
Where s (m, n, b) represents transmission data on an nth subcarrier of an mth symbol of an mth block. P is p 5 (k) For a raised cosine shaping filter (RRC),
is the length of the GFDM symbol.
Among six waveforms to be identified in the embodiment, three waveforms of OFDM, W-OFDM and F-OFDM are similar, the W-OFDM is obtained by windowing on the basis of OFDM, and the F-OFDM is obtained by filtering sub-carriers on the basis of OFDM. The FBMC, UFMC, GFDM waveforms are relatively complex, and the waveform characteristics are also obvious, for example, compared with other multi-carrier waveforms, the FBMC has smaller out-of-band leakage of the frequency spectrum, the UFMC filters for the sub-bands, and the GFDM has certain cycle characteristics. Therefore, the CNN is used for carrying out two classifications on six waveforms, one is three simpler waveforms of OFDM, W-OFDM and F-OFDM, and the other is FBMC, UFMC, GFDM three more complex waveforms. And then carrying out different processing and transformation on the two types of waveforms, and respectively carrying out three classification on the two types of waveforms, thereby accurately identifying six types of waveforms.
Therefore, referring to fig. 5, a multi-carrier waveform identification method based on convolutional neural network includes the following steps:
s1, acquiring a received signal of a waveform to be identified;
s2, performing two classification on the received signal according to a first type waveform and a second type waveform by using a preset first convolutional neural network model CNN# 1; wherein the first type of waveform comprises three waveforms of OFDM, W-OFDM and F-OFDM; the second type of waveform comprises three waveforms of FBMC, GFDM and UFMC;
s3, performing discrete Fourier transform and Haar wavelet transform on the waveform signals divided into the first type of waveforms in the step S2, and inputting a preset second convolutional neural network model CNN# 2 to perform three classification to obtain classification results of the received signals on three waveforms of OFDM, W-OFDM and F-OFDM;
and S4, carrying out Fourier transform on the waveform signals divided into the second type of waveforms in the step S2, inputting a preset third convolutional neural network model CNN# 3 for three classification, and obtaining classification results of the received signals on the three waveforms of FBMC, GFDM and UFMC.
Compared with the prior art, the invention provides a CNN-based multi-carrier waveform identification scheme aiming at six novel multi-carrier waveforms of OFDM, W-OFDM, F-OFDM, UFMC, FBMC and GFDM, and the characteristics of the waveforms are extracted by combining FSST, DFT and wavelet transformation by applying a convolutional neural network model, so that the identification accuracy is further improved, and the scheme still has higher identification accuracy at low signal-to-noise ratio.
As a preferred embodiment, the received signal is normalized before the step S2 is performed.
Specific: let R (n) be the square of the modulus obtained after DFT conversion of the multicarrier signal x (k), i.e
Wherein n=0, 1, …, N 6 -1，N 6 Is the length of the signal x (k). The wavelet transformation result of R (n) is
Wherein ψ (p) is the mother wavelet, and a and n are the scale and transform factors, respectively. Since the effect of the recognition scheme is independent of the choice of the type of parent wavelet, the scheme selects a Haar wavelet of lower complexity as the parent wavelet, i.e
For three waveforms in class 2, FSST transformation is carried out on the waveforms, characteristics in a time-frequency domain are extracted, then a third CNN is input for three classification, and three waveforms of FBMC, UFMC and GFDM can be respectively identified. The multicarrier signal f (t) may be decomposed into I parts, i.e
Wherein A is i (t) and phi i (t) represents the amplitude and frequency of the i-th part. The result of short-time Fourier transform (short-time Fourier transform, STFT) of f (t) is
V f (η，t)＝∫ R f(τ)g(τ-t)e -2iπη(τ-t) dτ，(11)
Based on the results of the STFT, FSST converts the variable (η, t) in the STFT into
Obtaining
Where g (0) is the value of the sliding window g (t) at time t=0, δ (·) is the dirac function,
is defined as follows
Wherein the function of the operator Re (·) is to take the real part of a complex number.
As a preferred embodiment, referring to fig. 6, the first convolutional neural network model, the second convolutional neural network model and the third convolutional neural network model are respectively composed of two convolutional layers and two fully-connected layers in sequence.
Specifically, the convolution layer can extract the corresponding mapping relation from the input data through the neurons of the layer. The full connection layer at the CNN terminal can perform nonlinear combination on the CNN terminal according to the learned characteristics and output related classification prediction, namely a classification result.
The three CNNs used in this embodiment are substantially identical in structure and model, except for the input data and output classifications. For the first CNN, the input data is waveform signal data obtained by normalizing six kinds of multi-carriers, and the input data is output as two kinds of waveforms. For the second CNN, the input data is waveform data after three waveforms are subjected to DFT and Haar wavelet transformation respectively, and the input data is output as classification results of three waveforms of OFDM, W-OFDM and F-OFDM. For the third CNN, the input data is waveform data of three waveforms after FSST conversion, and the classification result of the three waveforms is output as FBMC, UFMC, GFDM.
Further, 128 filters with the size of 1×3 are adopted in the first convolution layer, and a linear rectifying unit is used as an active layer.
Further, 64 filters of size 1×3 and an active layer ReLU are used in the second convolution layer.
Further, the first fully-connected layer has 128 neurons.
Further, the second full connection layer is output via a SoftMax.
More specifically, please refer to fig. 7, which is a diagram of the relationship between the recognition accuracy and the signal-to-noise ratio of the proposed multi-carrier waveform recognition scheme. As can be seen from the figure, as the signal-to-noise ratio is increased, the recognition accuracy of the six multi-carrier waveforms is improved correspondingly. The recognition accuracy of the three complex waveforms of GFDM, FBMC, UFMC is higher than that of the three simple waveforms of OFDM, W-OFDM and F-OFDM. The recognition accuracy of the multi-carrier waveform recognition scheme proposed by the technology is obviously better than that proposed by the prior art [ S.Duan, K.Chen, X.Yu, and M.Qian, "Automatic multicarrier waveform classification via PCA and convolutional neural networks," IEEE Access, vol.6, pp.51-51 373, sept.2018] (labeled [12] in the figure) and the prior art [ K.Zerhouni, E.M.Amhoud, and M.Chafii, "Filtered multicarrier waveforms classification: A deep learning-based application", IEEE Access, vol.9, pp.69 426-69 438, may 2021] (labeled [13] in the figure), especially in the environment of low signal-to-noise ratio. For example, when snr= -4dB, the recognition accuracy of the scheme proposed by the present technology for OFDM is 0.88, whereas in the above two prior arts, the recognition accuracy of OFDM is only 0.52 and 0.54. For FBMC, the recognition accuracy of the proposed solution is 0.94 when SNR = -4dB, whereas the accuracy of the two prior art techniques described above is only 0.68 and 0.78 for FBMC. Similarly, when snr= -4dB, the recognition accuracy of the scheme proposed by the present technology for F-OFDM is 0.93, which is higher than 0.78 in one of the prior arts.
Referring to fig. 8, a graph of average recognition accuracy and signal-to-noise ratio for the proposed multi-carrier waveform recognition scheme is shown. It can be seen from the graph that as the signal-to-noise ratio increases, the average recognition accuracy of the six multi-carrier waveforms increases gradually. The average recognition accuracy of the proposal provided by the technology is better than that of the recognition methods proposed by the two prior arts. For example, when snr= -4dB, the average recognition accuracy of the proposed solution of the present technology is 0.94, and the average recognition accuracy of the two recognition methods proposed by the above-mentioned prior art is 0.76 and 0.82. In the recognition scheme provided by the technology, the SNR when the average recognition accuracy reaches 1 is 0dB, and in the two prior arts, the SNR required for the recognition accuracy to reach 1 is 6dB and 14dB respectively.
The design proposal provided by the technology combines DFT, haar wavelet transformation, FSST and CNN, and can accurately identify six novel multi-carrier waveforms of OFDM, W-OFDM, F-OFDM, GFDM, FBMC and UFMC. First, six waveforms are divided into two types by using a 4-layer CNN, wherein one type is a waveform which is simply windowed and filtered based on OFDM, and the waveforms comprise OFDM, W-OFDM and F-OFDM. Another class is the more complex waveforms, including GFDM, FBMC and UFMC. For the type 1 waveform, DFT and Haar wavelet transformation are carried out to extract the edge characteristics of the waveform, and then three classification is carried out by 4 layers of CNNs, so that three waveforms of OFDM, W-OFDM and F-OFDM are respectively identified. For the type 2 waveform, FSST transformation is carried out on the waveform, and three classification is carried out on the waveform by 4 layers of CNNs, so that three waveforms of GFDM, FBMC and UFMC are respectively identified. The simulation result verifies the effectiveness of the scheme, is superior to the existing multi-carrier recognition scheme, and has stronger engineering implementation significance.
Example 2
Referring to fig. 9, a multi-carrier waveform recognition system based on convolutional neural network is characterized by comprising a signal receiving module 1, a first classifying module 2, a second classifying module 3 and a third classifying module 4; the first classification module 2 is respectively connected with the signal receiving module 1, the second classification module 3 and the third classification module 4; wherein:
the signal receiving module 1 is used for acquiring a received signal of a waveform to be identified;
the first classification module 2 is configured to perform two classifications on the received signal according to a first type waveform and a second type waveform by using a preset first convolutional neural network model; wherein the first type of waveform comprises three waveforms of OFDM, W-OFDM and F-OFDM; the second type of waveform comprises three waveforms of FBMC, GFDM and UFMC;
the second classification module 3 is configured to perform discrete fourier transform and Haar wavelet transform on the waveform signal divided into the first type of waveform in the step S2, and input a preset second convolutional neural network model to perform three classifications, so as to obtain classification results of the received signal on three waveforms of OFDM, W-OFDM and F-OFDM;
the third classification module 4 is configured to perform fourier transform on the waveform signal divided into the second type of waveform in the step S2, input a preset third convolutional neural network model to perform three classification, and obtain classification results of the received signal on three waveforms of FBMC, GFDM and UFMC.
Example 3
A storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a convolutional neural network-based multicarrier waveform identification method as described in embodiment 1.
Example 4
A communication device comprising a storage medium, a processor and a computer program stored in the storage medium and executable by the processor, which when executed by the processor implements the steps of the convolutional neural network-based multicarrier waveform identification method of embodiment 1.
It is to be understood that the above examples of the present invention are provided by way of illustration only and not by way of limitation of the embodiments of the present invention. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. Any modification, equivalent replacement, improvement, etc. which come within the spirit and principles of the invention are desired to be protected by the following claims.
Claims (10)
1. The multi-carrier waveform identification method based on the convolutional neural network is characterized by comprising the following steps of:
s1, acquiring a received signal of a waveform to be identified;
s2, performing two classification on the received signal according to a first type waveform and a second type waveform by using a preset first convolutional neural network model; wherein the first type of waveform comprises three waveforms of OFDM, W-OFDM and F-OFDM; the second type of waveform comprises three waveforms of FBMC, GFDM and UFMC;
s3, performing discrete Fourier transform and Haar wavelet transform on the waveform signals divided into the first type of waveforms in the step S2, and inputting a preset second convolution neural network model to perform three classification to obtain classification results of the received signals on three waveforms of OFDM, W-OFDM and F-OFDM;
s4, performing FSST transformation on the waveform signals divided into the second type of waveforms in the step S2, inputting a preset third convolutional neural network model for three classification, and obtaining classification results of the received signals on the three waveforms of FBMC, GFDM and UFMC.
2. The method for identifying a multicarrier waveform based on convolutional neural network according to claim 1, wherein said received signal is normalized before said step S2 is performed.
3. The convolutional neural network-based multicarrier waveform recognition method of claim 1, wherein the first convolutional neural network model, the second convolutional neural network model, and the third convolutional neural network model are each comprised of two convolutional layers and two fully-connected layers in sequence.
4. The method for identifying multi-carrier waveforms based on convolutional neural network as recited in claim 3, wherein 128 filters with a size of 1 x 3 are used in the first convolutional layer, and a linear rectifying unit is used as an active layer.
5. The method for identifying multi-carrier waveforms based on convolutional neural network as recited in claim 3, wherein 64 filters with a size of 1 x 3 and an active layer ReLU are used in the second convolutional layer.
6. The convolutional neural network-based multicarrier waveform identification method of claim 3, wherein the first fully-connected layer has 128 neurons.
7. The convolutional neural network-based multicarrier waveform identification method of claim 1, wherein the second full-connection layer is output via a SoftMax.
8. The multi-carrier waveform recognition system based on the convolutional neural network is characterized by comprising a signal receiving module (1), a first classification module (2), a second classification module (3) and a third classification module (4); the first classification module (2) is respectively connected with the signal receiving module (1), the second classification module (3) and the third classification module (4); wherein:
the signal receiving module (1) is used for acquiring a received signal of a waveform to be identified;
the first classification module (2) is used for performing two classifications on the received signal according to a first type waveform and a second type waveform by using a preset first convolutional neural network model; wherein the first type of waveform comprises three waveforms of OFDM, W-OFDM and F-OFDM; the second type of waveform comprises three waveforms of FBMC, GFDM and UFMC;
the second classification module (3) is configured to perform discrete fourier transform and Haar wavelet transform on the waveform signal divided into the first type of waveform in the step S2, input a preset second convolutional neural network model to perform three classifications, and obtain classification results of the received signal on three waveforms of OFDM, W-OFDM and F-OFDM;
the third classification module (4) is configured to perform FSST transformation on the waveform signal divided into the second type of waveform in the step S2, input a preset third convolutional neural network model to perform three classification, and obtain classification results of the received signal on three waveforms of FBMC, GFDM and UFMC.
9. A storage medium having a computer program stored thereon, characterized by: the computer program, when executed by a processor, implements the steps of a convolutional neural network-based multicarrier waveform identification method as claimed in any one of claims 1 to 7.
10. A communication device, characterized by: comprising a storage medium, a processor, and a computer program stored in the storage medium and executable by the processor, which when executed by the processor, performs the steps of the convolutional neural network-based multicarrier waveform identification method as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210368127.3A CN114679359B (en) | 2022-04-08 | 2022-04-08 | Multicarrier waveform identification method based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210368127.3A CN114679359B (en) | 2022-04-08 | 2022-04-08 | Multicarrier waveform identification method based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114679359A CN114679359A (en) | 2022-06-28 |
| CN114679359B true CN114679359B (en) | 2023-05-19 |
Family
ID=82077633
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210368127.3A Active CN114679359B (en) | 2022-04-08 | 2022-04-08 | Multicarrier waveform identification method based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114679359B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108600137A (en) * | 2018-04-28 | 2018-09-28 | 重庆邮电大学 | A kind of novel multicarrier recognition methods based on reverse transmittance nerve network |
| CN112364823A (en) * | 2020-11-30 | 2021-02-12 | 金陵科技学院 | Novel 5G multi-carrier signal identification method |
| CN113660185A (en) * | 2021-08-11 | 2021-11-16 | 金陵科技学院 | Multi-carrier signal type identification method based on extreme value distribution and wavelet transformation characteristics |
-
2022
- 2022-04-08 CN CN202210368127.3A patent/CN114679359B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108600137A (en) * | 2018-04-28 | 2018-09-28 | 重庆邮电大学 | A kind of novel multicarrier recognition methods based on reverse transmittance nerve network |
| CN112364823A (en) * | 2020-11-30 | 2021-02-12 | 金陵科技学院 | Novel 5G multi-carrier signal identification method |
| CN113660185A (en) * | 2021-08-11 | 2021-11-16 | 金陵科技学院 | Multi-carrier signal type identification method based on extreme value distribution and wavelet transformation characteristics |
Non-Patent Citations (1)
| Title |
|---|
| O'Shea，Timothy J等.Convolutional radio modulation recognition networks.《2016 International Conference on Engineering Application of Neutral Networks》.2016,全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114679359A (en) | 2022-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bhatti et al. | Shared spectrum monitoring using deep learning | |
| CN101227438B (en) | OFDM channel estimating method based on wavelet unbiased risk threshold value noise elimination | |
| Jeong et al. | Spectrogram-based automatic modulation recognition using convolutional neural network | |
| CN111541505B (en) | Time domain channel prediction method and system for OFDM wireless communication system | |
| Park et al. | Deep learning-based automatic modulation classification with blind OFDM parameter estimation | |
| CN109462443B (en) | 5G multi-carrier underwater acoustic communication method | |
| CN112702293A (en) | Signal parameter analysis method and system based on cyclic spectrum | |
| CN114679359B (en) | Multicarrier waveform identification method based on convolutional neural network | |
| Jiang et al. | Modulation recognition of underwater acoustic communication signals based on data transfer | |
| CN102739595A (en) | Method for reducing peak-to-average ratio of OFDM signal and device thereof | |
| CN116132237B (en) | OFDM signal detection and identification method based on periodic convolution and fractal | |
| An et al. | Multimodality-aided multicarrier waveform recognition in low SNR regimes based on denoised cyclic autocorrelation transformation | |
| CN114826857B (en) | OFDM signal generation method based on generation countermeasure network | |
| CN110035025A (en) | A kind of detection recognition method of the multicarrier mixed signal based on direct feature extraction | |
| CN115097397B (en) | Intermittent sampling forwarding interference resisting method and device and electronic equipment | |
| CN108600137B (en) | Novel multi-carrier identification method based on back propagation neural network | |
| CN114580470B (en) | OFDM and UFDM multi-carrier signal identification method based on non-uniform quantization map features | |
| An et al. | Blind multicarrier waveform recognition based on spatial-temporal learning neural networks | |
| Hasan et al. | Performance comparison of multicarrier communication systems over doubly-selective channels | |
| Chen et al. | Accurate Multicarrier Waveform Classification Using Convolutional Neural Networks | |
| CN107248964B (en) | Method for estimating code rate of underlay frequency spectrum sharing time-frequency overlapping signals | |
| Ghanim | Performance investigation of WOFDM for 5G wireless networks | |
| Zhou et al. | Deep Learning-Based Automatic Modulation Recognition in OTFS and OFDM systems | |
| Jajoo et al. | Modulation Classification for Overlapped Signals Using Deep Learning | |
| Guo et al. | Deep Learning Based RF Fingerprint Identification using Cyclic Shift Characteristic |
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 |