CN113269023A - Modulation identification method based on fraction low-order constellation diagram and deep learning - Google Patents
Modulation identification method based on fraction low-order constellation diagram and deep learning Download PDFInfo
- Publication number
- CN113269023A CN113269023A CN202110341451.1A CN202110341451A CN113269023A CN 113269023 A CN113269023 A CN 113269023A CN 202110341451 A CN202110341451 A CN 202110341451A CN 113269023 A CN113269023 A CN 113269023A
- Authority
- CN
- China
- Prior art keywords
- signal
- neural network
- deep neural
- constellation diagram
- order
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000010586 diagram Methods 0.000 title claims abstract description 29
- 238000013135 deep learning Methods 0.000 title claims abstract description 17
- 238000013528 artificial neural network Methods 0.000 claims abstract description 35
- 238000012549 training Methods 0.000 claims abstract description 23
- 238000012360 testing method Methods 0.000 claims abstract description 20
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000004519 manufacturing process Methods 0.000 claims abstract description 5
- 230000006870 function Effects 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 abstract description 4
- 238000010200 validation analysis Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 108010076504 Protein Sorting Signals Proteins 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 2
- 238000006731 degradation reaction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Digital Transmission Methods That Use Modulated Carrier Waves (AREA)
Abstract
A modulation identification method based on fraction low-order constellation diagram and deep learning mainly comprises the following steps: acquiring a signal and performing fractional low-order processing on the signal; calculating the characteristics of the fractional low-order constellation diagram and manufacturing a training set and a test set; designing and training a lightweight deep neural network; and testing the deep neural network and carrying out signal modulation identification. The method provided by the invention can obviously inhibit the influence of impulse noise on the signal modulation recognition accuracy rate, and meanwhile, the light-weight network can obviously reduce the calculation amount in the training and using processes.
Description
Technical Field
The invention relates to the technical field of modulation identification, in particular to a signal modulation identification method based on a fraction low-order constellation diagram and deep learning. The method is characterized by taking a fractional low-order constellation diagram as a characteristic and taking a deep neural network as a classifier, and realizes accurate identification of various signal modulations under the condition of pulse noise.
Background
Signal modulation identification is an important prerequisite for signal demodulation and content acquisition. In recent years, with the development of communication technology, the transmission environment of signals has become increasingly complex. Under the premise of not having sufficient prior knowledge of the signal, it is important to accurately identify the modulation mode of the signal.
Conventional signature-based signal modulation identification methods typically rely on manual selection of different kinds of signatures. However, due to the differences in personal experience and subjective judgment, this method is often labor and time consuming.
In recent years, with the rapid development of artificial intelligence technology, deep learning has been widely applied to many aspects such as computer vision, image processing, speech recognition, and the like as a branch of machine learning, and has exhibited excellent performance in the field of pattern recognition. Therefore, by means of the deep neural network involved in deep learning, the dependence of a feature selection stage on artificial experience and subjective judgment in the signal modulation identification process can be greatly reduced, and a plurality of signal modulation identification methods by means of deep learning appear. However, the existing methods and technologies do not solve two main problems often involved in the methods at the same time, namely: non-gaussian noise conditions and a lightweight deep neural network.
In view of this, the invention provides a modulation identification method based on a fractional low-order constellation diagram and deep learning. The invention firstly provides a concept of a fractional low-order constellation diagram which can inhibit impulse noise, namely typical non-Gaussian noise, then designs a lightweight deep neural network which can reduce the calculated amount, and finally provides a signal debugging and identifying method based on the fractional low-order constellation diagram and deep learning. From the perspective of pattern recognition, the fractional low-order constellation is a feature, and the lightweight neural network is a classifier. The method provided by the invention can obviously inhibit the influence of impulse noise on the signal modulation recognition accuracy rate, and meanwhile, the light-weight network can obviously reduce the calculation amount in the training and using processes.
Disclosure of Invention
In order to solve the problems of performance degradation and large calculation amount of the existing signal modulation identification method based on signal characteristics and deep learning under the condition of impulsive noise, the invention provides a modulation identification method based on a fraction low-order constellation diagram and deep learning.
The invention is realized by adopting the following scheme:
a: the signal is acquired and subjected to fractional low order processing.
A1: a signal is acquired.
A2: the signal is subjected to fractional low order processing.
B: and calculating the characteristics of the fractional low-order constellation diagram and manufacturing a training set and a test set.
B1: fractional low order constellation features of the signal are calculated.
B2: and fully mixing the characteristics, and forming a training set, a verification set and a test set by the mixed characteristics according to a certain proportion.
C: and designing and training a lightweight deep neural network.
C1: and designing a lightweight deep neural network.
C2: the deep neural network is trained using a training set.
C3: the deep neural network is validated using a validation set.
D: and testing the deep neural network and carrying out signal modulation identification.
D1: the deep neural network is tested using a test set.
D2: and carrying out signal modulation identification.
The invention has the beneficial effects that:
the method solves the problem of performance degradation of the existing signal modulation identification method under the typical non-Gaussian noise condition of impulsive noise, and reduces the problem of higher calculation amount introduced by a deep neural network. Meanwhile, the signal debugging and identifying method can identify a wide variety of debugging modes.
Drawings
Fig. 1 is a flowchart of a modulation identification method based on a fractional low-order constellation diagram and deep learning according to the present invention.
Figure 2 is a graph of a fractional low order mapping function in accordance with the present invention. Wherein each curve corresponds to a different parameter p.
Fig. 3 is a fractional low order constellation of signals in accordance with the present invention. Wherein, the signals take 2PSK, 4PSK, 8PSK, 16QAM, 32QAM and 64QAM as examples; the values of the parameter p related to the fractional low order mapping function are 1.2, 1.5 and 1.8 respectively.
Fig. 4 is a structural diagram of a lightweight deep neural network according to the present invention.
Fig. 5 is a graph of recognition accuracy for different methods under different noise conditions to which the present invention relates. The values of the parameter p related to the fractional low-order constellation diagram are respectively 1.1, 1.3 and 1.5, and the comparison method adopts the traditional constellation diagram as the characteristic.
Detailed Description
For better understanding, the technical solutions in the implementation process of the present invention are clearly and completely described below with reference to the accompanying drawings of the present invention.
A modulation identification method based on fractional low-order constellation and deep learning, the flow chart of the main steps is shown in fig. 1, specifically, it includes the following steps:
a: the signal is acquired and subjected to fractional low order processing.
The step A specifically comprises the following steps:
a1: a signal is acquired. The simulation signal can be generated by a computer according to the principles of different modulation modes, and simulated impulse noise is added to the simulation signal, and a real signal under the impulse noise condition can be acquired by a receiver and an antenna. Wherein, the intensity of the impulse noise is measured by adopting a generalized signal-to-noise ratio (GSNR), and the definition formula of the generalized signal-to-noise ratio is as follows:
wherein, PsRepresenting the power of the signal, PnRepresenting the generalized power of the noise. When Alpha stationary distributions are used to characterize non-Gaussian noise, Pnγ denotes the scale parameter of the Alpha stabilizing moiety.
A2: the signal is subjected to fractional low order processing. And performing fractional low-order processing on the signal by using a fractional low-order mapping function, wherein the specific formula is as follows:
yFLO(n)＝(y(n))<p-1>
＝aFLO(n)+jbFLO(n)
wherein n represents the discrete time variable corresponding to the sampling time, y (n) represents the sampled signal sequence, yFLO(n) represents a signal sequence processed by fractional lower-order (FLO), aFLORepresenting the real part of the signal, bFLOWhich represents the imaginary part of the signal,<·>representing a fractional low order operator that satisfies the following relationship:
wherein z represents a complex field
One complex number in (1), the upper corner indicates a conjugate operator.
A graph of a fractional low-order mapping function to which the present invention relates is shown in fig. 2. Wherein each curve corresponds to a different parameter p.
B: and calculating the characteristics of the fractional low-order constellation diagram and manufacturing a training set and a test set.
The step B specifically comprises the following steps:
b1: fractional low order constellation features of the signal are calculated. In the real part a of the signalFLOAs the abscissa, by the imaginary part b of the signalFLOAnd as a vertical coordinate, calculating the characteristics of the fractional low-order constellation diagram corresponding to the signals one by one according to the sampling time n.
B2: and fully mixing the characteristics, and forming a training set, a verification set and a test set by the mixed characteristics according to a certain proportion. Fully mixing the fractional low-order constellation diagram features obtained in the step B1, and then according to m1:m2:m3The proportions of (a) and (b) respectively constitute a training set, a verification set and a test set. In general, the ratio can be set to m1:m2:m3＝6:2:2。
The fractional low order constellation of the signal to which the present invention relates is shown in figure 3. Wherein, the signals take 2PSK, 4PSK, 8PSK, 16QAM, 32QAM and 64QAM as examples; the values of the parameter p related to the fractional low order mapping function are 1.2, 1.5 and 1.8 respectively.
C: and designing and training a lightweight deep neural network.
The step C specifically comprises the following steps:
c1: and designing a lightweight deep neural network. To reduce the amount of computation in the training, validation and testing process, a lightweight deep neural network is designed with fewer convolutional layers.
C2: the deep neural network is trained using a training set. Features in the training set are trained as inputs to the deep neural network in C1. In the training process, RMSprop is used as an optimization method of network parameters, cross entropy (cross entropy) is used as a loss function, and the learning rate is set to be 0.01.
C3: the deep neural network is validated using a validation set. Features in the validation set are validated as input to the deep neural network in C1.
The structure diagram of the deep neural network according to the present invention is shown in fig. 4, and specific network structure parameter settings are labeled in fig. 4.
D: and testing the deep neural network and carrying out signal modulation identification.
The step D specifically comprises the following steps:
d1: the deep neural network is tested using a test set. The features in the test set are tested as inputs to the deep neural network in C1.
D2: and carrying out signal modulation identification. And taking the fraction low-order constellation diagram characteristics of the signal with the unknown modulation mode as the input of the deep neural network in C1 to obtain the identification result of the signal modulation mode.
The recognition accuracy curves of the different methods under different noise conditions to which the present invention relates are shown in fig. 5. The values of the parameter p related to the fractional low-order constellation diagram are respectively 1.1, 1.3 and 1.5, and the comparison method adopts the traditional constellation diagram as the characteristic.
Claims (6)
1. A modulation identification method based on a fraction low-order constellation diagram and deep learning is characterized by comprising the following steps:
a: acquiring a signal and performing fractional low-order processing on the signal;
b: calculating the characteristics of the fractional low-order constellation diagram and manufacturing a training set and a test set;
c: designing and training a lightweight deep neural network;
d: and testing the deep neural network and carrying out signal modulation identification.
2. The modulation identification method according to claim 1, wherein the acquiring and performing fractional low order processing on the signal specifically comprises: and generating a simulation signal or acquiring a real signal under the impulse noise condition by using a receiver and an antenna, and then performing fractional low-order processing on the signal by using a fractional low-order mapping function.
3. The modulation identification method according to claim 1, wherein the calculating of the features of the fractional low-order constellation and the manufacturing of the training set and the test set specifically comprises: and taking the real part of the signal as an abscissa and the imaginary part of the signal as an ordinate, calculating the characteristics of the fractional low-order constellation diagram corresponding to the signal one by one according to the sampling time n, and then fully mixing the characteristics to prepare a data set.
4. The modulation identification method based on the fractional low-order constellation diagram and the deep learning of claim 1, wherein the deep neural network for designing and training lightweight specifically comprises: designing a lightweight deep neural network, training the deep neural network by using a training set and verifying the deep neural network by using a verification set.
5. The modulation identification method based on the fractional low-order constellation diagram and the deep learning of claim 4 is characterized in that RMSprop is adopted as an optimization method of network parameters in a training process, cross entropy is adopted as a loss function, and a learning rate is set to be 0.01.
6. The modulation identification method based on the fractional low-order constellation diagram and the deep learning according to claim 1, wherein the testing the deep neural network and performing the signal modulation identification specifically comprises: inputting the test set into a network to test the test set; then, the fractional low-order constellation diagram feature of the signal with unknown modulation mode is used as the input of the deep neural network in claim 4 to obtain the recognition result of the signal modulation mode.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2021103241851 | 2021-03-26 | ||
| CN202110324185 | 2021-03-26 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113269023A true CN113269023A (en) | 2021-08-17 |
| CN113269023B CN113269023B (en) | 2024-04-12 |
Family
ID=77228273
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110341451.1A Active CN113269023B (en) | 2021-03-26 | 2021-03-30 | Modulation identification method based on fractional low-order constellation diagram and deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113269023B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110071885A (en) * | 2019-04-17 | 2019-07-30 | 成都华日通讯技术有限公司 | A kind of deep learning method of discrimination of PSK digital signal subclass Modulation Identification |
| US20190273637A1 (en) * | 2018-04-30 | 2019-09-05 | Intel Corporation | Channel state information reference signal (csi-rs) and sounding reference signal (srs) triggering |
| CN110417694A (en) * | 2019-06-07 | 2019-11-05 | 吉林大学 | A kind of modulation mode of communication signal recognition methods |
| US20190342851A1 (en) * | 2018-02-09 | 2019-11-07 | Intel Corporation | Technologies to authorize user equipment use of local area data network features and control the size of local area data network information in access and mobility management function |
| CN111079347A (en) * | 2019-12-26 | 2020-04-28 | 华侨大学 | Signal-to-noise ratio estimation method based on deep learning by using constellation diagram |
| CN111585925A (en) * | 2020-04-18 | 2020-08-25 | 西北工业大学 | Robust real-time radio frequency signal modulation identification method based on deep learning |
| DE102020120784A1 (en) * | 2019-08-15 | 2021-02-18 | Intel Corporation | CHANNEL SCANNING FOR PHYSICAL DIRECT ACCESS CHANNEL SIGNALS IN NEW RADIO SYSTEMS THAT WORK IN THE UNLICENSED SPECTRUM |
-
2021
- 2021-03-30 CN CN202110341451.1A patent/CN113269023B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190342851A1 (en) * | 2018-02-09 | 2019-11-07 | Intel Corporation | Technologies to authorize user equipment use of local area data network features and control the size of local area data network information in access and mobility management function |
| US20190273637A1 (en) * | 2018-04-30 | 2019-09-05 | Intel Corporation | Channel state information reference signal (csi-rs) and sounding reference signal (srs) triggering |
| CN110071885A (en) * | 2019-04-17 | 2019-07-30 | 成都华日通讯技术有限公司 | A kind of deep learning method of discrimination of PSK digital signal subclass Modulation Identification |
| CN110417694A (en) * | 2019-06-07 | 2019-11-05 | 吉林大学 | A kind of modulation mode of communication signal recognition methods |
| DE102020120784A1 (en) * | 2019-08-15 | 2021-02-18 | Intel Corporation | CHANNEL SCANNING FOR PHYSICAL DIRECT ACCESS CHANNEL SIGNALS IN NEW RADIO SYSTEMS THAT WORK IN THE UNLICENSED SPECTRUM |
| CN111079347A (en) * | 2019-12-26 | 2020-04-28 | 华侨大学 | Signal-to-noise ratio estimation method based on deep learning by using constellation diagram |
| CN111585925A (en) * | 2020-04-18 | 2020-08-25 | 西北工业大学 | Robust real-time radio frequency signal modulation identification method based on deep learning |
Non-Patent Citations (3)
| Title |
|---|
| HOU CHANGBO等: ""Multimodal Feature Fusion Recognition of Modulated Signals Based on Image and Waveform Domain"", 《2020 7TH INTERNATIONAL CONFERENCE ON DEPENDABLE SYSTEMS AND THEIR APPLICATIONS (DSA)》, 26 January 2021 (2021-01-26) * |
| JIACHENG ZHANG等: ""Robust Sparse Representation for DOA Estimation With Unknown Mutual Coupling Under Impulsive Noise"", 《IEEE COMMUNICATIONS LETTERS》, vol. 24, no. 7, 24 March 2020 (2020-03-24), XP011798006, DOI: 10.1109/LCOMM.2020.2983038 * |
| 唐作栋;龚晓峰;雒瑞森;: "一种小波特征与深度神经网络结合的信号制式识别算法", 计算机工程与科学, no. 05, 15 May 2020 (2020-05-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113269023B (en) | 2024-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110163282B (en) | Modulation mode identification method based on deep learning | |
| CN107979554B (en) | Radio signal Modulation Identification method based on multiple dimensioned convolutional neural networks | |
| CN110855591B (en) | QAM and PSK signal intra-class modulation classification method based on convolutional neural network structure | |
| CN109274621B (en) | Communication protocol signal identification method based on depth residual error network | |
| CN108696331B (en) | Signal reconstruction method based on generation countermeasure network | |
| CN110619296B (en) | Signal noise reduction method based on singular decomposition | |
| CN112733811B (en) | Method for identifying underwater sound signal modulation modes based on improved dense neural network | |
| CN113014524B (en) | Digital signal modulation identification method based on deep learning | |
| CN113114599B (en) | Modulation identification method based on lightweight neural network | |
| CN110138459A (en) | Sparse underwater sound orthogonal frequency division multiplexing channel estimation methods and device based on base tracking denoising | |
| CN110071885A (en) | A kind of deep learning method of discrimination of PSK digital signal subclass Modulation Identification | |
| CN107276938A (en) | A kind of digital signal modulation mode recognition methods and device | |
| CN113037663B (en) | Improved code element rate estimation method suitable for non-constant envelope signal | |
| WO2021088465A1 (en) | Fast modulation recognition method using multilayer perceptron, and employing data fusion of multiple distribution tests | |
| CN113947151A (en) | Automatic modulation and identification method for wireless communication signals in offshore complex environment | |
| Ma et al. | Modulation classification method based on deep learning under non-Gaussian noise | |
| CN115186712A (en) | Modulated signal identification method and system | |
| CN113723556A (en) | Modulation mode identification method based on entropy weighting-multi-mode domain antagonistic neural network | |
| CN115001916A (en) | MCS identification method based on deep learning and blind identification | |
| CN109583575B (en) | Processing method for improving instrument vector signal analysis performance based on deep learning | |
| CN113518049B (en) | Modulation identification method based on fractional low-order polar coordinates and deep learning | |
| CN113269023B (en) | Modulation identification method based on fractional low-order constellation diagram and deep learning | |
| CN115913849A (en) | Electromagnetic signal identification method based on one-dimensional complex value residual error network | |
| CN115409056A (en) | Automatic modulation identification method for large dynamic signal-to-noise ratio | |
| CN115900802A (en) | Universal dynamic compensation method for multi-type sensors |
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 |