Robust communication radiation source intelligent identification method based on deep learning

A deep learning and intelligent recognition technology, applied in the field of machine learning, can solve the problems of discounted recognition accuracy and poor robustness, and achieve the effects of strong adaptability, improved matching degree, and high recognition sensitivity

Inactive Publication Date: 2021-04-06
北京理工大学重庆创新中心
View PDF1 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the previous classification methods rely too much on the accuracy and cleanliness of the extracted data features, which are less robust to the features, and the small noise, frequency offset, and phase shift of the signal will greatly reduce the recognition accuracy.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Robust communication radiation source intelligent identification method based on deep learning
  • Robust communication radiation source intelligent identification method based on deep learning
  • Robust communication radiation source intelligent identification method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0030]This example discloses a method of intelligent recognition based on deep learning robust communication radiation source, including:

[0031]A. Data pretreatment

[0032]Such asfigure 1 As shown, the signal preprocessing phase passes through the carrier frequency estimation, the lower frequency conversion, phase calculation, slimming, data standardization, and the like, the original data processing is a baseband signal data conducive to the extraction of radiation source subtle features in deep neural network.

[0033]A1. Bundry Estimation

[0034]Since there is an unknown frequency offset, the carrier frequency of the signal is first to be estimated before the intermediate frequency signal is downconverted. For QPSK signals, the four spectrums also have a strong discrete spectrum component in the quad-fold frequency, which is more powerful, and the carrier frequency estimation is simple, and the amount of operation is small based on the four-fold frequency. Higher, the carrier frequency o...

Embodiment 2

[0096]This example discloses a method of intelligent recognition based on deep learning robust communication radiation source, including:

[0097]The pretreatment step of signal sample data. The pre-processing step is cleaned on the signal sample data to make the sample data easy to train the learning model. The pre-processing step includes the process of carrier frequency estimation, down frequency, phase compensation, mining sample, and data standardization processing.

[0098]After pretreatment, the signal sample data is enhanced by using the signal-to-noise ratio evaluating the test set assessed in the test environment. The so-called test environment is an application environment, the test set is also the signal sample collected in the application environment, in this regard as the data enhancement reference to the training sample, making training samples to be more adapted to the actual environment.

[0099]Based on the design of the design model architecture, the CNN module is utilized...

Embodiment 3

[0103]The present embodiment will be described in accordance with the data acquired by the actual device, and the identification process will be described. Since the data used is collected by the actual equipment, there is a certain noise, frequency offset, and meta.

[0104]This embodiment contains 30 signal samples generated by 30 communication radiation sources. The signal is 70 MHz intermediate frequency signal sampling, the sample rate is 30 mSPS, the signal carrier frequency is 1000 MHz, the modulation method is QPSK, the modulation rate is 1M baud / s, the test data The blind selection module is an unknown radiation source signal that is not included in the training data.

[0105]First, pretreatment of the collected data signals, such asfigure 1 As shown, the specific training steps of the depth learning network are as follows:

[0106]Step 1, the carrier frequency estimation

[0107]For QPSK signals, there will be a strong discrete spectrum component in tetracy, and the carrier frequenc...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a robust communication radiation source intelligent identification method based on deep learning. The method comprises the steps of preprocessing signal sample data; performing data enhancement on the signal sample data; extracting signal sample data features by using a CNN module + LSTM model, classifying the features of each signal sample data by using a Softmax logistic regression model, identifying abnormal signal sample data according to the confidence of each signal sample data, and performing K-Means clustering on the features of the abnormal signal sample data. The design scheme of the invention is simple and easy to implement, and the method can be used for receiving signals with different carrier frequencies, signal-to-noise ratios and phase distortion degrees, has strong adaptability to intercepted signals in different communication environments, can be used for identifying historical radiation sources with labels, and can also be used for carrying out blind source classification on abnormal radiation sources without labels.

Description

Technical field[0001]The present invention relates to the field of machine learning, in particular, an intelligent identification method based on deep learning ropeless communication radiation source.Background technique[0002]The research of individual identification of communication radiation sources is late. In 1995, Choe delivered the first document on communication radiation source identification, and the transient signal based on wireless transmitter was identified. Transient signals are derived from a short unstable state where the communication station is just turned on or switched in operation mode. After the transient process ends, the radiation source will enter a long time, at which point the device is relatively small. And superimposed on a steady-state signal level, which makes the submissions of the radiation source in the steady state process become more difficult, but the steady state signal is easy to obtain, still have an important research value. At present, the a...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/049G06N3/08G06N3/045G06F2218/02G06F2218/08G06F2218/12G06F18/23213G06F18/241
Inventor 费泽松王新尧李维彪尹睿锐曾鸣
Owner 北京理工大学重庆创新中心
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap