Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Method and system for assisting deep learning based on blind equalization

A deep learning and deep learning network technology, applied in the field of deep learning based on blind equalization, can solve the problems of deterioration of modulation recognition performance, affecting the accuracy of modulation recognition, etc., and achieve the effect of improving the recognition accuracy.

Inactive Publication Date: 2020-06-02
NANTONG RES INST FOR ADVANCED COMM TECH CO LTD +1
View PDF3 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in this kind of multipath fading channel, there is a serious deterioration in the performance of modulation recognition, which greatly affects the accuracy of modulation recognition.

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
  • Method and system for assisting deep learning based on blind equalization
  • Method and system for assisting deep learning based on blind equalization
  • Method and system for assisting deep learning based on blind equalization

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0028] This specific embodiment discloses a method for assisting deep learning based on blind equalization, including the following steps:

[0029] Sampling, sampling the received modulated signal, and sending the sampled signal to the blind equalizer;

[0030] Equalization, using various step sizes to equalize the received signal;

[0031] Check and clear, check the signal data after equalization, and clear the empty data that failed the equalization;

[0032] Recognition learning, send all the successfully balanced data to the deep learning network, extract the hidden information in the data, and use the hidden information to identify the balanced signal.

[0033] Further, the equalization method adopts a multi-mode equalization method.

[0034] Further, the recognition learning method adopts a deep learning method based on a deep learning network, and the deep learning network is a deep learning network based on a Resnet network.

[0035] Feasible, wherein the equalizati...

Embodiment 2

[0038] This specific embodiment discloses a system applying the blind equalization-assisted deep learning method as in Embodiment 1, which includes an equalization module and a deep learning module, and the equalization module and the deep learning module form a bidirectional transmission link.

[0039] Further, the equalization module includes a channel response unit, a step size selection unit, a blind equalizer and an empty data clearing unit, a transmission link is provided between the empty data clearing unit and the deep learning module, and the channel response unit, the blind equalizer The equalizer and the empty data clearing unit are sequentially connected through a signal transmission link, and the front end of the blind equalizer is also provided with a step size selection unit, and the step size selection unit is used to judge the step size of the signal, and the blind equalizer The unknown signal is equalized according to the step size determined by the step size ...

Embodiment 3

[0044] This specific embodiment discloses a verification method for the blind equalization-assisted deep learning method and system pair in Embodiment 1 and Embodiment 2, and the verification is specifically performed through the following process.

[0045] in image 3 It is a model diagram of this experiment, assuming that the signals are three kinds of QAM signals, and the test environment is a sea area communication channel.

[0046] As shown in the figure, the received signal can be written as:

[0047]

[0048] Where r(t) is the signal received by the receiving end, s(t) is the transmitted signal, h(t,τ) is the impulse response of the channel at time t, τ represents the position where the multipath component appears, and n(t) is Additive white Gaussian noise.

[0049] Here, the blind equalizer adopts a mature multi-mode (MMA) algorithm, and the specific process of the algorithm is as follows:

[0050] Equalized output:

[0051]

[0052] where x n @[x(n),x(n-1),.....

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 method for assisting deep learning based on blind equalization, which belongs to the field of communication and comprises the following steps: sampling: sampling a received modulation signal, and sending the sampled signal to a blind equalizer; equalization: equalizing the received signal by using multiple step lengths; checking and clearing: checking the equalized signaldata, and clearing the empty data which fails to be equalized; and identification learning: sending all successfully equalized data to a deep learning network, extracting implicit information in thedata, and identifying equalized signals by utilizing the implicit information. The invention further discloses a system based on the method, the system comprises an equalization module and a deep learning module, and the equalization module and the deep learning module bidirectionally form a transmission link. The method has the advantages that the received signals are equalized by using a blind equalization method, and then the potential information is acquired by using the deep learning network for identification, so that the identification accuracy of unknown signals is improved.

Description

technical field [0001] The present invention relates to the communication field, in particular to a method and system for assisting deep learning based on blind equalization. Background technique [0002] Modulation identification is an important content in communication, and modulation identification realizes the classification of modulation types of received signals. It has wide application value in cognitive radio and military communication. Specifically, in cognitive radio systems, knowing the modulation used by other (primary or secondary) users can help current receivers better understand the interference environment, while in military communications, an eavesdropper must know the modulation type of the other party , in order to further decode the intercepted signal. Traditional modulation recognition methods are divided into likelihood-based (LB) methods and feature-based (FB) methods. The LB method calculates the likelihood function of the signal and compares it w...

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): H04L27/00
CPCH04L27/0012H04L2027/0038
Inventor 王珏纪雪飞张艳秋厉凯孙强曹娟徐晨杨永杰
Owner NANTONG RES INST FOR ADVANCED COMM TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com