Skip to main navigation Skip to search Skip to main content

Unsupervised Domain Adaptation based Modulation Classification for Overlapped Signals

  • School of Electronics and Information Engineering, Harbin Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

As the electromagnetic environment complexity increases, the importance of automatic modulation classification for overlapped signals (OS-AMC) becomes more evident. Existing deep learning methods utilizing supervised learning are heavily challenged because the actual application environment is more complicated than the ideal training environment, and the raw samples may not be labeled due to time constraints when it comes to an emergency. Hence, in this letter, we develop an unsupervised domain adaptation-based OS-AMC method to transfer the model trained under an additive white Gaussian noise (AWGN) channel to a multi-path channel. To the best of our knowledge, solving the OS-AMC problem in a complex environment has not been addressed yet.

Original languageEnglish
Title of host publicationProceedings - 2022 9th International Conference on Dependable Systems and Their Applications, DSA 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1014-1015
Number of pages2
ISBN (Electronic)9781665488778
DOIs
StatePublished - 2022
Externally publishedYes
Event9th International Conference on Dependable Systems and Their Applications, DSA 2022 - Wulumuqi, China
Duration: 4 Aug 20225 Aug 2022

Publication series

NameProceedings - 2022 9th International Conference on Dependable Systems and Their Applications, DSA 2022

Conference

Conference9th International Conference on Dependable Systems and Their Applications, DSA 2022
Country/TerritoryChina
CityWulumuqi
Period4/08/225/08/22

Keywords

  • Overlapped signals
  • deep adaptation network
  • deep learning
  • modulation classification
  • unsupervised domain adaptation

Fingerprint

Dive into the research topics of 'Unsupervised Domain Adaptation based Modulation Classification for Overlapped Signals'. Together they form a unique fingerprint.

Cite this