Fully connected feed-forward neural network based nonlinearity compensation method for polarization multiplexed transmission systems

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

Abstract

In this work we propose a receiver-side nonlinearity compensation method based on fully connected feed-forward neural networks applicable to polarization-division multiplexing transmission systems. We consider different neural network architectures and show that the use of information from both polarizations allows to effectively compensate the accumulated nonlinear distortion.

Original languageEnglish
Title of host publicationProceedings - International Conference Laser Optics 2020, ICLO 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728152332
DOIs
Publication statusPublished - 2 Nov 2020
Event2020 International Conference Laser Optics, ICLO 2020 - St. Petersburg, Russian Federation
Duration: 2 Nov 20206 Nov 2020

Publication series

NameProceedings - International Conference Laser Optics 2020, ICLO 2020

Conference

Conference2020 International Conference Laser Optics, ICLO 2020
Country/TerritoryRussian Federation
CitySt. Petersburg
Period02.11.202006.11.2020

Keywords

  • fully connected feed forward neural networks
  • machine learning
  • nonlinearity compensation
  • polarization-division multiplexing

Fingerprint

Dive into the research topics of 'Fully connected feed-forward neural network based nonlinearity compensation method for polarization multiplexed transmission systems'. Together they form a unique fingerprint.

Cite this