A Novel Learned Volterra-based Scheme for Time-domain Nonlinear Equalization

Nelson Castro, Stylianos Sygletos

Research output: Chapter in Book/Published conference outputConference publication

Abstract

We introduce a learned time-domain Volterra-based equalizer to address nonlinear transmission impairments in fibre systems. When compared to conventional unoptimized equalization schemes it can offer more than 2.2 dB performance gain at reduced computational complexity.

Original languageEnglish
Title of host publication2022 Conference on Lasers and Electro-Optics, CLEO 2022 - Proceedings
PublisherIEEE
Number of pages2
ISBN (Electronic)9781957171050
Publication statusPublished - 23 Sep 2022
Event2022 Conference on Lasers and Electro-Optics, CLEO 2022 - San Jose, United States
Duration: 15 May 202220 May 2022

Publication series

Name2022 Conference on Lasers and Electro-Optics, CLEO 2022 - Proceedings

Conference

Conference2022 Conference on Lasers and Electro-Optics, CLEO 2022
Country/TerritoryUnited States
CitySan Jose
Period15/05/2220/05/22

Bibliographical note

Funding Information:
We introduced a time domain nonlinear equalization scheme based on a simplified IVSTF algorithm. Using ML that combines gradient back propagation and pruning it can result in more than 2.2 dB SNR improvement and a drastic reduction of its computational complexity compared to conventional NLEs. Acknowledgements: The work was supported by the EPSRC Programme Grant TRANSNET (EP/R035342/1).

Fingerprint

Dive into the research topics of 'A Novel Learned Volterra-based Scheme for Time-domain Nonlinear Equalization'. Together they form a unique fingerprint.

Cite this