Design of time-domain learned Volterra equalisers for WDM systems

Nelson Castro; Sonia Boscolo; Andrew D. Ellis; Stylianos Sygletos

doi:10.23919/ONDM61578.2024.10582691

Design of time-domain learned Volterra equalisers for WDM systems

Nelson Castro, Sonia Boscolo, Andrew D. Ellis, Stylianos Sygletos

Aston Institute of Photonic Technologies (AiPT)

Research output: Chapter in Book/Published conference output › Conference publication

Abstract

We examine various design aspects of a learned time-domain multiple-input multiple-output (MIMO) Volterra-based equaliser and reveal their impact on the convergence and performance of the model. We show that appropriate parameter initialisation is vital for the model's convergence and scalability to a higher number of channels. This design optimisation enables the first demonstration of a 7 × 7 operation of the MIMO algorithm at one step per span, achieving 1.5 dB effective signal-to-noise ratio improvement over single-channel nonlinear equalisation, hence significantly enhancing the transmission performance in a wavelength-division multiplexing scenario.

Original language	English
Title of host publication	Proceedings of the 2024 International Conference on Optical Network Design and Modeling (ONDM)
Editors	David Larrabeiti-Lopez, Luca Valcarenghi, Carmen Mas-Machuca, Jose A. Hernandez-Gutierrez
Publisher	IEEE
Number of pages	3
ISBN (Electronic)	9783903176546
DOIs	https://doi.org/10.23919/ONDM61578.2024.10582691
Publication status	Published - 11 Jul 2024
Event	28th International Conference on Optical Network Design and Modelling - University Carlos III de Madrid, Madrid, Spain Duration: 6 May 2024 → 9 May 2024 https://ondm2024.uc3m.es/home

Conference

Conference	28th International Conference on Optical Network Design and Modelling
Abbreviated title	ONDM 2024
Country/Territory	Spain
City	Madrid
Period	6/05/24 → 9/05/24
Internet address	https://ondm2024.uc3m.es/home

Bibliographical note

Keywords

Volterra series
machine learning
nonlinearity equalisation
optical fibre systems

Access to Document

10.23919/ONDM61578.2024.10582691

N Castro et al_Design of Time-Domain Learned Volterra Equalisers for WDM Systems
Copyright © 2024 IFIP. This is an accepted manuscript of a paper presented at the 28th International Conference on Optical Network Design and Modelling.
Accepted author manuscript, 514 KBLicence: Unspecified

Cite this

Castro, N., Boscolo, S., Ellis, A. D., & Sygletos, S. (2024). Design of time-domain learned Volterra equalisers for WDM systems. In D. Larrabeiti-Lopez, L. Valcarenghi, C. Mas-Machuca, & J. A. Hernandez-Gutierrez (Eds.), Proceedings of the 2024 International Conference on Optical Network Design and Modeling (ONDM) IEEE. https://doi.org/10.23919/ONDM61578.2024.10582691

@inproceedings{893c90b9f86947a38e0960637d39bf93,

title = "Design of time-domain learned Volterra equalisers for WDM systems",

abstract = "We examine various design aspects of a learned time-domain multiple-input multiple-output (MIMO) Volterra-based equaliser and reveal their impact on the convergence and performance of the model. We show that appropriate parameter initialisation is vital for the model's convergence and scalability to a higher number of channels. This design optimisation enables the first demonstration of a 7 × 7 operation of the MIMO algorithm at one step per span, achieving 1.5 dB effective signal-to-noise ratio improvement over single-channel nonlinear equalisation, hence significantly enhancing the transmission performance in a wavelength-division multiplexing scenario.",

keywords = "Volterra series, machine learning, nonlinearity equalisation, optical fibre systems",

author = "Nelson Castro and Sonia Boscolo and Ellis, {Andrew D.} and Stylianos Sygletos",

note = "Copyright {\textcopyright} 2024 IFIP. This is an accepted manuscript of a paper presented at the 28th International Conference on Optical Network Design and Modelling. ; 28th International Conference on Optical Network Design and Modelling , ONDM 2024 ; Conference date: 06-05-2024 Through 09-05-2024",

year = "2024",

month = jul,

day = "11",

doi = "10.23919/ONDM61578.2024.10582691",

language = "English",

editor = "David Larrabeiti-Lopez and Luca Valcarenghi and Carmen Mas-Machuca and Hernandez-Gutierrez, {Jose A.}",

booktitle = "Proceedings of the 2024 International Conference on Optical Network Design and Modeling (ONDM)",

publisher = "IEEE",

address = "United States",

url = "https://ondm2024.uc3m.es/home",

}

Castro, N, Boscolo, S , Ellis, AD & Sygletos, S 2024, Design of time-domain learned Volterra equalisers for WDM systems. in D Larrabeiti-Lopez, L Valcarenghi, C Mas-Machuca & JA Hernandez-Gutierrez (eds), Proceedings of the 2024 International Conference on Optical Network Design and Modeling (ONDM). IEEE, 28th International Conference on Optical Network Design and Modelling , Madrid, Spain, 6/05/24. https://doi.org/10.23919/ONDM61578.2024.10582691

Design of time-domain learned Volterra equalisers for WDM systems. / Castro, Nelson; Boscolo, Sonia ; Ellis, Andrew D. et al.
Proceedings of the 2024 International Conference on Optical Network Design and Modeling (ONDM). ed. / David Larrabeiti-Lopez; Luca Valcarenghi; Carmen Mas-Machuca; Jose A. Hernandez-Gutierrez. IEEE, 2024.

Research output: Chapter in Book/Published conference output › Conference publication

TY - GEN

T1 - Design of time-domain learned Volterra equalisers for WDM systems

AU - Castro, Nelson

AU - Boscolo, Sonia

AU - Ellis, Andrew D.

AU - Sygletos, Stylianos

PY - 2024/7/11

Y1 - 2024/7/11

N2 - We examine various design aspects of a learned time-domain multiple-input multiple-output (MIMO) Volterra-based equaliser and reveal their impact on the convergence and performance of the model. We show that appropriate parameter initialisation is vital for the model's convergence and scalability to a higher number of channels. This design optimisation enables the first demonstration of a 7 × 7 operation of the MIMO algorithm at one step per span, achieving 1.5 dB effective signal-to-noise ratio improvement over single-channel nonlinear equalisation, hence significantly enhancing the transmission performance in a wavelength-division multiplexing scenario.

AB - We examine various design aspects of a learned time-domain multiple-input multiple-output (MIMO) Volterra-based equaliser and reveal their impact on the convergence and performance of the model. We show that appropriate parameter initialisation is vital for the model's convergence and scalability to a higher number of channels. This design optimisation enables the first demonstration of a 7 × 7 operation of the MIMO algorithm at one step per span, achieving 1.5 dB effective signal-to-noise ratio improvement over single-channel nonlinear equalisation, hence significantly enhancing the transmission performance in a wavelength-division multiplexing scenario.

KW - Volterra series

KW - machine learning

KW - nonlinearity equalisation

KW - optical fibre systems

UR - https://ieeexplore.ieee.org/document/10582691

UR - http://www.scopus.com/inward/record.url?scp=85199311219&partnerID=8YFLogxK

U2 - 10.23919/ONDM61578.2024.10582691

DO - 10.23919/ONDM61578.2024.10582691

M3 - Conference publication

BT - Proceedings of the 2024 International Conference on Optical Network Design and Modeling (ONDM)

A2 - Larrabeiti-Lopez, David

A2 - Valcarenghi, Luca

A2 - Mas-Machuca, Carmen

A2 - Hernandez-Gutierrez, Jose A.

PB - IEEE

T2 - 28th International Conference on Optical Network Design and Modelling

Y2 - 6 May 2024 through 9 May 2024

ER -

Design of time-domain learned Volterra equalisers for WDM systems

Abstract

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this