Unsupervised and supervised machine learning for performance improvement of NFT optical transmission

Oleksandr Kotlyar; Maryna Pankratova; Morteza Kamalian Kopae; Anastasiia Vasylchenkova; Jaroslaw E. Prilepsky; Sergei K. Turitsyn

doi:10.1109/BICOP.2018.8658274

Unsupervised and supervised machine learning for performance improvement of NFT optical transmission

Oleksandr Kotlyar, Maryna Pankratova, Morteza Kamalian Kopae, Anastasiia Vasylchenkova, Jaroslaw E. Prilepsky, Sergei K. Turitsyn

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

Abstract

We apply both the unsupervised and supervised
machine learning (ML) methods, in particular, the k-means
clustering and support vector machine (SVM) to improve the
performance of the optical communication system based on the
nonlinear Fourier transform (NFT). The NFT system employs the
continuous NFT spectrum part to carry data up to 1000 km using
the 16-QAM OFDM modulation. We classify the performance of
the system in terms of BER versus signal power dependence.
We show that the NFT system performance can be improved
considerably by means of the ML techniques and that the
more advanced SVM method typically outperforms the k-means
clustering.

Original language	English
Title of host publication	2018 British and Irish Conference on Optics and Photonics, BICOP 2018 - Proceedings
Publisher	IEEE
ISBN (Electronic)	978-153867361-4
ISBN (Print)	978-1-5386-7362-1
DOIs	https://doi.org/10.1109/BICOP.2018.8658274
Publication status	Published - 4 Mar 2019
Event	1st IEEE British and Irish Conference on Optics and Photonics (BICOP 2018) - London, United Kingdom Duration: 12 Dec 2018 → 14 Dec 2018

Conference

Conference	1st IEEE British and Irish Conference on Optics and Photonics (BICOP 2018)
Country/Territory	United Kingdom
City	London
Period	12/12/18 → 14/12/18

Bibliographical note

© 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

Funding: European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie Grant Agreements No.751561 (MP) and No.713694 (OK), EPSRC project TRANSNET (EP/R035342/1) (OK, MK & SKT) and the Leverhulme Trust project (RPG-2018-063) (JEP & SKT).

Keywords

Machine learning
k-means clustering
nonlinear Fourer transform
optical communications
support vector machine

Access to Document

10.1109/BICOP.2018.8658274

Unsupervised and supervised machine learning
© 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Funding: European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie Grant Agreements No.751561 (MP) and No.713694 (OK), EPSRC project TRANSNET (EP/R035342/1) (OK, MK & SKT) and the Leverhulme Trust project (RPG-2018-063) (JEP & SKT).
Accepted author manuscript, 408 KB

Cite this

@inproceedings{991066de23c946d98c5768cc6c2ee041,

title = "Unsupervised and supervised machine learning for performance improvement of NFT optical transmission",

abstract = "We apply both the unsupervised and supervisedmachine learning (ML) methods, in particular, the k-meansclustering and support vector machine (SVM) to improve theperformance of the optical communication system based on thenonlinear Fourier transform (NFT). The NFT system employs thecontinuous NFT spectrum part to carry data up to 1000 km usingthe 16-QAM OFDM modulation. We classify the performance ofthe system in terms of BER versus signal power dependence.We show that the NFT system performance can be improvedconsiderably by means of the ML techniques and that themore advanced SVM method typically outperforms the k-meansclustering.",

keywords = "Machine learning, k-means clustering, nonlinear Fourer transform, optical communications, support vector machine",

author = "Oleksandr Kotlyar and Maryna Pankratova and {Kamalian Kopae}, Morteza and Anastasiia Vasylchenkova and Prilepsky, {Jaroslaw E.} and Turitsyn, {Sergei K.}",

note = "{\textcopyright} 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Funding: European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie Grant Agreements No.751561 (MP) and No.713694 (OK), EPSRC project TRANSNET (EP/R035342/1) (OK, MK & SKT) and the Leverhulme Trust project (RPG-2018-063) (JEP & SKT).; 1st IEEE British and Irish Conference on Optics and Photonics (BICOP 2018) ; Conference date: 12-12-2018 Through 14-12-2018",

year = "2019",

month = mar,

day = "4",

doi = "10.1109/BICOP.2018.8658274",

language = "English",

isbn = "978-1-5386-7362-1",

booktitle = "2018 British and Irish Conference on Optics and Photonics, BICOP 2018 - Proceedings",

publisher = "IEEE",

address = "United States",

}

Kotlyar, O, Pankratova, M, Kamalian Kopae, M, Vasylchenkova, A, Prilepsky, JE & Turitsyn, SK 2019, Unsupervised and supervised machine learning for performance improvement of NFT optical transmission. in 2018 British and Irish Conference on Optics and Photonics, BICOP 2018 - Proceedings., 8658274, IEEE, 1st IEEE British and Irish Conference on Optics and Photonics (BICOP 2018), London, United Kingdom, 12/12/18. https://doi.org/10.1109/BICOP.2018.8658274

TY - GEN

T1 - Unsupervised and supervised machine learning for performance improvement of NFT optical transmission

AU - Kotlyar, Oleksandr

AU - Pankratova, Maryna

AU - Kamalian Kopae, Morteza

AU - Vasylchenkova, Anastasiia

AU - Prilepsky, Jaroslaw E.

AU - Turitsyn, Sergei K.

N1 - © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Funding: European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie Grant Agreements No.751561 (MP) and No.713694 (OK), EPSRC project TRANSNET (EP/R035342/1) (OK, MK & SKT) and the Leverhulme Trust project (RPG-2018-063) (JEP & SKT).

PY - 2019/3/4

Y1 - 2019/3/4

N2 - We apply both the unsupervised and supervisedmachine learning (ML) methods, in particular, the k-meansclustering and support vector machine (SVM) to improve theperformance of the optical communication system based on thenonlinear Fourier transform (NFT). The NFT system employs thecontinuous NFT spectrum part to carry data up to 1000 km usingthe 16-QAM OFDM modulation. We classify the performance ofthe system in terms of BER versus signal power dependence.We show that the NFT system performance can be improvedconsiderably by means of the ML techniques and that themore advanced SVM method typically outperforms the k-meansclustering.

AB - We apply both the unsupervised and supervisedmachine learning (ML) methods, in particular, the k-meansclustering and support vector machine (SVM) to improve theperformance of the optical communication system based on thenonlinear Fourier transform (NFT). The NFT system employs thecontinuous NFT spectrum part to carry data up to 1000 km usingthe 16-QAM OFDM modulation. We classify the performance ofthe system in terms of BER versus signal power dependence.We show that the NFT system performance can be improvedconsiderably by means of the ML techniques and that themore advanced SVM method typically outperforms the k-meansclustering.

KW - Machine learning

KW - k-means clustering

KW - nonlinear Fourer transform

KW - optical communications

KW - support vector machine

UR - https://ieeebicop.com/

UR - https://ieeexplore.ieee.org/document/8658274

UR - http://www.scopus.com/inward/record.url?scp=85063904926&partnerID=8YFLogxK

U2 - 10.1109/BICOP.2018.8658274

DO - 10.1109/BICOP.2018.8658274

M3 - Conference publication

SN - 978-1-5386-7362-1

BT - 2018 British and Irish Conference on Optics and Photonics, BICOP 2018 - Proceedings

PB - IEEE

T2 - 1st IEEE British and Irish Conference on Optics and Photonics (BICOP 2018)

Y2 - 12 December 2018 through 14 December 2018

ER -

Unsupervised and supervised machine learning for performance improvement of NFT optical transmission

Abstract

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this