Attention-aided partial bidirectional RNN-based nonlinear equalizer in coherent optical systems

Yifan Liu; Victor Sanchez; Pedro J. Freire; Jaroslaw E. Prilepsky; Mahyar J. Koshkouei; Matthew D. Higgins

doi:10.1364/oe.464159

Attention-aided partial bidirectional RNN-based nonlinear equalizer in coherent optical systems

Yifan Liu, Victor Sanchez, Pedro J. Freire, Jaroslaw E. Prilepsky, Mahyar J. Koshkouei, Matthew D. Higgins

Research output: Contribution to journal › Article › peer-review

Abstract

We leverage the attention mechanism to investigate and comprehend the contribution of each input symbol of the input sequence and their hidden representations for predicting the received symbol in the bidirectional recurrent neural network (BRNN)-based nonlinear equalizer. In this paper, we propose an attention-aided novel design of a partial BRNN-based nonlinear equalizer, and evaluate with both LSTM and GRU units in a single-channel DP-64QAM 30Gbaud coherent optical communication systems of 20 × 50 km standard single-mode fiber (SSMF) spans. Our approach maintains the Q-factor performance of the baseline equalizer with a significant complexity reduction of ∼56.16% in the number of real multiplications required to equalize per symbol (RMpS). In comparison of the performance under similar complexity, our approach outperforms the baseline by ∼0.2dB to ∼0.25dB at the optimal transmit power, and ∼0.3dB to ∼0.45dB towards the more nonlinear region.

Original language	English
Article number	32908
Pages (from-to)	32908-32923
Number of pages	16
Journal	Optics Express
Volume	30
Issue number	18
Early online date	24 Aug 2022
DOIs	https://doi.org/10.1364/oe.464159
Publication status	Published - 29 Aug 2022

Bibliographical note

Published by Optica Publishing Group under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.

Funding Information:
H2020 Marie Skłodowska-Curie Actions (813144); Leverhulme Trust (RP-2018-063); Engineering and Physical Sciences Research Council (EP/N509796/1, EP/R513374/1).

Keywords

Atomic and Molecular Physics, and Optics

Access to Document

10.1364/oe.464159Licence: CC BY 4.0

Liuetal_2022_VoR
Published by Optica Publishing Group under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.
Final published version, 2.7 MBLicence: CC BY 4.0

Cite this

@article{0e0838e1e1414eedb84a76fbbe831cb1,

title = "Attention-aided partial bidirectional RNN-based nonlinear equalizer in coherent optical systems",

abstract = "We leverage the attention mechanism to investigate and comprehend the contribution of each input symbol of the input sequence and their hidden representations for predicting the received symbol in the bidirectional recurrent neural network (BRNN)-based nonlinear equalizer. In this paper, we propose an attention-aided novel design of a partial BRNN-based nonlinear equalizer, and evaluate with both LSTM and GRU units in a single-channel DP-64QAM 30Gbaud coherent optical communication systems of 20 × 50 km standard single-mode fiber (SSMF) spans. Our approach maintains the Q-factor performance of the baseline equalizer with a significant complexity reduction of ∼56.16% in the number of real multiplications required to equalize per symbol (RMpS). In comparison of the performance under similar complexity, our approach outperforms the baseline by ∼0.2dB to ∼0.25dB at the optimal transmit power, and ∼0.3dB to ∼0.45dB towards the more nonlinear region.",

keywords = "Atomic and Molecular Physics, and Optics",

author = "Yifan Liu and Victor Sanchez and Freire, {Pedro J.} and Prilepsky, {Jaroslaw E.} and Koshkouei, {Mahyar J.} and Higgins, {Matthew D.}",

note = "Published by Optica Publishing Group under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI. Funding Information: H2020 Marie Sk{\l}odowska-Curie Actions (813144); Leverhulme Trust (RP-2018-063); Engineering and Physical Sciences Research Council (EP/N509796/1, EP/R513374/1). ",

year = "2022",

month = aug,

day = "29",

doi = "10.1364/oe.464159",

language = "English",

volume = "30",

pages = "32908--32923",

journal = "Optics Express",

issn = "1094-4087",

publisher = "The Optical Society",

number = "18",

}

TY - JOUR

T1 - Attention-aided partial bidirectional RNN-based nonlinear equalizer in coherent optical systems

AU - Liu, Yifan

AU - Sanchez, Victor

AU - Freire, Pedro J.

AU - Prilepsky, Jaroslaw E.

AU - Koshkouei, Mahyar J.

AU - Higgins, Matthew D.

N1 - Published by Optica Publishing Group under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI. Funding Information: H2020 Marie Skłodowska-Curie Actions (813144); Leverhulme Trust (RP-2018-063); Engineering and Physical Sciences Research Council (EP/N509796/1, EP/R513374/1).

PY - 2022/8/29

Y1 - 2022/8/29

N2 - We leverage the attention mechanism to investigate and comprehend the contribution of each input symbol of the input sequence and their hidden representations for predicting the received symbol in the bidirectional recurrent neural network (BRNN)-based nonlinear equalizer. In this paper, we propose an attention-aided novel design of a partial BRNN-based nonlinear equalizer, and evaluate with both LSTM and GRU units in a single-channel DP-64QAM 30Gbaud coherent optical communication systems of 20 × 50 km standard single-mode fiber (SSMF) spans. Our approach maintains the Q-factor performance of the baseline equalizer with a significant complexity reduction of ∼56.16% in the number of real multiplications required to equalize per symbol (RMpS). In comparison of the performance under similar complexity, our approach outperforms the baseline by ∼0.2dB to ∼0.25dB at the optimal transmit power, and ∼0.3dB to ∼0.45dB towards the more nonlinear region.

AB - We leverage the attention mechanism to investigate and comprehend the contribution of each input symbol of the input sequence and their hidden representations for predicting the received symbol in the bidirectional recurrent neural network (BRNN)-based nonlinear equalizer. In this paper, we propose an attention-aided novel design of a partial BRNN-based nonlinear equalizer, and evaluate with both LSTM and GRU units in a single-channel DP-64QAM 30Gbaud coherent optical communication systems of 20 × 50 km standard single-mode fiber (SSMF) spans. Our approach maintains the Q-factor performance of the baseline equalizer with a significant complexity reduction of ∼56.16% in the number of real multiplications required to equalize per symbol (RMpS). In comparison of the performance under similar complexity, our approach outperforms the baseline by ∼0.2dB to ∼0.25dB at the optimal transmit power, and ∼0.3dB to ∼0.45dB towards the more nonlinear region.

KW - Atomic and Molecular Physics, and Optics

UR - https://opg.optica.org/oe/fulltext.cfm?uri=oe-30-18-32908&id=495566

UR - http://www.scopus.com/inward/record.url?scp=85138666991&partnerID=8YFLogxK

U2 - 10.1364/oe.464159

DO - 10.1364/oe.464159

M3 - Article

SN - 1094-4087

VL - 30

SP - 32908

EP - 32923

JO - Optics Express

JF - Optics Express

IS - 18

M1 - 32908

ER -

Attention-aided partial bidirectional RNN-based nonlinear equalizer in coherent optical systems

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this