We present our data-collection and deep neural network (DNN)-training scheme for extracting the optical status from signals received by digital coherent optical receivers in fiber-optic networks. The DNN is trained with unlabeled datasets across multiple administrative network domains by combining federated learning and unsupervised learning. The scheme allows network administrators to train a common DNN-based encoder that extracts optical status in their networks without revealing their private datasets. An early-stage proof of concept was numerically demonstrated by simulation by estimating the optical signal-to-noise ratio and modulation format with 64-GBd 16QAM and quadrature phase-shift keying signals.
Takahito TANIMURA
https://orcid.org/0000-0001-5162-1104
Hitachi Ltd.
Riu HIRAI
Hitachi Ltd.
Nobuhiko KIKUCHI
Hitachi Ltd.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Takahito TANIMURA, Riu HIRAI, Nobuhiko KIKUCHI, "Physical Status Representation in Multiple Administrative Optical Networks by Federated Unsupervised Learning" in IEICE TRANSACTIONS on Communications,
vol. E106-B, no. 11, pp. 1084-1092, November 2023, doi: 10.1587/transcom.2022OBP0004.
Abstract: We present our data-collection and deep neural network (DNN)-training scheme for extracting the optical status from signals received by digital coherent optical receivers in fiber-optic networks. The DNN is trained with unlabeled datasets across multiple administrative network domains by combining federated learning and unsupervised learning. The scheme allows network administrators to train a common DNN-based encoder that extracts optical status in their networks without revealing their private datasets. An early-stage proof of concept was numerically demonstrated by simulation by estimating the optical signal-to-noise ratio and modulation format with 64-GBd 16QAM and quadrature phase-shift keying signals.
URL: https://global.ieice.org/en_transactions/communications/10.1587/transcom.2022OBP0004/_p
Copy
@ARTICLE{e106-b_11_1084,
author={Takahito TANIMURA, Riu HIRAI, Nobuhiko KIKUCHI, },
journal={IEICE TRANSACTIONS on Communications},
title={Physical Status Representation in Multiple Administrative Optical Networks by Federated Unsupervised Learning},
year={2023},
volume={E106-B},
number={11},
pages={1084-1092},
abstract={We present our data-collection and deep neural network (DNN)-training scheme for extracting the optical status from signals received by digital coherent optical receivers in fiber-optic networks. The DNN is trained with unlabeled datasets across multiple administrative network domains by combining federated learning and unsupervised learning. The scheme allows network administrators to train a common DNN-based encoder that extracts optical status in their networks without revealing their private datasets. An early-stage proof of concept was numerically demonstrated by simulation by estimating the optical signal-to-noise ratio and modulation format with 64-GBd 16QAM and quadrature phase-shift keying signals.},
keywords={},
doi={10.1587/transcom.2022OBP0004},
ISSN={1745-1345},
month={November},}
Copy
TY - JOUR
TI - Physical Status Representation in Multiple Administrative Optical Networks by Federated Unsupervised Learning
T2 - IEICE TRANSACTIONS on Communications
SP - 1084
EP - 1092
AU - Takahito TANIMURA
AU - Riu HIRAI
AU - Nobuhiko KIKUCHI
PY - 2023
DO - 10.1587/transcom.2022OBP0004
JO - IEICE TRANSACTIONS on Communications
SN - 1745-1345
VL - E106-B
IS - 11
JA - IEICE TRANSACTIONS on Communications
Y1 - November 2023
AB - We present our data-collection and deep neural network (DNN)-training scheme for extracting the optical status from signals received by digital coherent optical receivers in fiber-optic networks. The DNN is trained with unlabeled datasets across multiple administrative network domains by combining federated learning and unsupervised learning. The scheme allows network administrators to train a common DNN-based encoder that extracts optical status in their networks without revealing their private datasets. An early-stage proof of concept was numerically demonstrated by simulation by estimating the optical signal-to-noise ratio and modulation format with 64-GBd 16QAM and quadrature phase-shift keying signals.
ER -