Machine-learning-based telemetry for monitoring long-haul optical transmission impairments: methodologies and challenges [Invited]

Huazhi Lun; Xiaomin Liu; Meng Cai; Yihao Zhang; Ruoxuan Gao; Weisheng Hu; Lilin Yi; Qunbi Zhuge

doi:10.1364/JOCN.426826

Journal of Optical Communications and Networking
Vol. 13,
Issue 10,
pp. E94-E108
(2021)
•https://doi.org/10.1364/JOCN.426826

Machine-learning-based telemetry for monitoring long-haul optical transmission impairments: methodologies and challenges [Invited]

Huazhi Lun, Xiaomin Liu, Meng Cai, Yihao Zhang, Ruoxuan Gao, Weisheng Hu, Lilin Yi, and Qunbi Zhuge

Not Accessible

Your library or personal account may give you access

Get PDF
Email
Share
Get Citation
Copy Citation Text
Huazhi Lun, Xiaomin Liu, Meng Cai, Yihao Zhang, Ruoxuan Gao, Weisheng Hu, Lilin Yi, and Qunbi Zhuge, "Machine-learning-based telemetry for monitoring long-haul optical transmission impairments: methodologies and challenges [Invited]," J. Opt. Commun. Netw. 13, E94-E108 (2021)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Citation alert
Save article

Related Topics
Optics & Photonics Topics
?

The topics in this list come from the Optics and Photonics Topics applied to this article.

About this Article
History
- Original Manuscript: April 1, 2021
- Revised Manuscript: June 8, 2021
- Manuscript Accepted: June 9, 2021
- Published: July 6, 2021
Virtual Issues
Journal of Optical Communications and Networking Advanced Monitoring and Telemetry in Optical Networks (2021)

Abstract

Current management of optical communication systems is conservative, manual-based, and time-consuming. To improve this situation, building an intelligent closed-loop control system is becoming an active topic of the industry. One of the key techniques to achieve such a management system is physical layer impairment telemetry, with the help of which the controller can make proper instructions. However, it is challenging to implement an accurate telemetry module due to the complex mechanisms of various impairments. To overcome that, many studies have been done. In this paper, those recent studies are reviewed, and the design of telemetry is discussed systematically. We analyze metrics for evaluating system performance and mechanisms of various impairments comprehensively, which are the theoretical foundations for designing telemetry modules. We then summarize a unified workflow for designing telemetry modules based on the review of previous works. Its effectiveness is then verified by concrete use cases of our previous studies. Finally, we discuss the challenges of deploying machine-learning-based telemetry techniques in optical communication systems.

Full Article | PDF Article

More Like This

Machine-learning-based soft-failure localization with partial software-defined networking telemetry

Kayol S. Mayer, Jonathan A. Soares, Rossano P. Pinto, Christian E. Rothenberg, Dalton S. Arantes, and Darli A. A. Mello
J. Opt. Commun. Netw. 13(10) E122-E131 (2021)

Performance comparisons between machine learning and analytical models for quality of transmission estimation in wavelength-division-multiplexed systems [Invited]

Jianing Lu, Gai Zhou, Qirui Fan, Dengke Zeng, Changjian Guo, Linyue Lu, Jianqiang Li, Chongjin Xie, Chao Lu, Faisal Nadeem Khan, and Alan Pak Tao Lau
J. Opt. Commun. Netw. 13(4) B35-B44 (2021)

Machine-learning-based EDFA gain estimation [Invited]

Jiakai Yu, Shengxiang Zhu, Craig L. Gutterman, Gil Zussman, and Daniel C. Kilper
J. Opt. Commun. Netw. 13(4) B83-B91 (2021)

Previous Article Next Article

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Figures (25)

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Tables (4)

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Equations (15)

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Target	Reference	ML Algorithm	Input	Output	Evaluation Method	Performance
NLI	[40]	LSTM	Data samples after ADC	The power of NLI, OSNR	Experiment	NLI: The maximum absolute error is less than 1 dB.
	[40]	LSTM	Data samples after ADC	The power of NLI, OSNR	Experiment	OSNR: The maximum absolute error is less than 1 dB.
	[41]	DNN	Optical power spectral density	Optimum power	Experiment	The maximum error is 1.6 dB.
	[42]	ANN	Statics of distribution of the BER	${S N R}_{NLI}$ , OSNR	Experiment	The maximum normalized error is less than 4%.
OSNR	[43]	CNN	2D diagram of the constellation	OSNR	Experiment	The maximum absolute error is 0.7 dB.
	[44]	CNN	3D diagram of the constellation	OSNR	Experiment	The maximum error is less than 0.5 dB.
	[45]	CNN + LSTM	Waveform	OSNR	Experiment	The maximum error is about 1.3 dB.
	[46]	Gaussian process	Launch power, gain of EDFAs, noise figure of EDFAs	OSNR	Field trial	The maximum absolute error is about 1.2 dB
	[47]	ANN	Launch power, gain of EDFAs, noise figure of EDFAs	OSNR	Field trial	The mean absolute error is about 0.6 dB

Target	Reference	AI Algorithm	Input	Output	Evaluation Method	Performance
NLI/OSNR	[49]	ANN	PNC, and aspect ratio of the constellation diagram	${S N R}_{NLI}$ , OSNR	Experiment	NLI: The maximum absolute error is less than 1 dB. OSNR: The maximum error is 1 dB.
	[48]	ANN	ANC, PNC, GN model, link parameters, and signal parameters	${S N R}_{NLI}$	Simulation	NLI: The maximum absolute error is less than 0.8 dB.
	[50]	ANN	PNC	${S N R}_{NLI}$ , OSNR	Experiment	NLI: The maximum absolute error is less than 0.6 dB. OSNR: The maximum error is 0.15 dB.
	[51]	ANN	PNC	OSNR	Experiment	The maximum absolute error is less than 1.5 dB.
	[8]	ANN	GN model and launch power of the signal power	${S N R}_{NLI}$	Simulation	The maximum absolute error is about 0.3 dB.
In-line optical filtering	[31]	ANN	Frequency response of the adaptive equalizer, and the signal spectrum	Filtering penalty	Experiment	The maximum absolute error is less than 0.3 dB.

DSP Module	Available Information
CDC	Digital spectrum
CDC	Accumulated chromatic dispersion
Adaptive equalizer	Filtering effect
	Transceiver I/Q error
	Polarization effects
CPR	Laser phase noise
	Fiber nonlinear phase noise
	Frequency offset
Nonlinear compensation	Nonlinear noise
Clock recovery	Clock mismatch

Soft Failure Type	Number of WSSs	Training MAE/Testing MAE (GHz)
FS	8	0.11/0.12
FS	11	0.09/0.11
FT	8	0.08/0.11
FT	11	0.12/0.13

Target	Reference	ML Algorithm	Input	Output	Evaluation Method	Performance
NLI	[40]	LSTM	Data samples after ADC	The power of NLI, OSNR	Experiment	NLI: The maximum absolute error is less than 1 dB.
	[40]	LSTM	Data samples after ADC	The power of NLI, OSNR	Experiment	OSNR: The maximum absolute error is less than 1 dB.
	[41]	DNN	Optical power spectral density	Optimum power	Experiment	The maximum error is 1.6 dB.
	[42]	ANN	Statics of distribution of the BER	${S N R}_{NLI}$ , OSNR	Experiment	The maximum normalized error is less than 4%.
OSNR	[43]	CNN	2D diagram of the constellation	OSNR	Experiment	The maximum absolute error is 0.7 dB.
	[44]	CNN	3D diagram of the constellation	OSNR	Experiment	The maximum error is less than 0.5 dB.
	[45]	CNN + LSTM	Waveform	OSNR	Experiment	The maximum error is about 1.3 dB.
	[46]	Gaussian process	Launch power, gain of EDFAs, noise figure of EDFAs	OSNR	Field trial	The maximum absolute error is about 1.2 dB
	[47]	ANN	Launch power, gain of EDFAs, noise figure of EDFAs	OSNR	Field trial	The mean absolute error is about 0.6 dB

Abstract

Cited By

Figures (25)

Tables (4)

Equations (15)

Journal of Optical Communications and Networking