Author: Lu, Jianing
Title: Quality of transmission estimation and parameter monitoring techniques for next-generation optical networks
Advisors: Lu, Chao (EIE)
Degree: Ph.D.
Year: 2021
Subject: Optical communications -- Quality control
Fiber optics
Optical fiber communication
Hong Kong Polytechnic University -- Dissertations
Department: Department of Electronic and Information Engineering
Pages: xxvi, 157 pages : color illustrations
Language: English
Abstract: Fiber optical communication system is an essential component of our information society. Ever emerging applications and services have provided strong motivations for ultra-long transmission distance and ultra-high transmission capacity for the communication systems. As a result, high-spectral efficiency (SE) coherent fiber optical transmission together with powerful digital signal processing (DSP) technique has become the dominant technical approach by which the received signal can be efficiently processed in the digital domain. To further improve the utilization of the available system capacity, dynamic optical networking should be implemented. This requires accurate, efficient, and low-cost optical performance monitoring (OPM) techniques. One of the key functions of OPM is optical signal-to-noise ratio (OSNR) monitoring. Conventional OPM modules for OSNR monitoring assume the signal is only affected by amplified spontaneous emission (ASE) noise. However, high-speed optical signals suffer more severe impairments with the growing transmission distance and capacity. In particular, fiber nonlinearity becomes non-negligible, and conventional OPM methods must be improved to consider the effect of fiber nonlinearity. Moreover, in flexible optical networks, the optical parameters and impairments resulted from fast-varying and heterogeneous transmission links put forward more stringent requirements on the multi-functionality and affordable complexity of OPM modules. On the other hand, since next-generation optical networks are expected to be highly flexible elastic optical networks (EONs), it is desirable for them to make full use of physical-layer and time resources to maximize network capacity. Key enablers for maximizing the capacity of EONs are dynamic lightpath provisioning, baud rate, modulation format, and shaping factor adaptation based on quality of transmission (QoT) estimation by which the SNR of each channel is predicted. Due to the lack of an efficient and reliable QoT estimator, current optical networks incorporate large operating margins, including unallocated, system, and design margins, in order to ensure that the planned demand capacity as well as the service level agreements (SLAs) are met. This results in network over-dimensioning for both green-field and brown-field scenarios. Therefore, accurate QoT estimation is the key enabler for low-margin optical networking. However, in wavelength-division-multiplexed (WDM) and heterogenous dynamic link environments, the parameter uncertainties and fiber nonlinear distortions render QoT estimations a nontrivial task: the SNR estimation of one channel depends not only on the linear accumulation of ASE noise from inline Erbium-doped fiber amplifiers (EDFAs), but also on the signal power and lightpath of all channels in a complicated manner, and all the parameters are subject to a certain level of uncertainty. How to implement a practical and accurate QoT estimator is a significant research challenge. In view of the above key problems in next-generation optical transmission systems and networks, this thesis will focus on advanced multi-parameter monitoring techniques including the monitoring strategies of modulation format, timing and frequency offset (TO and FO), FO drift (FOD), linear noise and nonlinear noise. Meanwhile, efficient QoT estimators based on analytical model and machine learning (ML) model are investigated and compared in depth.
Firstly, assisted by sparse fast Fourier transform (S-FFT), high-accuracy and low-complexity multi-parameter monitoring techniques for flexible optical networks are proposed to monitor modulation format, TO, FO and FOD. Specifically, a blind and fast modulation format identification (MFI) scheme is proposed with high identification accuracy at low OSNR to identify quadrature phase shift keying (QPSK), 16, 32, and 64 quadrature amplitude modulation (QAM). Then, an accurate and low-complexity TO/FO synchronization scheme based on S-FFT is proposed. The proposed scheme consists of coarse timing/frequency synchronization, fine timing synchronization, and fine frequency synchronization. For the drift feature of FO, an S-FFT-based FOD monitoring scheme is further proposed, by which the optimum performance and the minimum hardware cost of FO estimation (FOE) are achieved simultaneously. In addition, to handle OSNR monitoring accuracy degradation induced by fiber nonlinearity, a modulation-format-transparent and accurate joint linear and nonlinear noise monitoring scheme is proposed based on the calculation of correlation between two spectral components at the upper and lower sideband of the signal spectrum. Different characteristics of flat linear noise spectrum and non-flat nonlinear noise spectrum are used to distinguish the influences on the correlation value from the two noise sources. The performance of the proposed scheme is numerically and experimentally verified in up-to-7 channel WDM systems over a 915 km single mode fiber (SMF) link. Finally, a comprehensive comparative study of QoT estimation for WDM systems using artificial neural network (ANN)-based ML models and Gaussian noise (GN) model-based analytical models is conducted. To obtain the best performance for comparison, all the system parameters for GN-based models are optimized in a brute-force manner. For ML models, the number of neurons, activation function, and the number of layers are optimized. In simulation settings, GN-based analytical models generally outperform ANN models while in experimental settings, however, inaccurate knowledge of various link parameters degrades GN-based models, and ML generally estimates the QoT with better accuracy. However, ML models are temporally less stable and less generalizable to different link configurations. Potential network capacity gains resulting from improved QoT estimators and reduced operating margins are also briefly studied.
Rights: All rights reserved
Access: open access

Files in This Item:
File Description SizeFormat 
5965.pdfFor All Users3.82 MBAdobe PDFView/Open

Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show full item record

Please use this identifier to cite or link to this item: