Abstract
dc:description.abstractIncreasing transmission data rates and bandwidth efficiency urgently need to be solved for optical fibre communication systems to enable the tremendous growth of Internet traffic. Optical communication networks are mainly limited by the power-dependent fibre nonlinearity which limits the capacity that can be achieved in such systems. This PhD thesis addresses the problem of maximising the information throughput from single- to multi-channel point-to-point transmission to optical networks. The first part of this thesis looks into the problem of maximising the spectral efficiency in the nonlinear fibre channel. Constellation shaping has been widely used to approach the theoretical limit of the capacity in the additive white Gaussian noise (AWGN) channel. However, its performance is limited in the nonlinear fibre channel. A novel nonlinearity-tolerant shaping scheme was proposed and demonstrated in this thesis to effectively mitigate the optical Kerr nonlinearity. Expanding optical bandwidth in wavelength-division multiplexing (WDM) systems is an achievable solution to solve the incoming growth in the demand for high data rates. Second, in ultra-wideband (UWB) systems, beyond the conventional 5 THz bandwidth, the inter-channel stimulated Raman scattering (ISRS) in the fibre becomes significant. The launch power optimisation of a point-to-point transmission minimising the impact of the power-dependent nonlinear interference (NLI) for the UWB systems is necessary to meet the future capacity requirement. The launch power optimisation can be more practical by machine learning (ML) based forward and inverse physical layer models which can reduce the computational complexity significantly as shown in the second part of this thesis. Third, the ISRS effect complicates the UWB optical network design and limits the network throughput. The larger margin that allows the network to operate satisfactorily needs to be reduced for the upcoming demand increase. A low-margin sequential loading scheme enabled by the inverse physical layer model is discussed. More network traffic has been shown by the heuristic algorithms performed in the proposed scheme. The network performance can be further improved by optimising the launch power and spectra assignment.
Degree
thesis:*- Name dc:type.qualificationname
- Doctor of Philosophy (PhD)
- Level dc:type.qualificationlevel
- Doctoral
- Grantor dc:publisher.institution
- University of Cambridge
- Year dc:date.issued
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Gan, Zelin
- Advisor dc:contributor.advisor
-
- Savory, Seb
Subjects
dc:subject × 2Rights
dc:rightsIdentifiers
dc:identifier.*- DOI dc:identifier.doi
- https://doi.org/10.17863/CAM.115144
- OAI identifier oai:identifier
- oai:www.repository.cam.ac.uk:1810/378914