Resource Allocation and Optimisation of Elastic Optical Networks in Nonlinear Regime

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


Elastic optical networks have emerged as a promising technology for the backbone networks to support increasing popular applications and service such as ultra high-definition video and inter data centre communication. These applications require to accommodate high-capacity and dynamic bandwidth, which pose a significant challenge to the backbone networks. This thesis aims at making the efficient use of the elastic optical networks and increasing the network utilisation subject to the transmission impairments constraints which degrade the quality of the optical signal.

This thesis begins with the quality of transmission model considering the nonlinear impairments and the amplified spontaneous emission noise as the dominant impairments in the elastic optical networks. Based on widely adopted Gaussian noise nonlinear impairments model, a novel load-aware nonlinear impairments estimation method is proposed. The accuracy of the proposed nonlinear model is evaluated, which effectively approximates the real performance when the link occupation exceeds 100 GHz for different signal power spectral densities and fibre span lengths.

Then, various novel resource allocation algorithms based on the proposed load-aware nonlinear impairments estimation strategy are proposed for the case where the traffic requests are sequentially loaded into the network. The proposed optimisation algorithm and its corresponding service reconfiguration functions achieve higher spectral efficiency and higher network throughput compared to the benchmark solutions.

This thesis also investigates the dynamical resource allocation problem in elastic optical networks with service provisioned and expired, utilising the proposed nonlinear impairments model. The proposed solution significantly improves the dynamical service acceptance ratio and yields higher network utilisation. Further, a novel traffic grooming scheme is presented to explore the potential capability of both bandwidth-variable transponders and the network spectrum resources. The proposed traffic grooming scheme shows tremendous transponders saving over the non-traffic grooming benchmark method and better service acceptance ratio over the traffic grooming benchmark.

Finally, a coordinated fibre span power optimisation and ROADM input power management scheme is introduced based on the proposed nonlinear impairments model. The experimental results show that the proposed coordinated power management scheme leads to 50% bit error rate reduction. The benefit of BER reduction is verified in the elastic optical network through extensive simulation, which depicts 15% higher network capacity compared to an ordinary power optimisation method.
Date of Award25 Jun 2019
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorReza Nejabati (Supervisor) & Dimitra Simeonidou (Supervisor)

Cite this