Master Thesis Defense: Yan Ma

Speaker: Yan Ma

Supervisor: Dr. B. Jaumard

Examining Committee: Drs. J. Yang, D. Goswami, A. Hanna (Chair)

Title: A Study on Reactive and Proactive Push-Pull/Make-Before-Break Defragmentation for Dynamic RMSA

Date: Friday, May 31, 2019

Time: 3:30 p.m.

Place: EV 12.163

ABSTRACT

In this thesis, we investigate several defragmentation techniques, with both proactive and reactive triggering strategies, in the context of dynamic Routing, Modulation and Spectrum Assignment (RMSA) in optical flexible networks.

Proactive defragmentation is executed periodically or according to some fragmentation degradation thresholds in order to maintain spectral defragmentation at an acceptable level, the defragmentation is independent of the request connection events. Reactive defragmentation, on the other hand, is performed when a new request is blocked due to insufficient spectral resources. In the context of dynamic traffic in a flexible optical network, we looked into different combinations of proactive/reactive push-pull and make-before-break defragmentations.

Extensive numerical results show that reactive push-pull defragmentation performs quite well in terms of network throughput and request blocking ratio. Consequently, it is efficient in order to improve network throughput. For proactive push-pull defragmentation, we investigated two different triggering events, namely, time-driven and throughput-driven. We observed that both triggering strategies have a good performance on maintaining network status. Throughput-driven strategy performs better when the network is heavily loaded, whereas time-driven strategy is a better option when the network is less loaded.