Author

Quoc V. Phung

Date of Award

2010

Document Type

Thesis - ECU Access Only

Publisher

Edith Cowan University

Degree Name

Doctor of Philosophy

School

School of Engineering

Faculty

Faculty of Computing, Health and Science

First Supervisor

Professor Daryoush Habibi

Abstract

Along with the ever-increasing volume of Internet and other communication traffic in
today’s networks, the advancement of key technologies such as Wavelength Division Multiplexing
(WDM) or Dense WDM (DWDM) could enable a tremendous improvement in
utilisation of the enourmous bandwidth that optical fibres are capable of providing. For
large-scale transmission deployment, optical networking offers excellent solutions due to
its transmission reliability; with low-bit error rate, massive capacity and high scalability.
Being the backbone of the network and responsible for the distribution of huge amount
of traffic, optical mesh networks are consequently faced with great expectations for the
reliability of the services delivered. The failure of an optical component, such as a fibre cut
or a link-down at a node, often has catastrophic consequences leading to communication
disruption, loss of data, services and profits. For instance, according to Gartner Group,
up to $500 million in US business losses would have been inflicted by network failures experienced
through 2004. A direct voice-call revenue loss from the failure of a major trunk
group is frequently quoted at amounting to more than $100,000 per minute. In another
case reported by the International Cable Protection Committee, a powerful earthquake
in Taiwan in December 2006 caused damage to four major fiber optics lines as well as
nine cables laid undersea. It took 49 days to repair all 21 damaged points. These cases
signify that protection design against network failures has a crucial and imperative role in
telecommunications networks today - being of even more significance in the case of optical
mesh networks. The central theme of this dissertation revolves around this vital research
area.
Protection design, however, can not cover for all scenarios of network failures. Indeed,
the protection expectation of traffic demands is varied. Majority of these, defined as gold
service class, require an average outage period of around 5 minutes per year with full
restorability against single span failures. Others, such as the platinum service class, may
require higher protection against dual span failures; or average outage period of 30 seconds
per year. This thesis first studies the minimum requirements regarding the connectivity of a physical network topology to ensure the success of restoration algorithms to achieve
Multiple Quality of Protection (MQoP). Furthermore, we propose the mathematical formulation
for determining the relationship between any two paths connecting each node-pair.
This principle plays an important role as the core of modeling potential and eligible sets of
disjoint and distinct path candidates for designing backup routes against network failures.
On the other hand, protection design in optical transport networks can be applied at
either Optical Channel sublayer (OCh) known as span-based, or, Optical Multiplex Section
sublayer (OMS), known as path-based. A variety of restoration techniques have been
designed and implemented for mesh networks including span restoration, path restoration,
p-cycles, etc. This thesis studies the design of these protection schemes for supporting
MQoP service classes. We propose a new optimisation model for Shared Backup Path
Protection (SBPP) with the objective of minimising the number of constraints and reducing
the model complexity. This model is much more tractable with large size networks. For pcycle
design, we propose a new mathematical model to construct sufficient a set of p-cycles
to protect failed working channels or flows. Unlike a conventional model which selects the
optimal set of p-cycles from all pre-determined cycles in the network, this approach uses
span-based variables to indicate whether or not a cycle crosses a span. Therefore, the
complexity of the model is significantly reduced in large scale networks where the number
of all cycles exponentially increases with network size.
Finally, it is noted that each of these protection schemes has its own advantages and
disadvantages in terms of complexity, capacity efficiency and restoration time. In addition,
the use of an individual technique for MQoPs may be neither possible or efficient. A
new mathematical model is proposed based on a mixed-protection formulation to support
MQoPs. Despite of the complexity of modeling, this optimisation formulation efficiently
utilises the advantages of each protection scheme with respect to different service classes
and, hence, maximise the efficiency. Consequently, the interaction between service classes
is formulated and investigated to provide the minimum amount of capacity utilisation for a
given network topology. The best mixture of service classes would determine whether or not
the existing physical topology is suitable to support the known or predicted requirements
of traffic demands.

Share

 
COinS