Year

2015

Degree Name

Doctor of Philosophy

Department

School of Electrical, Computer and Telecommunications Engineering

Abstract

The Information and Communication Technology (ICT) industry is responsible for a non-negligible proportion of the world's global energy consumption. Unfortunately, the energy usage of the Internet will continue to rise due to the rapid growth in the number of connected devices and bandwidth intensive applications. In addition, networks are nowadays over-provisioned to provide redundancy and to preserve Quality of Service (QoS) during peak periods. This means current networks are designed to handle the “worst-case” scenario in terms of failures and traffic demands. In other words, they are not designed to be energy efficient. However, many works have shown that traffic exhibits diurnal patterns that correspond to business hours and weekends. This observation and the dire need to conserve energy have spurred intense research efforts into green approaches that consolidate traffic onto the minimal number of links/switches/routers during off-peak periods. In particular, researchers have designed many green or energy-aware traffic engineering (TE) techniques that jointly optimize energy savings and QoS constraints such as maximum link utilization, and end-to-end delays.

This thesis contains a number of novel green TE techniques. First, it studies green TE techniques in Multi-Protocol Label Switching (MPLS) networks. Approaches in this area aim to establish as many arriving Label Switched Path (LSP) requests as possible while utilizing the minimum number of links/routers. However, no one has quantified these LSP establishment approaches in terms of the number of accepted LSP requests and the resulting energy savings. Therefore, this thesis studies six heuristics and proposes a novel metric that considers both energy savings and acceptance rates. In addition, it proposes a simple heuristic that selects paths that contain already established links and use the fewest number of new links.

FoR codes (2008)

0805 DISTRIBUTED COMPUTING

Download

Share

COinS
 

Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.