Towards a green optical Internet
thesisposted on 27.02.2017 by Dharmaweera, Madushanka Nishan
In order to distinguish essays and pre-prints from academic theses, we have a separate category. These are often much longer text based documents than a paper.
Increasing electricity consumption significantly affects all nations from environmental, social, and economic perspectives. Among the contributors to this problem is rising power consumption of the Internet’s backbone network due to expanding Internet traffic volumes. This thesis addresses the power consumption minimization problem of the Internet's backbone network. Existing solutions to reduce power consumption are identified and categorized into four approaches of network redesign, traffic engineering, power-aware networking, and load-adaptive operation. Network redesign solutions either modify core nodes by incorporating advanced devices or optimize the physical link topology. Traffic engineering solutions aggregate multiple traffic flows into fewer streams using wavelength/waveband grooming. Power-aware networking solutions vary the operating states of network devices according to traffic variability. Load-adaptive operation solutions use high-rate network devices with Mixed Line Rates (MLR) to serve traffic. To reduce power consumption of an Optical Circuit Switched (OCS) network, each chapter derives solutions from one or more approaches. First, network redesign and traffic engineering solutions are combined to develop a novel sparse grooming algorithm. The proposed algorithm selectively places core routers to reduce power consumption by 688 kW over a traditional network. Second, network redesign and power-aware networking solutions are integrated to develop two novel strategies that use different algorithms to reduce cost and power consumption of a survivable network. Cost is reduced by 40% and power consumption is reduced by 1.3 MWh in a day. Third, the effect of grouping policy and band size and configuration design parameters on power consumption of a waveband-groomed network is analyzed. Proposed algorithms show that power consumption of this Band Switched Network (BSN) can be reduced in different traffic environments by using specific design parameters. Fourth, traffic engineering and load-adaptive operation solutions are combined to develop three new schemes and two algorithms that assign line rates and wavelengths/wavebands in the network, respectively. Power consumption of this MLR-based BSN reduces by 80% and 17% over two Single Line Rate (SLR)-based traditional networks. Lastly, solutions from network redesign and power-aware networking are integrated to develop a novel burst assembly scheme that reduces power consumption of a sleep-mode-enabled Optical Burst Switched (OBS) network by approximately 0.5 kW than when traditional schemes are used.