Energy Efficient Core Networks with Clouds

The popularity of cloud based applications stemming from the high volume of connected mobile devices has led to a huge increase in Internet traffic. In order to enable easy access to cloud applications, infrastructure providers have invested in geographically distributed databases and servers. However, intelligent and energy efficient high capacity transport networks with near ubiquitous connectivity are needed to adequately and sustainably serve these requirements. In this thesis, network virtualisation has been identified as a potential networking paradigm that can contribute to network agility and energy efficiency improvements in core networks with clouds.
The work first introduces a new virtual network embedding core network architecture with clouds and a compute and bandwidth resource provisioning mechanism aimed at reducing power consumption in core networks and data centres. Further, quality of service measures in compute and bandwidth resource provisioning such as delay and customer location have been investigated and their impact on energy efficiency established. Data centre location optimisation for energy efficiency in virtual network embedding infrastructure has been investigated by developing a MILP model that selects optimal data centre locations in the core network. The work also introduces an optical OFDM based physical layer in virtual network embedding to optimise power consumption and optical spectrum utilization. In addition, virtual network embedding schemes aimed at profit maximization for cloud infrastructure providers as well greenhouse gas emission reduction in cloud infrastructure networks have been investigated. GreenTouch, a consortium of industrial and academic experts on energy efficiency in ICTs, has adopted the work in this thesis as one of the measures of improving energy efficiency in core networks.