Energy efficiency of next generation optical IP networks

© 2012 Dr. Michael Zhe FengWith the continuous growth of energy consumption in modern telecommunications networks, the topic of improving energy efficiency of global telecommunications has been gaining ever-increasing importance not only among the research community, but also across the ICT industry. Current trends indicate that the average annual rate of the Internet Protocol (IP) traffic growth is between 40% and 50% and this rate is not expected to decline any time soon. This growing volume of IP traffic, as a result of the fast-rising number of users and higher bandwidth requirements, will be accompanied by a corresponding increase in both the quantity and capacity of the transmission and switching equipment required in the network. As a result, the energy consumption of the global telecommunications network will also increase. This raises the question of whether the future growth and advancement of the telecommunications networks will ultimately be constrained by energy consumption rather than by bandwidth. Recent studies in the literature have shown that the total power consumption of national broadband telecommunications networks is currently dominated by the access network. But the total power consumption is expected to be dominated by the high-capacity core routers and switches consumption in the core networks as the fibre-based networks continue to expand in size and reach into the future. In this thesis we develop new design methodologies and architectures to reduce power consumption of the core networks, through optimising the network so that more of the traffic is switched at the more energy-efficient layers. We begin by providing a high-level overview of the power consumption by the Internet including the access, metropolitan/edge, and core networks. We then conduct a survey of the literature on existing work pertaining to improving energy-efficiency of core networks. We present a review of power consumption values for representative, commercially available routing and switching equipment deployed in core networks. This survey will contribute towards the development of the energy models in the following chapters. In Chapter 3, we present energy models for a variety of high-capacity core network node architectures, based on different combinations of key routing and switching technologies. We develop algorithms for an energy-efficient coordinated multi-layer switching scheme in the node. Using these algorithms, we investigate the potential of power savings that could be attained by appropriate applications of packet switching, electronic circuit switching and optical circuit switching in a node. In the following chapter, we present network energy models when the various node architectures considered in Chapter 3 are deployed upon a model and practical core network topology. We develop network routing and switching algorithms to optimise the traffic switching at the more energy-efficient layers in the network. We examine the network power savings that could be achieved through a coordinated cross-layer aggregation and bypass mechanism that is enabled by intelligent network automation technologies. These savings are relative to a network using full IP routing. In Chapter 5, we extend the analysis in Chapter 4 to include a range of different network topologies, beyond the model network analysed in Chapter 4. Furthermore, we generalise the model to include different types of traffic matrices representing traffic flows between source-destination pairs in the network. We analyse the impact of applying distance-dependent traffic matrices upon the power savings of the network. We will conclude this thesis by summarising the key findings, along with making suggestions for potential research directions in the area of energy-efficient networking, relating to the work that is presented in this thesis