There is an increasing awareness and concern about human consumption of fossil fuels, and how its consequences are affecting the climate. Global warming has been proven beyond doubt, and there is a broad consensus among climate scientists that greenhouse gas (GHG) emissions from human activity are causing it. A significant share of these emissions originates in the burning of fossil fuels to generate electricity. Global electricity production has grown about 3% annually in the past, corresponding to a doubling every 23 years on average. Clearly such growth cannot be sustainable if we are to reduce our emissions. In order to counter this trend, we will not only need to change the way we generate electricity by switching to renewable or other low-carbon energy sources, but also reduce the need for production by making the consuming sectors more energy efficient: we should strive to achieve more results with less energy. Information and communications technology (ICT) may not be the first sector that comes to mind when we think about our carbon footprint, but with a share in global GHG emissions that is about the same size as that of the global airline industry, and more potential for growth, it is far from negligible. Indeed, the ICT sector is growing fast, finding its way to many aspects of our everyday lives, and if no efforts are made to improve its energy efficiency, we can expect its share in our footprint to increase further. The general topic of this dissertation is green ICT, and more specifically, green networking. This research direction, which has become increasingly popular over the past decade, is aimed at reducing the GHG emissions from ICT and networking. The general recommendation above about reducing emissions remains applicable here: greening of communication networks can be achieved by combining low-carbon (renewable) energy sources with techniques to make networks more energy efficient, so that less energy is wasted. This work only looks at the use phase electricity consumption (excluding manufacturing and end-of-life disposal), as this makes up the largest portion of the overall carbon footprint of ICT.
Worldwide electricity consumption of communication networks
A first contribution of this dissertation is that it provides a detailed estimate of the contribution of networks to worldwide electricity consumption. Having upto-date estimates for the overall consumption of networks is an important first step towards assessing the impact of introducing new energy-efficient technologies that may reduce this consumption. We present a methodology that was developed to estimate the worldwide use phase electricity consumption of telecom operator networks, office networks and customer premises equipment. The model is described in detail, with particular attention to the modeling of telecom operator networks, as they take up the biggest share. To estimate this part, we use reports on the electricity consumption of telecom operators. The advantage of this approach is that it takes into account overheads and complexities that are hard to capture in a theoretical analysis based on network models. The methodology is subsequently applied to a data set for the years 2007-2012. The results show that the impact of networks on the environment is growing rapidly, at a rate of 10% per year, a growth that is mainly driven by growing numbers of subscribers and increasing bandwidth demands. The growth rate for networks is faster than that of the overall electricity consumption, so the share of networks in global electricity consumption is increasing. By 2012, almost two percent of all worldwide electricity was consumed by communication networks. This corresponds to 350 TWh (350 billion kilowatt hours), which is more than four times the electricity consumption of the whole of Belgium in that same year. These results emphasize the importance of green networking to stabilize or reduce the energy needs of communication networks.
Comparing next-generation optical access technologies
The green networking research that is presented in this dissertation focuses mostly on optical access. The access network is the part of the network that provides a physical connection to the end users through which they can connect to the Internet. In the case of optical access, this physical connection is an optical fiber that transports information by means of light. We look at passive optical network (PON) architectures (which require no active equipment in the field), because they are the most likely candidates for commercialization, with some types already deployed in the field. Several PON implementations exist and were being developed at the time we published this study, with subsequent generations each time offering higher data rates (in response to growing user demands) and longer reach. To avoid these innovations coming at the cost of increased energy consumption, we developed a model to compare candidates for future optical access, and to be able to make a recommendation on which technology is most energy efficient. The model takes into account the properties of each technology, user demands (with access rates up to 1 Gb/s), split ratio optimization (changing the network layout to achieve the required quality of service in the most energy-efficient way possible), dynamic bandwidth allocation (dividing network capacity among users dynamically taking into account their instantaneous needs), and considers a realistic geographical deployment for a major city (to take into account the implications of reach when determining the required amount of equipment). With all these parameters factored in, we can make a fair comparison between technologies by comparing the power per subscriber under various conditions. As such, we can make a recommendation towards network providers, vendors and standardization bodies that wish to offer the most energyefficient technology adapted to the network needs. Choosing the most energy-efficient technology can not only help them to reduce their GHG emissions, but also to reduce their operational expenditures, as energy prices are on the rise. On the other hand, the results also reveal that energy consumption will rise with the introduction of newer, faster next-generation PONs, unless additional measures are taken to reduce their power consumption.
Energy-efficient GreenTouch architecture for optical access
This dissertation also presents a second model to evaluate the power consumption of optical access. The GreenTouch model3 takes into account the metro/aggregation portion of the network as well as the access (to evaluate long reach technologies), and integrates various energy-saving approaches such as sleep modes, virtualization and hardware innovations in a single framework. The model is used to evaluate three scenarios: (1) a baseline optical access & metro architecture for 2010, using the most energy-efficient technologies that were available at that time; (2) a business-as-usual architecture for 2020 in which current trends are projected to continue without special efforts to improve energy efficiency, resulting in a 29-fold energy efficiency improvement compared to the baseline; and (3) a GreenTouch scenario for 2020, integrating energy efficiency improvement techniques that can be ready for commercial deployment by 2020 if energy efficiency is made a primary target for network design. Thanks to this energy-aware design a 257-fold improvement in energy efficiency can be achieved relative to the baseline scenario. Evidently, the footprint of optical access networks can be reduced greatly compared to current levels—rom the order of watts per subscriber to the order of a couple of hundred milliwatts.
Reducing the energy requirements of networks while maintaining their original quality of service is of essential importance, but another scenario may also become relevant in the future. This dissertation introduces a so-called “ost-peak”future scenario, in which we can no longer rely on fossil fuels as our main resource for electricity production (fossil fuels are “ast their peak”, but instead are replaced by alternative energy sources. This is a relatively new research direction in the field of communication networks, therefore our first contribution on the topic is to give an extensive motivation, showing why we expect such a scenario may occur in the near future. We refer to fossil fuel depletion, insecure energy supply and climate change as the main reasons to push a withdrawal from fossil fuels. Secondly, we also assess the impact this will have on energy availability, showing how this may result in a less reliable and constant energy supply than the one we know today, and how temporary energy shortages may result from this. Since this is a relatively new research domain, the effects of such shortages on communication network infrastructures are not yet clearly defined. We therefore formulated a number of research questions, such as: "if we only have a small fraction of the normal operational energy available, what fraction of the network service can we still offer?”A wireless case study for an existing mobile network in a city is included to show one possible practical solution that can make optimal use of limited available energy: in the simulated network, well-chosen network equipment is switched from an active to an energy-saving state to offer the maximal possible service under the given energy constraint. We also describe a framework for future research in this domain, including a basic assessment of the post-peak potential of technologies in access, core, and data centers; and we also propose new research directions that must be explored into if we want ICT infrastructures to be able to cope with post-peak energy limitations.