Green Networking E-Book

• Herausgeber: John Wiley & Sons
• Kategorie: Wissenschaft und neue Technologien
• Sprache: Englisch

Seitenzahl: 329

Veröffentlichungsjahr: 2012

Beschreibung

This book focuses on green networking, which is an important topic for the scientific community composed of engineers, academics, researchers and industrialists working in the networking field. Reducing the environmental impact of the communications infrastructure has become essential with the ever increasing cost of energy and the need for reducing global CO2 emissions to protect our environment.

Recent advances and future directions in green networking are presented in this book, including energy efficient networks (wired networks, wireless networks, mobile networks), adaptive networks (cognitive radio networks, green autonomic networking), green terminals, and industrial research into green networking (smart city, etc.).

Leseprobe

Table of Contents

Introduction

Chapter 1. Environmental Impact of Networking Infrastructures

1.1. Introduction

1.2. Some definitions and metrics

1.3. State of the sites of consumption of the networks: the case of wired networks

1.4. Academic and industrial initiatives

1.5. Perspectives and reflections on the future

1.6. Bibliography

PART 1. A STEP TOWARDS ENERGY-EFFICIENT NETWORKS

Chapter 2. A Step Towards Energy-efficient Wired Networks

2.1. Introduction

2.2. Models of energy consumption

2.3. Energy-saving strategies

2.4. The problem of energy-efficient routing

2.5. Conclusion

2.6. Bibliography

Chapter 3. A Step Towards Green Mobile Networks

3.1. Introduction

3.2. Processes and protocols for green networks

3.3. Architecture and engineering of green networks

3.4. Components and structures for green networks

3.5. Conclusion

3.6. Bibliography

Chapter 4. Green Telecommunications Networks

4.1. Introduction

4.2. Data centers

4.3. Wireless telecommunications networks

4.4. Terrestrial telecommunications networks

4.5. Low-cost and energy-efficient networks

4.6. The role of virtualization in “green” techniques

4.7. Conclusion

4.8. Bibliography

PART 2. A STEP TOWARDS SMART GREEN NETWORKS AND SUSTAINABLE TERMINALS

Chapter 5. Cognitive Radio in the Service of Green Communication and Networking

5.1. Introduction

5.2. Cognitive radio: concept and standards

5.3. Various definitions of green in cognitive radio

5.4. Clean solutions offered by cognitive radio

5.5. Use case: “Smart buildings”

5.6. Conclusion

5.7. Bibliography

Chapter 6. Autonomic Green Networks

6.1. Introduction

6.2. Autonomic networks

6.3. Self-configuring

6.4. Self-optimizing

6.5. Self-protecting

6.6. Self-healing

6.7. Conclusion

6.8. Bibliography

Chapter 7. Reconfigurable Green Terminals: a Step Towards Sustainable Electronics

7.1. Sustainable electronics?

7.2. Environmental impact of electronic products during their lifecycle

7.3. Reduce, reuse, recycle and reconfigure

7.4. Examples of reconfigurable terminals

7.5. Conclusion

7.6. Bibliography

PART 3. RESEARCH PROJECTS ON GREEN NETWORKING CONDUCTED BY INDUSTRIAL ACTORS

Chapter 8. Schemes for Putting Base Stations in Sleep Mode in Mobile Networks: Presentation and Evaluation

8.1. Motivation

8.2. Putting macro base transceiver stations in sleep mode

8.3. Sleep mode in small-cell heterogeneous networks

8.4. Conclusion and considerations on implementation

8.5. Bibliography

Chapter 9. Industrial Application of Green Networking: Smarter Cities

9.1. Introduction

9.2. Smart cities and green networking

9.3. Techniques involved

9.4. Conclusion

9.5. Bibliography

List of Authors

Index

First published 2012 in Great Britain and the United States by ISTE Ltd and John Wiley & Sons, Inc.

Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may only be reproduced, stored or transmitted, in any form or by any means, with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms and licenses issued by the CLA. Enquiries concerning reproduction outside these terms should be sent to the publishers at the undermentioned address:

ISTE Ltd27-37 St George’s RoadLondon SW19 4EUUK

John Wiley & Sons, Inc.111 River StreetHoboken, NJ 07030USA

www.iste.co.uk

www.wiley.com

© ISTE Ltd 2012

The rights of Francine Krief to be identified as the author of this work have been asserted by her in accordance with the Copyright, Designs and Patents Act 1988.

Library of Congress Cataloging-in-Publication Data

Green networking / edited by Francine Krief.

pages cm

Includes bibliographical references and index.

ISBN 978-1-84821-378-4

1. Telecommunication--Energy conservation. 2. Telecommunication--Environmental aspects. 3. Computer networks--Environonmental aspects. 4. Sustainable engineering. 5. Computer networks--Energy conservation. 6. Green technology. I. Krief, Francine.

TK5102.5.G734 2012

384.028′6--dc23

2012028169

British Library Cataloguing-in-Publication Data

A CIP record for this book is available from the British Library

ISBN: 978-1-84821-378-4

Introduction

Information and Communication Technology (ICT) is increasingly omnipresent in most human activities, and is viewed primarily in terms of its contribution to economic productivity and to our wellbeing. However, with the necessity to reduce our overall CO2 emissions to protect our environment, and the constant increase in the cost of energy, the carbon footprint of these technologies has become a cause for concern. Today, the energy impact of ICT is evaluated at 2% of greenhouse gas emissions (equivalent to that of all aviation the world over); in heavily industrialized countries, this figure can reach up to 10%. In addition, the energy consumed by ICT is increasing by about 15-20% each year. The development of new types of applications such as HDTV, new usages such as ubiquitous networks and the explosion in the volume of data traffic on 3G/3G+, and soon to be 4G/LTE networks, would suggest that this consumption is not likely to slow down any time soon.

However, at present, energy consumption is far from being optimized. Communication networks are usually over-dimensioned and designed with redundant capacity. Numerous networking devices consume a considerable amount of energy, even when they are not being heavily used, or not used at all; for instance, this is the case with the Base Transceiver Stations in cellular networks.

The concept of green networking has recently emerged. This encapsulates all approaches and measures employed to reduce the volume of greenhouse gas emissions due to the process of communication.

The objective of this book is to offer an overview of the mechanisms and procedures used for implementing energy-efficient networks and limiting their carbon footprint. Some of these devices are already operating — particularly in mobile networks; others are near the point of becoming operational; finally, some proposals and promising directions for future research are presented.

This book, which is one of the first ever to present a “state of the art” on the advances and research projects in the domain of green networking is made up of nine chapters, and is structured in three parts.

Chapter 1 introduces the problem of reducing the electricity consumption of ICT and particularly for telecommunications infrastructures, given that their CO2 emissions are increasing greatly.

The next three chapters discuss the energy efficiency of communication networks, with each chapter focusing on a particular technology. Together, they make up Part 1 of this book, entitled “A Step Towards Energy-efficient Networks”.

Chapter 2 looks at energy gains in operational wired networks, meaning networks for which the design phase has been completed and the infrastructure is already in place. These communication networks are usually over-dimensioned and built with redundant capacity. For that reason, the opportunities they represent for making energy savings are considerable. Several strategies for optimizing energy consumption are presented in this chapter — in particular, those that lend themselves to energy-efficient routing — which are then evaluated on the basis of “real-world” scenarios.

While the model of the consumption of wired communication links remains rather imprecise, the same is absolutely not true for radio links, which constitute an undeniable hotbed of optimization opportunities, because the processing of a radiofrequency signal is a very energy-hungry operation. In addition, of the various actors involved in telecommunications networks, mobile network operators are the main consumers of energy today.

Chapter 3 discusses the environmental impact of mobile networks and the necessity to reduce their energy consumption. Its primary focus is on the techniques employed to reduce power in the radio access network, given that this is the part which consumes most energy. Two other axes for optimization to reduce greenhouse gas emissions are then presented: the first relates to the architecture and engineering aspects of mobile networks, and the second to the components and structures of these networks.

Chapter 4 introduces the energy consumption of the data centers which form the memory of the Internet, and increasingly its computing power and applications. The focus then moves onto low-cost access networks with low energy consumption. Two solutions would seem to suggest themselves here: femtocells and mesh networks. Finally, the chapter highlights virtualization techniques, which facilitate more appropriate multiplexing and the shutdown of needless hardware resources.

The next three chapters deal with the contribution of new technologies to improve the energy efficiency of networks. They constitute Part 2 of this book, entitled “A Step Towards Smart Green Networks and Sustainable Terminals”. The issue is the anticipated gains, which will become greater still as these new technologies bring more users into the fold.

Chapter 5 discusses cognitive radio networks, which are emerging as a new concept in wireless communications, capable of dealing with the lack of radio resources. Thanks to its agility and capacity to intelligently adapt the parameters of communication, cognitive radio can be exploited to render wireless devices more energy efficient. Several possibilities for exploration are proposed in this chapter.

Chapter 6 applies the concept of autonomic networking to green networks. Green networks would then become capable of self-organization and self-adaptation in order to maintain efficient and environmentally-friendly operation, even in the presence of changeable conditions. Firstly, the four main self-functions are introduced — these are self-configuring, self-optimizing, self-protecting and self-healing. Their respective contributions in the context of green networks are later described and illustrated, taking the particular examples of wireless sensor networks, which by nature are energy-constrained, and smart grids, which contribute to decreasing greenhouse gas emissions.

Chapter 7 studies the environmental impact of communication terminals throughout their lifecycle. This impact, while it is less than that of other industries such as transport, is far from negligible when we consider the particularly high rate of replacement of mobile telephones and smartphones, which is linked to the problem of electronic waste. In addition, in the prediction of a digital society which is more respectful of the environment, it is essential to reduce the environmental impact of electronic products, which will occupy a phenomenally important place in that society in the future. This impact decreases greatly with a longer lifespan of the product. This chapter proposes an interesting avenue for improvement: to design reconfigurable hardware systems in order to increase their functional lifespans. Today, this solution is feasible, thanks in particular to the advent of reconfigurable hardware circuits, and offers better recycling.

The final part of this book, entitled “Research Projects on Green Networking Conducted by Industrial Actors”, is made up of two chapters. The first has been written by a telecoms operator, and the second by a large group specialized in products and systems of ITC.

In the context of the struggle against global warming, mobile operators are looking for ways to reduce the energy consumption of their equipment. However, this reduction must not impact the Quality of Service offered to the customers.

Chapter 8 describes research projects relating to mechanisms for putting Base Transceiver Stations in sleep mode, and their impacts on the overall consumption of mobile networks. This is one of the flagship solutions which drastically reduces energy consumption. Sleep mode is applied to various scenarios for deployment of networks: a conventional cellular network and a heterogeneous network made up of small cells. Sleep mode can offer significant gains in certain use cases, whilst still preserving the Quality of Service perceived by the user.

The smart city can be viewed as an industrial application of green networking.

Chapter 9 presents the concept of a smart city, which involves developing new generations of urban infrastructures heavily influenced by green networking. Indeed, numerous devices will be interconnected — particularly sensors, actuators, video cameras, Base Transceiver Stations, data servers, PCs for command centers, users’ smartphones, etc. The usage of these resources must be optimized by using energy-efficient techniques. This chapter gives an overview of different aspects of green networking which are able to reduce the number or the consumption of the devices produced and the networks put in place: low-consumption communication protocols, assistance with the deployment of a wireless sensor network, low-consumption processor treatments and finally, integration and use of sensors to help in deciding on energy-saving policies. The chapter closes with an example of the use of these technologies to respond to the need for energy management.

Problems linked to green networking are at the heart of research in the field of networks today, and of the societal stakes we find ourselves facing. A great deal of work has still to be done to reduce the consumption of the protocols, communication architectures and networking equipment without damaging the Quality of Service and security.

Taking account of the carbon footprint has to be a constant concern of R&D engineers in telecommunications. However, while ICTs do consume energy, they can also contribute to a reduction in our global CO2 emissions, e.g. by limiting our movements and optimizing the energy consumption of our dwellings and cities.

Chapter 1

Environmental Impact of Networking Infrastructures1

1.1. Introduction

Over the past decade, the issues relating to the cost of information and communication technology (ICT) have increased considerably. These issues stem from a number of disciplines: ecology, economics, politics and societal matters. “Green IT” encapsulates efforts taken to reduce the energy footprint related to ICT, or at the very least to slow its growth.

The environmental cost of ICT is a hot topic, because it is highly controversial. The most alarming estimations associate an impact of around 10% of electricity worldwide for ICT, and 2% of global energy, which is constantly increasing. These measurements are backed up, and projected to increase by around 10% a year over the next ten years [EPA 07]. In France, energy consumption by ICT is estimated to be between 55 and 60 TWh per year, which equates to 13.5% of electricity consumption by the end applications [TIC 08]. To express this as a financial cost, we must remember that the price of 1 kWh in France is €0.10, which is equivalent to around €900,000 for 1 megawatt/year. Yet these figures only take account of the costs of direct usage and the electricity impact, overlooking the aspects relating to the production, transport and recycling of the products.

As the GeSI (Global e-Sustainability Initiative) has shown (see Figure 1.1), the constantly-increasing trend is such that if it continues, in 2020 ICT will produce 1.43 Gtons of CO2, which represents 2.7% of the total carbon footprint, distributed between usage (1.08 Gtons) and the rest (production, transport, recycling — 0.35 Gtons) [GES].

Figure 1.1.Evolution in CO2 emissions from ICT: 2002–2020 [GESI]

It is a necessity to develop an ICT sector which is sustainable for the planet. Indeed, it is inexplicable that a sector as omnipresent as this in the daily lives of citizens should not be concerned with reducing CO2 emissions, controlling energy, and making a firm commitment to collective responsibility. The millions of subscribers to communication and entertainment services using ICT are responsible, and must become active participants in the changes made in (and for) the future. Every connection to an Internet network, every usage of a social network, every request sent to a search engine and every video watched entails a cost — a cost which it is helpful to understand and lay bare, in order to be able to reduce it later on.

Behind the powerful tools and services which today’s society cannot go without, there is a massive infrastructure of processing and communication: the platforms of multinational players in Internet and social networks, and of the banking systems, contain millions of machines distributed between data centers, sometimes organized into a “cloud”; the planet’s communication networks, comprising both wired and wireless technologies (fiber optic, copper, satellite, WiFi, GSM) provide a constantly-increasing throughput of communication to channel texts, images, videos, etc. — ultimately, bits of information. If the predictions of a 60% annual growth in Internet traffic, the majority of which comes from the entertainment sector (online games, higher-resolution audio/video jukeboxes, etc.), are correct, by 2025 there will be a transatlantic traffic of more than 400 Tbit/s.

Network operators are among the most electricity-hungry players in the industry. In 2011, Telecom Italia [TEL] estimated its consumption to represent 1% of Italy’s total electricity consumption (in comparison to 0.7% in 2008). Similarly, British Telecom estimates its share of the UK’s electricity consumption to be 0.7% (2.3 TWh) — a proportion similar to that declared by NTT in Japan. These figures take account of all sites where electricity is expended — from the headquarters to the network infrastructure and the data centers associated with it. For instance, for Telecom Italia’s infrastructure, 65% of the electricity is consumed by the networks (both landline and mobile) and 10% by the data centers. However, the figures do not include consumption by the equipment in the premises of the end users. In France, a study conducted by IDATE [IDA] states the overall energy consumption of the telecoms sector at 8.5 TWh in 2012 (by contrast to 6.7 TWh in 2008). This consumption is distributed between the infrastructure (wired and mobile networks: 46% of which data centers accounted for 6% in 2008), the ADSL boxes in users’ homes (24%) and both hardwired and wireless telephones (18 %), also in the users’ premises. Note that the total consumption of domestic routers in France is estimated at 3.3 TWh in 2012 (40 million boxes).

CO2 emissions due to communications networks are also increasing greatly, as shown by another analysis carried out by the GeSI (Figure 1.2). The part played by broadband networks has burgeoned (from 3% to 14%), the mobile network increases less in terms of percentage (from 43% to 51%), but doubles in terms of its absolute value, whereas the contribution of peripheral infrastructure increases from 12 to 20%. Only narrow band networks are expected to see their emissions decrease in the future, disappearing little-by-little and improving their efficiency to the benefit of other types of network. According to IDATE, every European mobile phone user is responsible for the emission of 17 kg of CO2 every year, while landline and Internet users are responsible for 44 kg of CO2.

Figure 1.2.CO2 emissions due to networks

1.2. Some definitions and metrics

We consider that there are two types of energy consumed by ICT equipment (servers, storage, networks, etc.):

— static energy: energy consumed by the “idle” power supply to the equipment (for instance, a router which is not carrying any data at the time, a server which is not performing any service, etc.);

— dynamic energy: that proportion of the energy consumed by the machine when it is in use.

Schematically, ICT infrastructures are made up of two levels: a hardware level and a software level. Thus, when we are interested in their environmental impact, it is helpful to look, at both these levels, at what can be measured, calculated, improved, sometimes considering the two levels independently, and sometimes combining them.

In order to be able to compare software and hardware infrastructures, we have to define a “meter stick” of sorts — a universal standard by which to measure: the energy consumed is one such meter stick, but it is not enough. Obviously, we have to take account of the work performed by the infrastructure in question, whether it be a computation or communication infrastructure.

Numerous metrics have been put forward and are still being put forward today, which demonstrates that the research discipline which deals with green IT is still in its infancy.

Here, we shall cite some of the metrics that have been proposed, but with no pretense at exhaustivity:

— Joules/bit represents the energy needed to process one bit of information. This metric can serve for computation, storage and communication. Thus, in networking equipment, for each component we can distinguish the energy needed to process and transfer one bit of information;

— PUE (Power Usage Effectiveness) represents the energy efficiency of computation and storage infrastructures such as Clusters and Clouds (data centers in general). This is the ratio between the power injected into the data center and the power used by the machines (thus eliminating energy losses, cooling etc.). Contemporary data centers have a PUE between 1.1 for the best, and 2 for the least efficient. This metric includes the computation, storage and communication aspects for the part measuring the power of the machines;

— CUE (Carbon Usage Effectiveness) considers the impact in terms of CO2 emissions. By contrast with the previous metric, this takes account of the type of energy by which the data center is supplied. It is measured in kilograms of CO2 per kilowatt hour. It is interesting that this is one of the “scarce approaches” which take account of the environmental impact beyond the quantity of electricity consumed. Similarly, metrics such as WUE (Water Usage Effectiveness) and ERE (Energy Reuse Effectiveness) amongst others enable us to compare the infrastructures based on their environmental aspects. Note that, to date, no metric integrating the production and recycling of the components has become available. These metrics (PUE, CUE, ERE, etc.) were presented by the Green Grid consortium [GRE b].

In what follows, our primary focus is on the infrastructures of interconnecting networks. Hence, the metrics which we shall use are mostly related to the energy per se, and to the energy expressed as a function of the number of bits processed.

1.3. State of the sites of consumption of the networks: the case of wired networks

Evaluating the energy consumption of an interconnecting network is no easy task: we have to coordinate a set of measurements of the instantaneous power consumed by each component of the network and its activity time. It is a challenge to measure this power: the quality of the measurement, its precision and its frequency depend on the measuring devices used. Indeed, the power measurement recorded by an external wattmeter measuring the whole of a device, and the measurement recorded component-by-component, will give neither the same degree of precision nor the same quality. Both types of evaluation are to be found in the existing body of literature.

A still more difficult task is to measure the consumption caused by a given particular communication, because we have to take account of the sharing of resources in equipment which is able to process several data flows at the same time.

Studies and projections concerning the energy consumption of networks are few and far between. In 2008, Asami et al. [ASA 08], studying the case of Japan, projected that, even if low-consumption electronic components were used, the energy consumption of IP routers would, in the 2030s, exceed Japan’s total energy-producing capacity (such as it was in 2005).

In 2010, Zhang et al. [ZHA 10] looked at optical networks. Taking the figures from 2009 as a basis, the authors estimated that the energy consumption of optical networks will increase by 120% (× 2.2) between now and 2017.

Bolla et al. [BOL 10] estimated the overall consumption of wired networks in Italy between 2015 and 2020. Their results, summarized in Table 1.1, show that the part of the network in users’ homes represents in total 79% of overall energy consumption (1,947 GWh/year), for 17.5 million access points. This fundamental study clearly shows where efforts in terms of research and awareness-raising are needed most desperately.

Table 1.1.Percentage of energy consumption for the different types of networks

In the same study, the authors focus on optical routers, analyzing the energy needed for the various components of such a router. This study comes in the wake of a previous article [TUC 07] about optical switches. In both cases, we note that:

— the proportion of energy attributable to electrical supply and cooling is around 35%;

— the proportion due to the control level (mainly updating of the routing tables) is 10%;

— the part due to the data level (decoding of the IP header, IP transfer, inputs/outputs, buffers, etc.) represents 55% of energy consumption;

— these percentages do not change whether the technology used is entirely optical or electronic, and in time the difference between optical and electronic technology will decrease.

Given that the largest part of the consumption is due to the processing of the header, it is clearly advantageous to work on this point — e.g. by decreasing the number of hops, decreasing the work in the network core, privileging data flows passing through without staying for any length of time on intermediary nodes, etc.

The energy efficiency of equipment based on CMOS technology increases by a multiplicative factor of 1.65 every 18 months (Dennard’s scaling law). In [KOO 11], the study over decades of energy efficiency shows that pieces of equipment such as servers’ efficiency doubles every 1.57 years: the number of computations per Joule continued to double at this pace between 1949 and 2010. Expressed differently, a hundred-fold decrease is/will be observed in the cost in energy for a fixed workload every ten years. At the same time, processing capacities increase a hundredfold every 10 or 11 years. Ultimately, therefore, the individual electrical power of the machines has remained fairly constant over the past 60 years [AEB 11].

In [BOL 11], the authors show the increase in power of networking machines in view of their performance (see Figure 1.3). While their capacity (measured in Mbits processed per second) was multiplied by 2.5 every 18 months, their energy is multiplied by 1.65 over that same time period.

Figure 1.3.Energy and capacity of routers over time

The traffic predictions associated with networks are difficult to evaluate. Recently, a report from [CISCO] mentioned an explosion in traffic (by 2020), with this traffic being generated by new types of application (HDTV) and new uses (mobile and ubiquitous networks).

Currently, a great many pieces of networking equipment consume a considerable amount of energy even when they are hardly being used (if at all). Thus, the static part of their consumption is very significant in comparison to the dynamic part. The design of network components whose electrical consumption is proportional to their use (traffic, workload, volume of data exchanged, etc.) is one of the ambitious objectives of researchers currently working in this domain.

1.4. Academic and industrial initiatives

For a number of years, governmental, intergovernmental and industrial initiatives have been set up in order to pool “live” strengths on the subject. This section details some of the most noteworthy initiatives in terms of reducing the energy footprint of networks.

One of the most ambitious initiatives is spearheaded by the GreenTouch consortium [GRE d]. GreenTouch is aiming, by 2015, to make recommendations of advances to reduce the energy consumption of networks by a factor of 1,000. This reduction is accompanied by projections of the Quality of Service (QoS) and traffic support in the future. Structured into working groups (focusing on optical core networks, routing and switches, wireless mobile communications, access networks, and services), the consortium draws on the support of academics and industrial actors for approaches which combine hardware and software aspects. A set of focused projects, demonstrators and prototypes (both software and hardware) validate GreenTouch’s proposals.

TREND [TRE] is a “network of excellence” on energy-efficient networking. Funded by the European Commission’s FP7 research program, this cooperative of 12 research centers and industrial actors seeks to quantitatively measure the current and future demand for telecoms infrastructures and design sustainable networks on a reduced scale. Note that a great many of the figures relating to the consumption of networks cited in this chapter are taken from TREND.

The European project ECONET (low Energy COnsumption NETworks) [ECO] focuses on reducing the energy of wireline networks by favoring a set of dynamic technologies (sleep mode and performance adaptation). The objective is to reduce the electrical consumption of network equipment by 50% in the medium term and 80% in the longer term, while preserving the same level of end-to-end Quality of Service.

The COST Action IC0804 [ICO] is an open collaborative action financed by the European program COST as part of FP7. The action focuses on the energy efficiency of large-scale systems. Two specialized working groups on wired networks and wireless networks bring together European researchers working on these topics [PIE 10; PIE 11].

The European project PrimeEnergyIT (2009–2012) [PRI], financed by Intelligent Energy in Europe, is exploring the various technologies (servers, networks, storage, cooling), metrics and test banks, and certifications related to small and medium-scale data centers. In order to encourage thinking about green hardware from the stage of the very design of the data centers, PrimeEnergyIT provides a set of recommendations aimed at those who hold the European public spending purse strings on data centers and computation. PrimeEnergyIT also makes a variety of educational materials freely available.

The Canadian initiative GreenStar [GRE c] proposes to set up a medium-scale experimental network using only “green” energy (solar or wind power). This network links various academic centers the world over, and operates in “best effort” mode: if the energy production conditions (sunshine/wind) allow it, the network is available and transports information. If not, the machines of the network enter sleep mode. An adapted software environment is made available to its users.

1.5.Perspectives and reflections on the future

The reduction of electrical consumption of communication networks can be viewed from different angles: financial, philosophical or environmental. This reduction is part of a more general context, which is that of the effort to limit the use of resources (be they fossil, nuclear or green) so as to reduce the generation of greenhouse gases. Certain researchers believe that the approach chosen to improve energy efficiency is not aggressive enough [BIL]. In any case, the human race is going to have to face a significant increase in global temperature, and we must prepare for that. It seems inevitable that the human factor cannot be ignored in this globalizing approach, and that new Quality of Service paradigms must be put at users’ disposal. “Green networking”, and more generally green IT, must be taken as a necessary innovative factor for research organizations and enterprises [HER 12].

In a globalized green networking approach, it seems probable that “energy-hungry” technological solutions will give way to more energy-efficient solutions (e.g. DSL in comparison to optical networks). In addition, closer interaction between energy providers and massive consumers seems indispensable in order to balance supply and demand. Thus, many researchers see the proposals for “smart grids” as a future axis for green, energy-efficient networks.

1.6.Bibliography

[AEB 11] AEBISCHER B., “ICT and Energy”, ICT for a Global Sustainable Future Symposium, http://www.cepe.ethz.ch/publications/Aebischer_Paradisio_ClubofRome_15-12-11_14-12-11.pdf, December 2011.

[ASA 08] ASAMI T., NAMIKI S., “Energy Consumption Targets for Network Systems”, ECOC 2008, Brussels, Belgium, September 2008.

[BOL 10] BOLLA R., BRUSCHI R., CHRISTENSEN K., CUCCHIETTI F., DAVOLI F., SINGH S., “The potential impact of green technologies in next-generation wireline networks: is there room for energy saving optimization?”, IEEE Communication Magazine, November 2010.

[BOL 11] BOLLA R., BRUSCHI R., DAVOLI F., CUCCHIETTI F., “Energy efficiency in the future internet: a survey of existing approaches and trends in energy-aware fixed network infrastructures”, IEEE Communications Surveys and Tutorials (COMST), 13 (2), May 2011.

[CIS] CISCO, Cisco visual networking index: Forecast and methodology, 2010–2015, 1 June 2011.

[GES] GLOBAL E-SUSTAiNiBiLiTY INITIATIVE (GeSI), SMART 2020: Enabling the low carbon economy in the information age report by The Climate Group on behalf of the e-Sustainability Initiative (GeSI), 2008.

[HER 12] HERZOG C., LEFEVRE L., PIERSON J.M., “Green IT for innovation and innovation for Green IT: the virtuous circle”, Human Choice and Computers (HCC10) International Conference, Amsterdam, September 2012.

[KOO 11] KOOMEY J., BERARD S., SANCHEZ M., WONG H., “Implications of historical trends in the electrical efficiency of computing”, Annals of the History of Computing, IEEE, vol. 33, Issue:3, p. 46–54, March 2011.

[PIE 10] PIERSON J.M., HLAVACS H., Proceedings of the COST Action IC0804 on Energy Efficiency in Large Scale Distributed Systems, 1st Year, IRIT, Toulouse, July 2010.

[PIE 11] PIERSON J.M., HLAVACS H., Proceedings of the COST Action IC0804 on Energy Efficiency in Large Scale Distributed Systems, 2nd Year, IRIT, Toulouse, July 2011.

[TUC 07] TUCKER R., “Will optical replace electronic packet switching”, SPIE Newsroom, 2007.

[TUC 08] TUCKER R.S., BALIGA J., AYRE R., HINTON K., SORIN W.V., “Energy consumption in IP networks”, Optical Communication, ECOC 2008, September 2008.

[ZHA 10] ZHANG YI, CHOWDHURY P., TORNATORE M., MUKHERJEE B., “Energy efficiency in telecom optical networks”, Communications Surveys & Tutorials, IEEE, 12 (4), 2010.

Websites

[BIL] Bill Saint Arnaud. http://green-broadband.blogspot.fr.

[ECO] https://www.econet-project.eu.

[EPA] EPA, US Environmental Protection Agency ENERGY STAR Program, Report to congress on server and data center energy efficiency, available online: www.energystar.gov/ia/partners/prod development/downloads/epa datacenter report congress final1.pdf, August 2007.

[GRE a] www.green500.org.

[GRE b] The GreenGrid, www.greengrid.org.

[GRE c] http://www.greenstarnetwork.com.

[GRE d] www.greentouch.org.

[ICO] IC0804, www.cost804.org.

[IDA] http://www.fftelecoms.org/sites/default/files/contenus_lies/007.15_idate_presentation_conference_de_presse.pdf.

[PRI] http://www.efficient-datacenter.eu.

[TIC 08] Rapport TIC et Développement Durable, France, http://www.cgedd.developpement-durable.gouv.fr/IMG/pdf/005815-02_rapport_cle2aabb4.pdf.

[TEL] http://www.telecomitalia.com/content/tiportal/it/innovation/events/conferences/giornata_studio_efficienzaenergeticapercheecome/jcr%3Acontent/rightParsys/linklist/linkdownloadParsys/download_1/file.res/02_Cucchietti_Energia.pdf.

[TRE] www.fp7-trend.eu.

1 Chapter written by Laurent LEFÈVRE and Jean-Marc PIERSON.

PART 1

A Step Towards Energy-efficient Networks

Chapter 2

A Step Towards Energy-efficient Wired Networks1

2.1. Introduction

Whether the phenomenon stems from an increased awareness of the consequences for the environment, from a financial opportunity or from a question of reputation and business, the reduction of greenhouse gas emissions has become a primary objective in recent times. Individuals, companies and governments alike are expending a great deal of energy in reducing the energy expenditure of many sectors of activity. In parallel, information and communications technology (ICT) is increasingly present in the majority of human activities, and it is estimated that 2% of greenhouse gas emissions could be attributed to such technology, with this proportion increasing to 10% in heavily industrialized countries [GLO 07; WEB 08].

While these figures may not seem excessive at present, they will undoubtedly increase in years to come. With the dawn of cloud computing, the computation and communication infrastructures require ever-higher degrees of performance and availability. This necessitates the use of powerful hardware, which consumes a great deal of energy both because of its direct function and also of its need for cooling. In addition, the demands in terms of availability necessitate the design of superfluous setups, built on a gargantuan scale to deal with a peak load. Hence, the infrastructures are often under-used, and adapting their level of performance to the workload actually required of them is a means of optimization that appears promising on a number of levels.

The Internet, for instance, can be represented as a core network, interconnecting multiple heterogeneous access networks. These networks exhibit numerous differences in terms of technologies used, performances expected and workloads. Consequently, they offer different energy-saving opportunities. However, because of the lack of operational data and the never-ending wheel of technological advancement, it is no easy task to characterize the different sources of energy consumption and their causes, and it is impossible to reach a lasting consensus. In 2002, Roth et al. [ROT 02] estimated that local networks, by way of concentrators and switches, were responsible for around 80% of energy consumption by the Internet. In 2005, Nordman and Christensen [NOR 05] attributed half of the total consumption to switch matrix interface cards. In 2009, a study conducted by Deutsche Telekom [LAN 09] predicted that by 2017, the consumption of the core network would have reached the same level as that of the access networks, whereas Bolla et al. [BOL 11] affirm that this consumption ought to remain negligible.

From a strictly environmentalist viewpoint, the objective of green networking is to reduce the volume of greenhouse gas emissions due to the communication process. The use of renewable energy sources or of low-consumption electronics (e.g. induction devices) constitutes an obvious path for improvement. In addition, there are numerous optimization strategies related to the physical design of the infrastructure itself. For instance, it is possible to place the energy-consuming elements (data centers, etc.) close to the points of energy production so as to avoid line losses when transporting energy over long distances. It is also possible to give preference to places where the outside temperature is low all year round, thereby reducing the need for air conditioning by way of simple ventilation.

These strategies may have a significant impact on the actual energy consumption of the infrastructure; yet their influence remains marginal when it comes to the networks. For example, the delocalization of energy-consuming elements imposes constraints on the architecture of the network and alters the volume and the profile of global traffic. It is essentially a question of planning and static optimization. In this chapter, however, we shall only focus on those aspects which have a direct bearing on the dynamic function of the networks, once the design phase has been completed and the infrastructure is in place — that is, on the communication protocols. Similarly to computation infrastructures, communication networks are generally oversized and designed with a great deal of redundant capacity. Oversizing is a natural phenomenon, whereby designers can allow for changes in the volume of traffic due to new usage. In addition, because there is no management of Quality of Service (QoS), the evaluation of the traffic load at any given time is generally carried out on the basis of a measurement or an estimation of the peak traffic. As a result, during periods of low usage, the network is active but under-used, and consumes energy needlessly, even if the traffic profiles are often regular and perfectly well known. For instance, the Website What Europeans do at Night [WED] shows that the traffic experiences peaks during the day and troughs at night. Redundancy is necessary in order to ensure a satisfactory level of reliability and fault tolerance, but necessitates the installation of surplus machines which remain on constant alert in order to take up the baton as soon as they detect a fault. The entire issue of green networking consists of exploring possibilities for optimization while seeking to limit their impact on the QoS or fault tolerance.

In this chapter, we are only interested in forms of optimization that are applicable to a fixed infrastructure network. After presenting various models of energy consumption in section 2.2, we explore different techniques for saving energy both at the level of applications and of infrastructures in section 2.3. Then, as an example, we present a formulation of the problem of energy-efficient routing in section 2.4