January/February 2013 (Vol. 17, No. 1) pp. 10-13
1089-7801/13/$31.00 © 2013 IEEE
Published by the IEEE Computer Society
Published by the IEEE Computer Society
|Challenges in ICT Energy Reduction|
|In This Issue|
PDFs Require Adobe Acrobat
Steadily rising energy costs and the need to reduce global greenhouse gas emissions to protect our environment have turned energy into one of the primary technological challenges of our time. Without direct action, a temperature increase of more than 5°C is predicted by the end of this century. 1 In this context, information and communications technology (ICT) is expected to play a major role in reducing worldwide energy requirements by optimizing energy generation, transportation, and consumption.
Recent research, however, also reveals staggering facts about how ICT is becoming a major component of the energy consumption budget. Detailed surveys show that ICT is responsible for roughly 4 percent of world energy consumption, 2 and this percentage is expected to double in the next decade, if things don't drastically change. Other parallel studies worldwide reached similar results. One determined that "the CO2 emission of the ICT industry alone exceeds the carbon output of the entire aviation industry" (see http://greenict.org.uk/sites/default/files/An%20Inefficient%20Truth%20-%20Full%20Report.pdf). A revealing case study in the telecommunications sector comes from Italy, where the incumbent telecommunications operator consumes more than 2 terawatts a year, or roughly 1 percent of the total national electricity demand, second only to the Italian railway system. 3 Similar considerations, and even more pessimistic ones (as in Japan 4 ), also hold for other developed countries.
Challenges in ICT Energy Reduction
So, which ICT sectors most impact global energy consumption? Should improvement efforts be focused on a specific field? Some studies show that no sector dominates ICT consumption, 2 , 5 so we need energy-efficient solutions in all of them, from data centers to network devices and infrastructures to user devices.
Consider data centers: estimates are that in the US, data center electricity consumption reached 1.5 percent of national consumption; the growing demand for the services data centers offer translates into a yearly 12 percent increase in their energy needs. 6 These needs are continuously and exponentially increasing with the advent and generalization of cloud applications (for example, email or photo storage and editing) and virtualization approaches (such as the virtualization of network elements).
Current estimates about the carbon footprint of network elements indicate that on average, each server produces eight tons of CO 2 per year, each PC or laptop produces four tons per year, each router 20 tons, and each Ethernet switch five tons. Studies on the full ICT device life cycle, including manufacturing, the use phase, and disposal, show that the use phase has the largest overall energy costs, accounting for up to 85 percent. 7
In today's telecommunications networks, most power consumption comes from wireless and fixed access networks, whose equipment, while consuming less than that of core networks, exists in much larger numbers. For example, in cellular access networks, one Node B consumes only about 1,500 W per hour, but altogether these devices contribute 80 percent of all mobile network energy consumption. 8 , 9 In fixed access networks, the current trend is to replace copper-based technologies (used in the majority of current ADSL and VDSL accesses) with optical fibers, due to the latter's significantly higher bandwidth, lower power consumption, and reduced sensitivity to increased bit rates. This could suggest that, in the long run, the percentage of energy used in the network bone might become predominant because of router power consumption. Indeed, routers have shown a large augmentation of power consumption as bit rates increase. If the present trend of increasing data rates continues, the energy share of core and aggregation networks will largely increase compared to that of access networks.
As a result of these trends, some projections indicate that next-generation Internet applications will require electricity in amounts that can't be generated or transported to major metropolitan areas. We must thus find our way to a sustainable future Internet.
The current situation has generated a keen interest in energy-saving approaches by all the actors in ICT, who view reducing their networks' energy consumption as an opportunity for cost reduction and an important aspect for promoting their image in the media.
Different solutions are under investigation for data centers, including algorithms to distribute the load so as to free up servers and put them in sleep mode, sensors that can identify which servers to shut down given environmental conditions, and specific physical layouts that reduce the need for cooling. Some ingenious approaches include moving data centers and service farms to colder areas (for example, mountain resorts) — where cooling is less of a problem than it is in large cities, which are normally located in warmer climates — or placing generators where energy is needed, so that no loss is incurred between producer and consumer.
On the network side, operators and providers are pushing their suppliers to produce more energy-parsimonious equipment. One of the most promising techniques for reducing energy consumption in wireless and fixed networks is dynamic stand-by of network equipment. This approach exploits the fact that during low-load periods, a fraction of the deployed equipment (such as cellular base stations) becomes unnecessary and can enter some low-consumption mode. 10 In addition, energy awareness is having a greater impact on how equipment works, and new protocols and algorithms are under study that would reduce energy consumption while jointly satisfying users' quality-of-service expectations.
Naturally, all these techniques can improve the current situation that has resulted from limited attention to the energy issue in network design, planning, and management phases in the past. However, new, comprehensive, energy-aware approaches to networking are necessary for a systematic evolution toward sustainable networks. We must consider the network as a whole, including core networks, wired and wireless access networks, and customer premises network equipment, data centers, and server farms. Advanced approaches must also consider the entire network's energy consumption in the design, planning, and management phases to achieve a comprehensive and sustainable approach to energy-efficient networking.
In this special issue, we aim to provide a snapshot of ongoing efforts toward a global solution to these urgent challenges that will lead, in the end, to a new energy-efficient and sustainable Internet.
In This Issue
Reducing network energy consumption and thus developing a sustainable Internet will require a massive introduction of photonic technologies. In the core network, the foreseeable huge increase in traffic will call for massive capacity, so core networks' energy consumption is expected to dramatically grow and potentially threaten Internet sustainability. In "Hybrid Optical Switching for an Energy-Efficient Internet Core," by Matteo Fiorani, Maurizio Casoni, and Slavisa Aleksic, the focus is the beneficial effects that photonic technologies can bring to this field. Today, core networks use optical transmission, whereas data switching and control information processing are implemented in the electronic domain, whose functions demand significant energy. This article compares the consumption of a traditional core network with that of two hybrid optical switching network solutions that integrate, at different degrees, optical and electronic domains. The results show some interesting performance/savings trade-offs, and that optical technologies allow for large energy savings in return for some performance reduction.
Optical technologies can also be effective at the access network level. In "Improving Energy Saving in Time-Division Multiplexing Passive Optical Networks," by S.H. Shah Newaz, Ángel Cuevas, Gyu Myoung Lee, Noël Crespi, and Jun Kyun Choi, the authors achieve energy savings using sleep modes in optical network units. Interestingly, the proposed architecture combines optical technologies with low-consuming wireless sensor networks that support the information exchange that is needed to properly decide when to activate and deactivate optical devices.
"HetNets Powered by Renewable Energy Sources: Sustainable Next-Generation Cellular Networks," by Giuseppe Piro, Marco Miozzo, Giuseppe Forte, Nicola Baldo, Luigi Alfredo Grieco, Gennaro Boggia, and Paolo Dini, discusses the problem of network sustainability in terms of energy generation. Focusing on cellular access networks, which constitute the most energy-hungry segment of mobile networks and by far the most energy-demanding access technology, the authors investigate feeding base stations in a heterogeneous network with renewable energy sources. In such a network, jointly using several types of base stations, including low-power ones, leads to significant savings and allows for sustainability when combined with renewable sources.
Finally, we shift attention from access networks to data centers, but still consider the effectiveness that using renewable sources can have for sustainability. The contribution "Powering a Data Center Network via Renewable Energy: A Green Testbed," by Kim Khoa Nguyen, Mohamed Cheriet, Mathieu Lemay, Michel Savoie, and Bobby Ho, presents a wide-area testbed designed to virtually migrate data centers between geographically distributed nodes according to the availability of renewable energy. The network obtains global power optimization via a local efficient use of energy — that is, by using energy where it's available.
The articles in this special issue are representative examples of the ongoing effort the scientific community is undertaking, an effort that covers broad topics, technologies, and methodologies, and that witnesses how crucial the issue of sustainability is for the future of the Internet, and of our planet.
This work received funding from the European Commission Seventh Framework Program under grant agreement 257740 (Network of Excellence TREND).
Alberto Conte is a research group leader at Alcatel-Lucent Bell Labs, France. His research interests include the energy efficiency of cellular networks, autonomous self-organizing systems, and wireless network architectures. Conte has MSs in telecommunications engineering from Politecnico di Torino, Italy, and Institut Eurecom, France. Contact him at email@example.com.
Michela Meo is an associate professor at Politecnico di Torino, Italy. Her research interests include green networking, traffic measurement and characterization, and peer-to-peer systems. Meo has a PhD in electronic and telecommunications engineering from Politecnico di Torino. Contact her at firstname.lastname@example.org.