Green Networking - A Literature Survey

Aleksi Penttinen

Green networking is a rapidly expanding area of research. This survey browses through the recent
green networking literature with two aims. First, the survey attempts to sketch the main ideas in
the research on different parts of communications networks. Second, the aim is to recognize areas
where mathematical modeling approaches are applied.
Contents
1 Introduction 1

2 Surveys on green networking 1

3 Energy consumption and measurements 2

4 Core and optical networks 2

5 Access Networks 3
5.1 Wired access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
5.2 Wireless access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

6 User end 4

7 Server end 4

8 Analytical work and models 6

9 Concluding remarks 7
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

1 Introduction
Research on energy-efficient communications networks1 , especially with respect to environmental
concerns is commonly referred to as green networking. First efforts towards green networking can
been traced back to the papers by Gupta and Singh [22] and Christensen et al. [10], but during
the last three years the interest has increased tremendously. This development is clearly visible
in the number of publications made on the topic, cf. the recent surveys [4, 7], and the number of
conferences and workshops organized in this field indicate that the interest has not even peaked
yet (see, e.g., [19] for an updated list). Whereas the research community has traditionally aimed
at maximizing performance of communication systems2 , the paradigm is now rapidly shifting to-
wards how to produce sufficient performance with minimum energy cost. Indeed, overprovisioning
and redundancy – positive qualities in the traditional network design – are inherently conflicting
objectives with green networking. Alongside with the research also the standardization and regu-
lation bodies are gearing up for the energy-aware communications technology of the future. Thus,
the development will fundamentally change the way communication and networking research is
conducted: The viewpoint of sustainability will be present in almost all the future research.
The importance of the area is also recognized within EU. Key projects within FP7 framework
are EARTH [11], OPERA-Net [40] and ECONET [12], and there is also a Network of Excellence
TREND [50] dealing with the energy-efficient communications and networking. European Institute
of Innovation and Technology (EIT) ICT Labs has a thematic action line coined as Smart Energy
Systems & Green ICT [49] in which also the Department of Communications and Networking
(Aalto ELEC) participates actively.
This survey explores the recent developments in the field of green networking. The aim is to
capture the central ideas and approaches to different parts of the networks rather than provide a
comprehensive account of all research. Despite the interest in this field has surfaced only recently,
the amount of work is quickly becoming overwhelming. Focus here is on developments since 2010
with an emphasis on wired networks (there exists a separate account on energy efficiency in wireless
access networks at Comnet). Also several surveys that cover the older work are briefly discussed.

2 Surveys on green networking

Green networking is still at its infancy but it has already received considerably attention. Accord-
ingly, there are several extensive surveys from slightly different viewpoints that summarize the
prior work related to the concept.
A recent survey by Bianzino et al. [4] provides a comprehensive treatment of recent green
networking techniques focusing on wired networking. They classify the existing research into four
categories; (i) adaptive link rate, (ii) interface proxying, (iii) energy-aware interfaces, and (iv)
energy-aware applications. Adaptive link rate studies approach energy savings by scaling down
the transmission speed of an interface (even to zero, i.e. to a sleep mode). Interface proxying refers
to attempts to ”outsource” network operations from CPUs and such to low-power devices on, e.g.,
network interface card or – in a case of a LAN – to a external device, closer to the physical network.
Energy-aware infrastructure is related to the design and management of networks. This includes
1
The environmental impact of network equipment arises from energy consumption during usage phase, although
manufacturing plays also a significant role (e.g., [48]).
2
With the sole exception of prolonging the battery life of wireless end devices in the context of wireless networks,
see [29] for a survey.

1
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

clean-slate designs and retrofitted approaches. The last category, energy-aware applications, calls
for energy-efficient software designs.
Another survey focusing on energy efficiency in the future Internet infrastructure [7] reviews
a representative selection of projects and standardization activities that are ongoing in the area
in addition to providing a detailed account of green networking papers. In their taxonomy, they
have identified three categories of research; (i) re-engineering, (ii) dynamic adaptation, and (iii)
sleeping/standby. Re-engineering aims at dimensioning and optimizing the network elements or
devising completely new efficient equipment designs. Dynamic adaptation refers to modulating
the capacities of packet processing engines and of network interfaces to meet actual requirements.
The last category, sleeping/standby, considers the case where the network elements (or parts of)
are switched off to a sleep state.
A comprehensive survey on the topics in optical networks is provided by Zhang et al. [59]. They
review the recent literature over different domains where optical technology is applied; core, metro,
and access networks. In addition, standardization efforts toward energy efficiency are summarized.

3 Energy consumption and measurements

In a typical ISP/telco network configuration the core network represents about 30% of the power
consumption while access networks weigh 70% [7]. Lange et al. [34] attempts to predict the future
energy consumption distribution of a universal network operator’s broadband telecommunications
(wireline) network. While the home networks are currently responsible for the most part of the
energy consumption, it is envisaged that IP/MPLS core networks and data centers will dominate
the growth in the future.
Measurement of energy efficiency is not necessarily an easy task. The authors in [6] survey
the different energy-efficiency metrics proposed in the literature and show that the results may
substantially depend on the chosen metrics. Measurement of legacy equipment is challenging as
the equipment are not designed to allow measurements and invasive approaches would be require
lot of work and cause interruptions in service. Phillips et al. [43] propose a regression approach to
address the problem.

4 Core and optical networks

General approach to save energy in the Internet core is to route the traffic around some routers
during low traffic periods and put the respective routers or line cards to sleep [22]. Chabarek et
al. [8] argue that energy awareness should be taken into account already in the design phase in
deploying the routers over a set of point-of-presences. They suggest that power-consuming packet
processing operations should be limited to a subset of routers that allow energy efficient operation.
Panarello et al. [41] propose a congestion control technique coupled with a physical layer re-
source/power management primitives which results in a new access router functionality design
(coined as the green router). In a follow-up paper [35], the authors improved the functionality
of the approach by adding a measurement mechanism where the router estimates the minimum
available capacity from source-destination path in the Internet. By scaling the power consump-
tion (assuming adaptive rate scaling) to match the minimum of the available capacity and user
requirements QoS targets can be met with minimum energy consumption.
Vasić and Kostić [54] present a traffic engineering approach that balances the traffic in the
network so that links and routers can be put into a sleep state or, alternatively, the link rates can

2
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

be lowered to save energy. The authors report 21% and 16% sleep ratios for links and routers,
respectively. Xia et al. [55] present a traffic engineering approach to optical backbone networks.
Traffic grooming and optical bypassing are discussed in the context of reducing the energy cost
of the network. In addition, they model the power consumption of a network by an auxiliary
graph that captures transmission/amplification costs as well as conversion costs between optical
and electronic domains and devise a power-aware scheme that produces improved results compared
to plain traffic grooming.
Fisher et al. [16] recognize that in the core network many of the physical links are actually
bundles (e.g., a 40 Gbps link may consist of four OC-192 cables 10 Gbps each). Although the
present technology does not support quick sleep modes for individual line cards (or bringing sleeping
interfaces back up) it is envisaged that such features become commonplace in the near future. Such
a degree of freedom could be utilized to shut down many cables in the bundles during the off-peak
hours resulting in considerable energy savings. The optimization model of the problem is NP-
complete, but the authors propose LP-based heuristics resulting in 79% energy savings on the
Abilene backbone.
Tucker presents a comprehensive analysis of the minimum energy consumption in optical net-
work in a series of two papers, [52, 53]. The approach represents the ideal case where the energy
consumption is limited only by the Shannon bound on receiver sensitivity and depends on modu-
lation format, fiber losses, system length, and noise in optical amplifiers. The first paper focuses
especially on deriving a lower bound for optically amplified transport systems, while the second
studies the network equipment. In [52], Tucker studies the energy-efficiency i.e., the energy per
bit transmitted and show that the energy consumption is minimized by locating repeaters appro-
priately. He derives a lower bound for energy efficiency and points out that the difference of the
practical equipment in use and the bound can be explained by inefficiencies in the equipment. Ac-
cordingly, the key to improving the energy efficiency lies in reducing these inefficiencies. The article
contains also models for spectral efficiency of optical communications and detailed descriptions on
component energy usage. It is expected that as technologies improve, the optimum repeater spac-
ing becomes as large as 10000 km which highlights the need to minimize the number of instances
where conversion is made between optical and electrical formats. The network perspective to op-
tical equipment energy consumption is considered in [53]. Various switching devices are described
by quantitative models and it is concluded that the energy cost of the switching infrastructure is
much larger than that of transport infrastructure. Consequently, the switches and routers should
receive priority in devising energy-efficient technologies.

5 Access Networks
5.1 Wired access
The case of local area networks differs from WANs by the fact that the recent hardware designs
(e.g., by Intel and Broadcom) allow operation in low-power idle modes. In WANs the power usage
remains largely independent of the load as the chassis and the line cards draw the most power. In
the Ethernet low-power mode the idle periods can be optimized, see e.g. [21].
Tsiaflakis et al. [51] model DSL power allocation under crosstalk interference and consider
optimization models to attain four different fairness notions. Also an optimization algorithm
(subgradient type) is proposed to solve the problems.

3
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

5.2 Wireless access

Energy efficiency in wireless access networks has been studied intensively. Originally the focus was
on battery life considerations and multihop wireless networks. Jones [29] provided a comprehensive
survey on energy efficient network protocols for wireless networks already in 2001, but the line of
research is still very active today, cf. [28].
Karray [30] derives analytical models for spectral efficiency and energy efficiency of OFDMA
network. The models allow studying the efficiencies as a function of various parameters such as
propagation exponent and cell radius.
Marsan and Meo [38] study energy savings obtained through coordination of two overlapping
cellular access networks (e.g., of different operators) during the daily peak hours both networks
can be on, but one could be switched off during low traffic and the users of that network roam in
the other. In a related paper, Marsan et al. [37] the authors consider the case in where several
WLAN access points have overlapping coverage. The authors develop a model that can be used
to assess the effectiveness of different policies used to switch off access points. They also propose
two simple policies based on the number of associated and active users, respectively.
While not strictly access networks, Koutitas [32] studies planning of wireless broadcasting
networks (DAB and DVB-T). The transmitter locations were optimized using a genetic algorithm.
He shows that significant energy (and CO2 emission) savings can be obtained in a mountainous
scenario when the network is planned explicitly using a green strategy.

6 User end
In [5] the authors study PC web-browsing and identify significant power consumption related to
e.g. flash players and tabbed browsing. It is likely that there will be further focus on web content
energy consumption and usage behavior. The area of mobile device energy consumption is studied
intensively by several Finnish research groups, cf. [39, 57, 33].
In certain environments even power availability causes problems. In such environments, inno-
vative energy generation means can be applied together with delay/distruption tolerant networking
to provide network access in a sustainable fashion [45].

7 Server end
Massive data centers, facilities used to house servers and associated components, have been built
during the last few years to provide various Internet services and applications. Rising energy prices
and environmental concerns have prompted the industry and governments to scrutinize the energy
consumption of the centers due to their global significance: Although there are no recent ratings
available, older estimations such as 1% of the world energy consumption [31] or 1.5% of the total
U.S. energy consumption [15] illustrate the enormous scale.
In a data center virtually all the input electrical power eventually ends up as heat. A typical
high-availability dual power path data center with N+1 CRAC units operating at a typical load
of 30% design capacity exhibits the power flows shown in Table 1 [46]. Roughly half of the energy
consumption comes from the IT load and the other half from facility functions such as cooling.
While the servers use the most of the IT equipment power share also the network equipment
connecting the servers plays a major role [24].
Naturally also the energy-efficiency considerations of data centers emphasize the facility view-
point. Major international efforts are on-going to harmonize the energy efficiency metrics for data

4
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

IT Equipment 47 %
Chiller 23 %
CRAC/CRAH 15 %
UPS 6%
PDU 3%
Humidifier 3%
Lighting / aux. devices 2%
Switchgear / generator 1%

Table 1: Power flow distribution in a typical data center [46]

centers to enable the owners of data centers to assess and improve their performance. The Green
Grid [18] is a global consortium of IT companies and professionals seeking to improve energy ef-
ficiency in data centers and business computing ecosystems around the globe. The consortium
participates also in a taskforce together with, e.g., U.S., European and Japanese authorities which
has achieved the first steps towards standardized energy-efficiency metrics. The recent memo [17]
describes the progress and identifies a general need for the following metrics:
• IT energy efficiency

• Facility and infrastructure energy efficiency

• Effect of energy re-use and sustainable energy technology

The harmonization attempts have been successful in the second category. The recent agreement
promotes power usage effectiveness (PUE) as the main metric assessing the facility level energy
efficiency. PUE is defined as the total energy of the data center divided by the IT energy con-
sumption (see measurement details in [17]) and it reflects the overhead of supporting components,
especially cooling. A value 1 is perfect, whereas 1.7 is regarded as high [20]. Google reports the
PUE of its data centers (those with actual IT load above 5MW) already below 1.2 [25]. The inverse
of PUE is referred to as Data Center Infrastructure Efficiency (DCiE) and it is generally given as
a percentage value (i.e. 100 % being ideal) [18].
The work in the first and the third categories, however, is still on-going. Several metrics
are in consideration, but so far a general consensus has not been reached. Examples of the IT
efficiency metrics include IT Equipment Efficiency (ITEE) and IT Equipment Usage (ITEU) [26],
and Data Center Energy Productivity (DCeP) Proxies [23]. The key challenge in defining the IT
energy efficiency is to define what is considered as ”useful work” by the IT Equipment, i.e., the
what is the output that could be compared against energy consumption. Examples of the metrics
addressing renewable technologies and energy re-use include, Energy Reuse Effectiveness (ERE)
[14], Carbon Usage Effectiveness (CUE) [13], and On-site Energy Generation Efficiency (OGE)
and Energy Carbon Intensity (ECI) [27].
Recent research has already addressed various aspects of improving the energy efficiency of data
centers, see [58] for a recent survey. Energy management schemes include low-power CPUs, efficient
power supplies, water cooling (or even outdoor air cooling, cf. [42]), switching off network elements
(e.g., [24], [47]) and improved software (e.g., virtualization). A common technological problem is
the current designs are not energy proportional and do not allow addressing the performance/power
trade-off efficiently for varying loads. Ideally equipment would consume no energy at zero load
and reach full power only at maximum load [3]. Dynamic voltage and frequency scaling has been

5
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

a successful energy-saving technique on laptop computers. In a networked setting, e.g., in data

centers, usage of DVFS is more challenging as it may have crucial impact on the response times.
Chen et al. [9] propose means of predicting the response times enabling performance-aware DVFS
for multitier applications.
In data center networks the servers and cooling account for 70% of the energy budget. The
network of the center consumes 10-20% of the its total power. The energy consumption of the
switches is typically such that most of the consumption can be associated with the switch being
powered on and increasing the traffic from zero to maximum load has only a minor effect [36].
The authors in [24] propose an active power management scheme for data center networks which
results up to 50% power savings by switching network elements on and off. In [47] a formal model
of energy-aware routing in data centers is established and solved using efficient heuristics.
Computer systems manufacturers participate in various consortia which aim at standardizing
the performance benchmarks used in trade. In light of recent development in energy costs and
energy-awareness, organization such as Transaction Processing Performance Council, Standard
Performance Evaluation Corporation, and Storage Performance Council have developed test to
measure the energy consumption in computer systems [44].

8 Analytical work and models

One of the starting points of this survey was to study what kind of mathematical modeling work
has been done on the topic. One of the key topics is related to speed scaling, which may turn
out to be important feature in the future communications equipment that are scalable, or energy
proportional [3], in the sense that when the service rate is scaled down also the energy consumption
goes down proportionally.
In this field, important work has been done with respect to queueing models. The trade-off
between response time optimality, fairness and robustness of such systems has been addressed
in detail in [1]. The authors shortest-remaining-processing-time (SRPT) and processor sharing
disciplines in a setting where jobs have random lengths and energy costs the weighted sum of
which is attempted to minimize. Robustness refers to models that utilize known parameters of
future traffic - such a model becomes inherently vulnerable if the parameter values are not correct.
Along the same lines [2] address the question that what is the minimum energy that is required to
keep the network stable? They assume that speed scaling can be used to reduce the link rates to
save energy. They devise different policies for servers to increase the rate according to the queue
length and show scalability results on the maximum queue size. In particular, the worst-case
models are addressed in this body of work.
Besides speed scaling some modeling work has been done in the context of wireless access
networks. Marsan and Meo [38] utilize simple models to assess energy savings in a two-operator
scenario and in [37] overlapping WLANs are considered. However, the models attempt to illustrate
the benefits of the proposed schemes on a coarse scale rather than aim at accurate analysis.
Finally, there are some studies that shed light on the energy consumption in a way that they
support mathematical modeling tasks. Tucker [52, 52] examines the energy consumption of optical
networks in detail and Karray [30] develops analytical models for spectral and energy efficiencies in
OFDMA networks (e.g., LTE). A short general review of energy consumption models at different
levels is given by Xu et al. [56].

6
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

9 Concluding remarks
With wireless end devices most of the low-level network operations are well developed already
and it seems that optical networking is currently best aligned towards development of energy
savings in the near future. With optical networks the whole energy consumption chain is quite
well understood as a whole. Furthermore, optical core connections can be generally viewed as bit
pipes which simplifies the energy efficiency considerations.
Other areas of communications and networking are, however, more fragmented from the green
networking point of view. Piecemeal advances are made on various layers on the protocol stack with
sometimes unclear, or even incompatible, assumptions on other layers both above and below the
focus of development. Such development of particular technology may induce additional difficulties
in some other part of the stack. What is more, the technologies close to the user end have to respond
to more diverse needs and varying conditions compared to optical networking. The effects of user
behavior cannot be neglected as eventually it is the users that generate the traffic.
The problem on green networking should be viewed from a rather general perspective to truly
get to the roots of energy consumption. A holistic approach on energy-efficiency could attempt to
address following questions:

• What information needs to be communicated?

– ”Is that ad really necessary, does it really need a flash player?”

• What are the requirements for the information to be communicated?

• How the information is communicated with minimum energy but still meeting the require-
ments?

At a more detailed level, future research will be seeking advances through piecemeal and dis-
ruptive approaches. Legacy equipment poses a problem that will take a long time to disappear.
Backward compatibility requirements will cast a shadow over the green networking even longer and
limit the scale on which improvements can be made. Piecemeal advances within well-understood
functionalities are welcome as they are most easily implemented in practice. However, it is im-
portant to ensure that the developments do not have side effects in other parts of the energy
consumption chain.
Another promising direction of research is to find out is there a possibility to save energy by
completely new disruptive ideas. This line of research could be approached through mathematical
modeling. With the exception of speed scaling there is almost no work done on this field so far.
A good starting point could be to address the fundamental building blocks of communications,
i.e. queueing and server models to gain understanding on energy-performance trade-off and to
devise energy-optimal operational policies. The challenge lies in determining the relevant energy
consumption models.
Finally, whereas many of the green networking improvements aim at utilizing the temporal
variations of user needs the spatial component has yet to be addressed. This question has a
resemblance to problems in logistics; it is more beneficial to store products locally than deliver
them one-by-one from overseas. By local replication of services and multicasting technologies also
the energy consumption could be addressed.

7
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

References
[1] Lachlan L.H. Andrew, Minghong Lin, and Adam Wierman. Optimality, fairness, and robust-
ness in speed scaling designs. SIGMETRICS Perform. Eval. Rev., 38:37–48, June 2010.

[2] Matthew Andrews, Spyridon Antonakopoulos, and Lisa Zhang. Energy-aware scheduling
algorithms for network stability. In INFOCOM 2011. The 30th Conference on Computer
Communications. IEEE, April 2011.

[3] L.A. Barroso and U. Holzle. The case for energy-proportional computing. Computer, 40(12):33
–37, December 2007.

[4] A. Bianzino, C. Chaudet, D. Rossi, and J. Rougier. A survey of green networking research.
Communications Surveys Tutorials, IEEE, 2010. to appear.

[5] Aruna Prem Bianzino, Anand Raju, and Dario Rossi. Greening the internet: Measuring web
power consumption. IT Professional, 13(1):48 –53, Jan.-Feb. 2011.

[6] Aruna Prem Bianzino, Anand Kishore Raju, and Dario Rossi. Apples-to-apples: a framework
analysis for energy-efficiency in networks. SIGMETRICS Perform. Eval. Rev., 38:81–85,
January 2011.

[7] R. Bolla, R. Bruschi, F. Davoli, and F. Cucchietti. Energy efficiency in the future internet:
A survey of existing approaches and trends in energy-aware fixed network infrastructures.
Communications Surveys Tutorials, IEEE, PP(99):1 –22, 2010.

[8] J. Chabarek, J. Sommers, P. Barford, C. Estan, D. Tsiang, and S. Wright. Power aware-
ness in network design and routing. In INFOCOM 2008. The 27th Conference on Computer
Communications. IEEE, pages 457 –465, April 2008.

[9] Shuyi Chen, Kaustubh R. Joshi, Matti A. Hiltunen, Richard D. Schlichting, and William H.
Sanders. Blackbox prediction of the impact of dvfs on end-to-end performance of multitier
systems. SIGMETRICS Perform. Eval. Rev., 37:59–63, March 2010.

[10] K. Christensen, B. Nordman, and R. Brown. Power management in networked devices. Com-
puter, 37(8):91 – 93, August 2004.

[11] Energy-aware radio network technologies EARTH. https://www.ict-earth.eu/.

[12] Low energy consumption networks ECONET. http://www.econet-project.eu/.

[13] Christian Belady (Ed.), Dan Azevedo, Michael Patterson, Jack Pouchet, and Roger Tipley.
Carbon Usage Effectiveness (CUE): A Green Grid Data Center Sustainability Metric. The
Green Grid, 2010. White Paper 32.

[14] Mike Patterson (Ed.), Bill Tschudi, Otto VanGeet, Jud Cooley, and Dan Azevedo. ERE: A
Metric for Measuring the Benefit of Reuse Energy from a Data Center. The Green Grid, 2010.
White Paper 29.

[15] EPA. U.S. environmental protection agency’s data center report to congress.
http://tinyurl.com/2jz3ft, 2007.

8
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

[16] Will Fisher, Martin Suchara, and Jennifer Rexford. Greening backbone networks: reducing
energy consumption by shutting off cables in bundled links. In Proceedings of the first ACM
SIGCOMM workshop on Green networking, Green Networking ’10, pages 29–34, New York,
NY, USA, 2010. ACM.

[17] Global Metrics Harmonization Task Force. Harmonizing global metrics for data center energy
efficiency. http://tinyurl.com/6jwfwjp, 2011.

[18] The green grid. http://www.thegreengrid.org.

[19] List of GreenIT Conferences, 2011. http://www.greenit-conferences.org/.

[20] Albert Greenberg, James Hamilton, David A. Maltz, and Parveen Patel. The cost of a cloud:
Research problems in data center networks. In ACM SIGCOMM CCR, 2009.

[21] M. Gupta and S. Singh. Using low-power modes for energy conservation in ethernet lans. In
INFOCOM 2007. 26th IEEE International Conference on Computer Communications. IEEE,
pages 2451 –2455, May 2007.

[22] Maruti Gupta and Suresh Singh. Greening of the internet. In Proceedings of the 2003 confer-
ence on Applications, technologies, architectures, and protocols for computer communications,
SIGCOMM ’03, pages 19–26, New York, NY, USA, 2003. ACM.

[23] Jon Haas, Mark Monroe, John Pflueger, Jack Pouchet, Peter Snelling, Andy Rawson, and
Freeman Rawson. Proxy Proposals for Measuring Data Center Productivity. The Green Grid,
2009. White Paper 17.

[24] Brandon Heller, Srini Seetharaman, Priya Mahadevan, Yiannis Yiakoumis, Puneet Sharma,
Sujata Banerjee, and Nick McKeown. Elastictree: saving energy in data center networks. In
Proceedings of the 7th USENIX conference on Networked systems design and implementation,
NSDI’10, pages 17–17, Berkeley, CA, USA, 2010. USENIX Association.

[25] Google (Inc.). Data center efficiency measurements. http://tinyurl.com/6fsch42.

[26] Green IT Promotion Council / Japan. Concept of new metrics for data center energy efficiency.
http://tinyurl.com/4gzg9j2, 2010.

[27] Green IT Promotion Council / Japan. Proposal of OGE and ECI. http://tinyurl.com/66zgvcr,
2011.

[28] Canming Jiang, Yi Shi, Y. Thomas Hou, and Sastry Kompella. On optimal throughput-
energy curve for multi-hop wireless networks. In INFOCOM 2011. The 30th Conference on
Computer Communications. IEEE, April 2011.

[29] Christine E. Jones, Krishna M. Sivalingam, Prathima Agrawal, and Jyh Cheng Chen. A
survey of energy efficient network protocols for wireless networks. Wirel. Netw., 7:343–358,
2001.

[30] M.K. Karray. Spectral and energy efficiencies of ofdma wireless cellular networks. In Wireless
Days (WD), 2010 IFIP, pages 1 –5, October 2010.

[31] Jonathan G Koomey. Worldwide electricity used in data centers. Environmental Research
Letters, 3(3):034008, 2008.

9
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

[32] G. Koutitas. Green network planning of single frequency networks. Broadcasting, IEEE
Transactions on, 56(4):541 –550, December 2010.

[33] E. Lagerspetz and S. Tarkoma. Mobile search and the cloud: The benefits of offloading. In 1st
IEEE PerCom workshop on Pervasive Communities and Service Clouds., 2011. to appear.

[34] C. Lange, D. Kosiankowski, R. Weidmann, and A. Gladisch. Energy consumption of telecom-

munication networks and related improvement options. Selected Topics in Quantum Electron-
ics, IEEE Journal of, PP(99):1 –11, 2010.

[35] Alfio Lombardo, Carla Panarello, and Giovanni Schembra. Achieving energy savings and qos
in internet access routers. SIGMETRICS Perform. Eval. Rev., 38:76–80, January 2011.

[36] Priya Mahadevan, Puneet Sharma, Sujata Banerjee, and Parthasarathy Ranganathan. A
power benchmarking framework for network devices. In Proceedings of the 8th International
IFIP-TC 6 Networking Conference, NETWORKING ’09, pages 795–808, Berlin, Heidelberg,
2009. Springer-Verlag.

[37] Marco Ajmone Marsan, Luca Chiaraviglio, Delia Ciullo, and Michela Meo. A simple analytical
model for the energy-efficient activation of access points in dense wlans. In Proceedings of the
1st International Conference on Energy-Efficient Computing and Networking, e-Energy ’10,
pages 159–168, New York, NY, USA, 2010. ACM.

[38] Marco Ajmone Marsan and Michela Meo. Energy efficient management of two cellular access
networks. SIGMETRICS Perform. Eval. Rev., 37:69–73, March 2010.

[39] Jukka.K Nurminen and Antti.P Miettinen. Analysis of the energy consumption of javascript
based mobile web applications. In 2nd International Conference on Mobile Lightweight Wire-
less Systems MOBILIGHT 2010, pages 124–135, May 2010.

[40] Optimising power efficiency in mobile radio networks (opera-net). http://opera-

net.org/default.aspx.

[41] Carla Panarello, Alfio Lombardo, Giovanni Schembra, Luca Chiaraviglio, and Marco Mellia.
Energy saving and network performance: a trade-off approach. In Proceedings of the 1st
International Conference on Energy-Efficient Computing and Networking, e-Energy ’10, pages
41–50, New York, NY, USA, 2010. ACM.

[42] Mikko Pervilä and Jussi Kangasharju. Running servers around zero degrees. In Proceedings
of the first ACM SIGCOMM workshop on Green networking, Green Networking ’10, pages
9–14, New York, NY, USA, 2010. ACM.

[43] Steven Phillips, Sheryl L. Woodward, Mark D. Feuer, and Peter D. Magill. A regression ap-
proach to infer electricity consumption of legacy telecom equipment. SIGMETRICS Perform.
Eval. Rev., 38:61–65, January 2011.

[44] Meikel Poess, Raghunath Othayoth Nambiar, Kushagra Vaid, John M. Stephens, Jr., Karl
Huppler, and Evan Haines. Energy benchmarks: a detailed analysis. In Proceedings of the
1st International Conference on Energy-Efficient Computing and Networking, e-Energy ’10,
pages 131–140, New York, NY, USA, 2010. ACM.

10
A! Aalto University
Department of Communications and Networking
Survey on green networking
Aleksi Penttinen

[45] T. V. Prabhakar, Akshay Uttama Nambi S.n, H. S. Jamadagni, Krishna Swaroop, R. Venkate-
sha Prasad, and I. IG.M.M. Niemegeers. A novel dtn based energy neutral transfer scheme
for energy harvested wsn gateways. SIGMETRICS Perform. Eval. Rev., 38:71–75, January
2011.
[46] Neil Rasmussen. Electrical efficiency measurement for data centers. APC, 2010. White Paper
154 v.2.
[47] Yunfei Shang, Dan Li, and Mingwei Xu. Energy-aware routing in data center network. In
Proceedings of the first ACM SIGCOMM workshop on Green networking, Green Networking
’10, pages 1–8, New York, NY, USA, 2010. ACM.
[48] Biplab Sikdar. Environmental impact of ieee 802.11 access points: a case study. SIGMETRICS
Perform. Eval. Rev., 38:66–70, January 2011.
[49] Smart energy systems & green ICT. http://eit.ictlabs.eu/action-lines/thematic-action-
lines/smart-energy-systems-green-ict/.
[50] Towards real energy-efficient network design TREND. http://www.fp7-trend.eu.
[51] Paschalis Tsiaflakis, Yung Yi, Mung Chiang, and Marc Moonen. Fair greening for dsl broad-
band access. SIGMETRICS Perform. Eval. Rev., 37:74–78, March 2010.
[52] R. S. Tucker. Green optical communications—part I: Energy limitations in transport. Selected
Topics in Quantum Electronics, IEEE Journal of, PP(99):1 –16, 2010.
[53] R. S. Tucker. Green optical communications—part II: Energy limitations in networks. Selected
Topics in Quantum Electronics, IEEE Journal of, PP(99):1 –14, 2010.
[54] Nedeljko Vasić and Dejan Kostić. Energy-aware traffic engineering. In Proceedings of the
1st International Conference on Energy-Efficient Computing and Networking, e-Energy ’10,
pages 169–178, New York, NY, USA, 2010. ACM.
[55] Ming Xia, M. Tornatore, Yi Zhang, P. Chowdhury, C. Martel, and B. Mukherjee. Greening
the optical backbone network: A traffic engineering approach. In Communications (ICC),
2010 IEEE International Conference on, pages 1 –5, May 2010.
[56] Jue Xu, Chuang Liu, Yang Yang, Xiaohu Ge, and Tao Chen. An overview of energy efficiency
analytical models in communication networks. In Wireless Communications and Signal Pro-
cessing (WCSP), 2010 International Conference on, pages 1 –6, October 2010.
[57] Bin Yu, Le Wang, and Jukka Manner. Energy-efficient web access on mobile devices. In The
2010 International Symposium on Energy-aware Computing and Networking, in conjunction
with the 2010 IEEE/ACM International Conference on Green Computing and Communica-
tions (GreenCom-2010), December 2010.
[58] Hang Yuan, C.-C.J. Kuo, and I. Ahmad. Energy efficiency in data centers and cloud-based
multimedia services: An overview and future directions. In Green Computing Conference,
2010 International, pages 375 –382, August 2010.
[59] Yi Zhang, P. Chowdhury, M. Tornatore, and B. Mukherjee. Energy efficiency in telecom
optical networks. Communications Surveys Tutorials, IEEE, 12(4):441 –458, 2010.

11