ArticlePDF Available

Abstract and Figures

Energy consumption is becoming an increasingly important issue throughout the community. For network operators in particular it is a concern as networks expand to deliver increasing traffic levels to increasing numbers of customers. The majority of the energy used by the Internet today is consumed in the access network, and this will continue to be the case for the short-to-mid- term future. Access technologies should thus be a prime focus for energy use mitigation. In this article, we present a detailed analysis of energy consumption in current and future access networks. We present the energy consumption of DSL, HFC networks, passive optical networks, fiber to the node, point-to-point optical systems, UMTS (W-CDMA), and WiMAX. Optical access networks are the most energy efficient of the available access technologies.
Content may be subject to copyright.
IEEE Communications Magazine • June 2011
70 0163-6804/11/$25.00 © 2011 IEEE
This work was supported
by the Australian
Research Council and by
Cisco Systems.
INTRODUCTION
The Internet has revolutionized the way in which
we seek and disseminate information, transact
business, educate, and entertain. Traffic growth
on the consumer Internet has been high and sus-
tained, with growing numbers of Internet cus-
tomers using increasingly sophisticated
applications, and using them more often. The
rollout of broadband access networks has both
facilitated and been driven by these increasing
demands. Service providers and network opera-
tors have invested heavily in deploying and
upgrading these new access networks, investing
as well in large data centers and expanding core
network capacity. In general, these investment
decisions have been driven by the traditional
design metrics of capital cost, operational cost,
and capacity requirements. Energy usage has
always been considered, but in the context of
operational cost rather than as an issue in its
own right. In today’s world, the traditional net-
work design metrics alone are no longer sustain-
able, and energy needs to become one of the
principal design parameters for future networks
and equipment.
It has been estimated that the IT industry
today is responsible for a total of 2 percent of
the electrical energy consumed in a typical Orga-
nization for Economic Cooperation and Devel-
opment (OECD) country [1]. Within this total,
the energy used in the switching, transmission,
and access networks delivering the consumer
Internet today has been estimated to be approxi-
mately 0.5 percent of typical national consump-
tion, with a rising trend as customer traffic levels
increase [2, 3]. Moreover, in the short- to medi-
um-term future, the majority of the total net-
work energy will be consumed in the access
network.
This article reviews the range of access net-
work technologies that might be used as network
operators move to deliver higher-speed customer
access, with a special focus on energy usage as
average customer data rates increase [4]. Wise
technology choices for future access networks
will be an important first step in helping our
industry to meet its challenges in a more energy-
constrained future [5]. We focus here on the
energy consumption of digital subscriber line
(DSL), hybrid fiber coaxial (HFC) networks,
passive optical networks (PONs), fiber to the
node (FTTN), point-to-point optical (PtP) sys-
tems, WiMAX, and Universal Mobile Telecom-
munications System (UMTS) using wideband
code-division multiple access (W-CDMA). We
find that optical access networks are the most
energy efficient of the available access technolo-
gies.
POWER CONSUMPTION MODEL
In this section we describe an energy model of
the access network, and consider the energy con-
sumption of a number of wired and wireless
access technologies. There are several different
access technologies in use today, and more are
in development [4]. Figure 1 is a schematic dia-
gram of the seven access network technologies
we consider here. These technologies include
DSL and HFC networks as well as a number of
high-speed access technologies: PON, FTTN,
PtP, WiMAX, and UMTS. In Fig. 1, thin lines
indicate optical links while thick lines indicate
copper links.
The energy consumption of each access net-
work can be split into three components: the
energy consumption in the customer premises
equipment (i.e., the modem), the remote node
ABSTRACT
Energy consumption is becoming an increas-
ingly important issue throughout the community.
For network operators in particular it is a con-
cern as networks expand to deliver increasing
traffic levels to increasing numbers of customers.
The majority of the energy used by the Internet
today is consumed in the access network, and
this will continue to be the case for the short-to-
mid-term future. Access technologies should
thus be a prime focus for energy use mitigation.
In this article, we present a detailed analysis of
energy consumption in current and future access
networks. We present the energy consumption of
DSL, HFC networks, passive optical networks,
fiber to the node, point-to-point optical systems,
UMTS (W-CDMA), and WiMAX. Optical
access networks are the most energy efficient of
the available access technologies.
ENERGY EFFICIENCY IN COMMUNICATIONS
Jayant Baliga, University of Melbourne and National ICT Australia
Robert Ayre, Kerry Hinton, and Rodney S. Tucker, University of Melbourne
Energy Consumption in Wired and
Wireless Access Networks
BALIGA LAYOUT 5/19/11 8:31 AM Page 70
IEEE Communications Magazine • June 2011 71
or base station (base transceiver station, BTS),
and the terminal unit (which is located in the
local exchange/central office). The per-customer
power consumption Paof all seven access tech-
nologies in Fig. 1 can be expressed in the form
(1)
where PCPE, PRN, and PTU are the powers con-
sumed by the customer premises equipment,
remote node or base station (if there is one),
and terminal unit, respectively. NRN and NTU are
the number of customers or subscribers that
share a remote node and the number of cus-
tomers that share a terminal unit, respectively.
The last term on the right side of Eq. 1 includes
a factor of 1.5 to account for additional over-
heads such as external power supplies, electricity
distribution losses, and cooling requirements in
the building that houses the terminal equipment
[6, 7]. The equipment in the remote node and at
the customer premises is cooled naturally by the
surrounding environment.
In this article, we estimate the energy con-
sumption of a range of access technologies,
based on representative data from manufactur-
ers’ data sheets for commercial equipment.
Table 1 lists commercial equipment for each of
the seven access networks. The equipment listed
in Table 1 is not necessarily best in class for
energy efficiency, but we believe it is representa-
tive of 2010-era access network equipment.
Table 2 lists representative values of the
parameters in Eq. 1 for each of the access
technologies considered here. The number of
users per remote node and terminal unit for
the two wireless technologies (WiMAX and
UMTS) correspond to per-user capacities of
0.25 Mb/s. The number of users per remote
node and terminal unit for the wired technolo-
gies correspond to configurations where the
ports on the remote node equipment and ter-
minal unit are fully occupied. In the following
paragraphs we describe each access technology
and explain the details of the parameters used
in developing an energy model for each access
technology.
DIGITAL SUBSCRIBER LINE
DSL is provided through copper pairs originally
installed to deliver a fixed-line telephone service
[4]. A DSL modem at each customer home con-
nects via a dedicated copper pair to a DSL access
multiplexer (DSLAM) at the nearest central
office (telephone exchange).
For the comparison presented here, we con-
sider a modern ADSL2+ access service. This
technology can in theory provide maximum
speeds of 24 Mb/s downstream to a customer
close to the central office and 1 Mb/s upstream.
However, to account for the typical degrada-
tion in performance due to line length, line
loss, crosstalk, and noise, we assume a maxi-
mum access rate of 15 Mb/s. We consider a
typical DSLAM capable of serving 1008 cus-
tomers, having a full-duplex switching capacity
of 2 Gb/s, and consuming approximately 1.7
kW. The customer modem is modeled as con-
suming 5 W.
HYBRID FIBER COAXIAL NETWORK
Cable distribution networks were initially
deployed to deliver television services, and today
also deliver Internet and telephony services. Typi-
cally, the television program material is compiled
from national and regional sources at a headend
distribution center in each regional city. This
material is distributed on radio frequency (RF)
modulated optical carriers through optical fiber to
local nodes, where the optical signal is converted
into an electrical signal. That electrical signal is
then distributed to customers through a tree net-
work of coaxial cables, with electrical amplifiers
placed as necessary in the network to maintain
signal quality. Hence, these networks are com-
monly termed hybrid fiber coaxial networks.
The electrical signal sent toward the cus-
tomer on the coaxial cable includes an array of
modulated RF carriers representing the individ-
ual television channels, generally extending from
either 50 or 65 MHz up to a frequency of
500–900 MHz, depending on the network. A
reverse channel to the node or head-end is also
provided in the band below 50 MHz. Broadband
PP P
N
P
N
aCPE
RN
RN
TU
TU
=++
15.,
Figure 1. Schematic of network structure with access network options including
digital subscriber line (DSL) and hybrid fiber coaxial (HFC) networks as well
as a number of promising candidates for future high speed access technologies
- passive optical network (PON), fiber to the node (FTTN), point-to-point
optical (PtP), WiMAX and UMTS. Thin lines indicate optical links and thick
lines indicate copper links.
Access network
Modem
Modem
DSLAM
OLT
Ethernet
switch
Ethernet
switch
Ethernet
switch
Ethernet
switch
RF
combiner
DSL
PON
FTTN
using VDSL
PtP
WiMAX
UMTS
HFC
Exchange/
central office
ONU
ONU
Modem
Splitter
DSLAM
Remote node
Modem
Modem
BTS
Modem
Modem
Modem
Modem Modem
OMC
OMC
BTS
Node
BALIGA LAYOUT 5/19/11 8:31 AM Page 71
IEEE Communications Magazine • June 2011
72
Internet access is provided by using one or more
of the downstream RF channels to deliver high-
speed data, and one or more of the low-frequen-
cy reverse channels to send data from the
customer into the network (upstream).
Each television customer has a set-top box,
which demodulates the incoming signal for dis-
play on a television receiver. The data/Internet
customer has a cable modem connected to
his/her computer or network.
The topology of an HFC network is illustrat-
ed in more detail in Fig. 2. As before, thin lines
indicate optical links while thick lines indicate
copper links. In our model for Internet access
via a HFC network, we include:
• “Head-end” equipment, where video RF
carriers are combined in a broadband net-
work platform (BNP) with data-supporting
RF carriers onto transmission fibers
• Field-deployed node equipment, which con-
verts the optical signals into electrical sig-
nals suitable for cable distribution
• A network of electrical RF amplifiers and
splitters, so that each node can support a
number of customers spread over many
streets
In each customer’s premises, a cable
modem; Universal Broadband Routers
(UBRs) are an essential part of the HFC
data network, but in this analysis we focus
on the energy consumption of the access
network and do not include the energy con-
sumption of UBRs in our calculations.
We model the network using current
DOCSIS-based equipment, employing 6 MHz
RF channels and 256-quadrature amplitude
modulation (QAM) to deliver 38 Mb/s per RF
channel. Each node in the cable network receives
four sets of RF data carriers on separate fibers
and the video program carriers on another fiber.
These data carriers are combined with video
program carriers and distributed on four lines of
coaxial cables. Based on the neighborhood topol-
ogy, the cable is branched, the signal re-ampli-
fied, and signals for individual customers tapped
off along the route. We allow for four RF carri-
ers to be assigned to convey data signals on each
of these coaxial cable links, so each cable tree
supports a total of 152 Mb/s.
In the final distribution network link, we
allow 15 customers to be served from a single
electrical line amplifier. When the offered
Table 1. Representative equipment used in model of access networks.
Table 2. Values of access network parameters used.
Terminal unit Remote node Customer premises equipment
ADSL Alcatel Stinger FS+ N/A D-Link DSL502
HFC Motorola GX2 Motorola SG4000 Quad Node
Motorola BLE100 RF Amplifier Motorola SB6120
PON Hitachi 1220 N/A Wave7 ONT-G1000i
FTTN Hitachi 1220 NEC AM3160 NEC VF200F6
PtP Cisco 4503 N/A TC Communications TC3300
WiMAX Cisco 4503 Motorola WAP 450 Series Alvarion BreezeMAX USB 200
Zyxel MAX-200M1
UMTS Cisco 4503 Motorola Horizon 3G-nx Sierra Wireless AirCard USB 306
PTU (kW) NTU PRN (W) NRN PCPE (W) Technology
limit
Per-user
capacity
ADSL 1.7 1008 N/A N/A 5 15 Mb/s 2 Mb/s
HFC 0.62 480 571 120 6.5 100 Mb/s 0.3 Mb/s
PON 1.34 1024 0 32 5 2.4 Gb/s 16 Mb/s
FTTN 0.47 1792 47 16 10 50 Mb/s 2 Mb/s
PtP 0.47 110 N/A N/A 4 1 Gb/s 55 Mb/s
WiMAX 0.47 24400 1330 420 5 22 Mb/s 0.25 Mb/s
UMTS 0.47 15300 1500 264 2 20 Mb/s 0.25 Mb/s
Based on the neigh-
borhood topology,
the cable is
branched, the signal
re-amplified, and sig-
nals for individual
customers tapped
off along the route.
We allow for four RF
carriers to be
assigned to convey
data signals on each
of these coaxial
cable links, so that
each cable tree
supports a total of
152 Mb/s.
BALIGA LAYOUT 5/19/11 8:31 AM Page 72
IEEE Communications Magazine • June 2011 73
capacity per customer is low, the coaxial cable
distribution network requires few nodes to
support many customers and is highly
branched; in such cases we allow one trunk
amplifier to support up to eight line ampli-
fiers. Each node requires at least one video
and one data port on the BNP that combines
the RF signals, and a number of RF data chan-
nels from the UBR. When modeling high data
loads with low oversubscription, several UBRs
may be required in a city.
The BNP consumes 620 W, while serving up
to four nodes. The number of customers served
by a node depends on the number of RF carriers
available for data, in both downstream and
upstream directions, and the per-customer traffic
level. A quad node consumes 256 W; the trunk
and line amplifiers each consume 35 W. In Table
2 the power consumption of the HFC remote
node includes the power consumption of the
node, trunk amplifier, and necessary electrical
amplifiers in a typical installation.
The RF amplifiers, nodes, and head-end RF
combining equipment in the HFC network are
shared between data and broadcast television
services; thus, the energy consumption of this
equipment should be shared between the ser-
vices. On the basis of the subscriber numbers for
each service in one provider’s network, we allo-
cate 40 percent of energy consumption of this
equipment to supporting Internet access.
We have dimensioned the network on the
basis of downstream capacity delivered to cus-
tomers. There are, however, many instances
where the upstream capacity of the reverse
channels may be limited by high ambient RF
noise levels, and this limits the number of cus-
tomers that can be served from a node and
cable network tree. Thus, in assuming a net-
work limited by download capacity, we offer a
conservative (i.e., lower) power consumption
estimate.
PASSIVE OPTICAL NETWORK (PON)
Fiber to the premises installations most com-
monly use a PON technology, in which a single
fiber from the network node feeds one or more
clusters of customers through a passive splitter
[4]. An optical line terminal (OLT) is located at
the central office, and serves a number of access
modems or optical network units (ONUs) locat-
ed at each customer premises. Each customer
ONU in a cluster connects via a fiber to the
splitter, and from there shares the same fiber
connection to the OLT. ONUs communicate
with the OLT in a time multiplexed order, with
the OLT assigning time slots to each ONU based
on its relative demand.
The number of customers that share a con-
nection to an OLT is generally 32 or 64. For the
network energy model, we consider a gigabit
PON (GPON) access network, providing asym-
metric 2.4 Gb/s downstream, 1.2 Gb/s upstream
from the ONU to the OLT and 32 customers
sharing a connection to an OLT. The OLT
equipment shelf is capable of supporting 32
GPON lines (1024 customers), has a backhaul
capacity of 16 Gb/s, and draws 1.34 kW. The
splitter is unpowered. The ONU is a basic model
providing only data connectivity, and draws 5 W.
FIBER TO THE NODE USING VDSL
Fiber-to-the-node (FTTN) technology makes use
of existing copper pairs [4]. Dedicated fiber is
provided from a network switch to a DSLAM in
a street cabinet close to a cluster of customers,
and high-speed copper pair cable technologies
such as very-high-speed DSL (VDSL) or
ADSL2+ are used for the final feed to the cus-
tomer premises. This accommodates the distance
limitations of high-speed pair-cable technologies,
and enables high-speed broadband service deliv-
ery without the cost of providing new cable entry
to the customer premises.
In an FTTN network using VDSL, a remote
node houses a VDSL DSLAM which communi-
cates with several homes through the copper
wire and connects back to an Ethernet switch in
the central office/local exchange through a fiber
link. A typical VDSL2 line card supports 16 cus-
tomers and consumes approximately 42 W. An
additional 5 W is consumed for an ONU to link
the remote VDSL DLSAM to an OLT back in
the central office. The VDSL2 customer modem
consumes 10 W and has a peak access rate of 50
Mb/s. The Ethernet switch has 116 optical Giga-
bit Ethernet ports and 64 Gb/s of switching
capacity, with a power consumption of 474 W.
For the model, we dimension the fiber backhaul
capacity to suit the customer traffic level, but set
upper limits on number of customers per Ether-
net switch at 1792, and the maximum number of
DSLAMs per Ethernet switch at 112. The four
remaining ports on the Ethernet switch are used
to provide backhaul capacity.
POINT-TO-POINT ACCESS OPTICAL NETWORK
The highest access speed is achieved using a
dedicated fiber between each customer premises
and the network terminal unit in a PtP configu-
ration [4]. The customer premises employs an
optical media converter (OMC) to convert
between the electrical signal used inside the
home and the optical signal used in the access
network.
For typical central office equipment we con-
sider an Ethernet switch providing 116 optical
Gigabit Ethernet ports and 64 Gb/s of switching
capacity, with a power consumption of 474 W.
An OMC at each home converts between the
electrical signal used in the home network to an
optical signal for transmission over fiber, and
consumes 4 W. In this architecture there is no
remote node. Each Ethernet switch connects to
110 homes, with the six remaining ports used to
provide backhaul capacity.
Figure 2. Layout of an HFC network.
Node
Video plus data
Data
Video
program
CMTS/
UBR
Node
Node
RF
combiner
ModemModem
BALIGA LAYOUT 5/19/11 8:31 AM Page 73
IEEE Communications Magazine • June 2011
74
WIMAX
WiMAX is a high-speed wireless access technol-
ogy. WiMAX was initially designed to provide
fixed-point or nomadic wireless access services,
but its design standards have since been amend-
ed to support full mobility. In our model, we
focus on the use of WiMAX in a stationary set-
ting, where each home uses an indoor modem to
connect to a base station. The WiMAX base sta-
tion is remotely located and uses fiber or point-
to-point wireless backhauls to connect to the
metropolitan and edge network. The area cov-
ered by a base station is referred to as a cell, and
users in a cell share the total available band-
width. Per-user bandwidth can be significantly
increased by creating multiple sectors in a cell
through the use of directional antennas. In a
three-sector configuration, each antenna covers
a 120° sector.
WiMAX provides access rates of up to 70
Mb/s under ideal conditions. However, typically
in urban areas there is not a clear line of sight
between the user and the base station, and the
combination of reduced signal level and multi-
path interference limits access speeds to about
35 Mb/s at distances up to about 7 km, with
degraded speeds at higher distances.
For the comparison in this article, we model
the base station at the remote antenna site as
using a point-to-point fiber link to communicate
to an upstream Ethernet switch. For the base
station we assume a dual-antenna multiple-input
multiple-output (MIMO) system with three sec-
tors and mast-mounted power amplifiers. Each
sector is modeled as providing 35 Mb/s in total
to all users in the sector, and the total base sta-
tion consumption is 1330 W. The fixed point
indoor customer premises unit may be a stand-
alone modem or a USB key style modem. Stand-
alone modems typically achieve higher
throughputs than USB modems but also con-
sume more power. We model the home modem
as consuming an average of 5 W to account for
the diversity of possible devices in a given cover-
age area.
At low average per-user traffic levels, all
users within a cell coverage area will receive
adequate service, subject to propagation con-
ditions. At higher average per-user traffic
levels, fewer users could be adequately served
by each base station sector, and either more
sectors or more base stations would be need-
ed. This leads to a rapid increase in the
equipment power consumption at higher traf-
fic levels.
UMTS
UMTS is a cellular mobile system that can pro-
vide high-speed broadband access capability. It
includes radio access to a base station, and
from there connections to the core networks
for data and voice. For this model we adopt
the more commonly used W-CDMA variant,
and focus on the broadband data access com-
ponent of the network. Users may connect to a
base station through their mobile phone, USB
modem, or standalone modem. The base sta-
tion is often located remote from its network
access controller, and uses fiber or point-to-
point wireless backhaul to connect to the con-
troller. Through the radio network controller a
mobile user can connect to other mobile
phones, the public switched telephone network,
or the Internet. As with WiMAX, capacity can
be greatly increased through the use of multi-
ple sectors.
The spectral efficiency of UMTS was greatly
increased through the introduction of high-speed
downlink packet access (HSDPA), high-speed
uplink packet access (HSUPA), and, most
recently, evolved high-speed packet access
(HSPA+). HSPA+ allows for theoretical down-
link speeds of 42 Mb/s and uplink speeds of 11
Mb/s. However, typically interference in urban
areas limits downlink speeds to about 30 Mb/s
and uplink speeds to about 6 Mb/s. For our
energy consumption model, the base station con-
nects via Ethernet to an upstream switch, and
from there to the radio network controller. A
typical outdoor base station consumes 1.5 kW
and supports three sectors. Each sector has an
average downlink throughput of 22 Mb/s. The
user modem is a USB modem that consumes
less than 2 W.
OVERSUBSCRIPTION
We characterize the capacity available to each
customer by the headline access rate advertised
and sold to customers by the Internet service
provider (ISP). However, backhaul networks
connecting the access network to the
metropolitan and edge networks are dimen-
sioned by network operators to provide some
lower worst-case minimum transmission rate to
every customer, taking advantage of the bursty
nature of customer Internet traffic. The ratio
of the advertised access rate to this minimum
per-user rate is referred to as the oversubscrip-
tion rate. Although the oversubscription rate
applied by network providers is typically much
higher for wireless access networks than for
wired access networks, to facilitate a fair com-
parison we model the same across all access
networks. Note that as the use of the consumer
Internet for streaming real-time services
increases, high oversubscription ratios will
become unsustainable.
We model each access network in terms of a
headline access rate of AMb/s per customer and
an oversubscription rate M. During the busiest
period of the day, the minimum capacity avail-
able to a customer is A/M, the per-user capacity.
Statistical multiplexing typically occurs at the
DSLAM in ADSL, at the OLT in PON and
FTTN, at the UBR in HFC, at the small Ether-
net switch in the case of PtP, and at the base sta-
tion switch for WiMAX and UMTS.
MARKET SHARE AND TAKE-UPRATE
In many markets, customers choose from a
range of Internet access options; for example,
customers may be able to choose between DSL,
HFC and UMTS network providers. A network
provider, when building an access network, esti-
mates the percentage of households that will
buy the service in the short to medium term,
referred to as the take-up rate, but will also
Although the over-
subscription rate
applied by network
providers is typically
much higher for
wireless access
networks than for
wired access
networks, to
facilitate a fair
comparison we
model the same
across all access
networks.
BALIGA LAYOUT 5/19/11 8:31 AM Page 74
IEEE Communications Magazine • June 2011 75
install additional capacity to cater for future
growth in take-up.
In markets with regulated competitive access
to infrastructure such as pair cable, a similar but
slightly different parameter is market share. Com-
peting ISPs commonly install DSL equipment in
the same area, and customers can purchase ser-
vices from a number of infrastructure-based ISPs,
each of which has slightly overprovisioned to
cater for future growth.
We combine these factors into one parame-
ter and refer to it as underutilization. The power
consumption of current networking equipment
does not typically scale with utilization [8];
therefore, underutilization decreases the energy
efficiency of equipment. To accommodate
underutilization, we increase the power con-
sumption of all access network equipment,
except the customer premises equipment, by 25
percent. This increase in power consumption
corresponds to network equipment utilization of
80 percent.
ENERGY CONSUMPTION
We use the model described earlier to calculate
the total per-customer power consumption for
typical deployments of each of the seven access
networks illustrated in Fig. 1. We also use the
model to project the future energy consumption
of these access networks. The power consump-
tion of each of the access networks has been cal-
culated for a range of “headline” access service
rates, with a constant oversubscription factor of
20. That oversubscription figure is low in situa-
tions where customers predominantly use tradi-
tional web services such as email and browsing,
but could be considered high for future scenar-
ios which include mass use of real-time video on
demand services.
For wired access technologies such as ADSL
and VDSL, we assume that all customer access
ports are fully occupied. For technologies with a
shared access resource such as HFC, wireless,
and PON, we again assume that all physical
ports are utilized, but in addition we share the
resource among as many customers as could be
served at the particular average service rate and
oversubscription factor. As service rates
increase, fewer customers can be served, and
more equipment (base stations, HFC nodes,
PON linecards, etc.) must be provisioned, with
an increase in the per-customer power con-
sumption.
POWER CONSUMPTION PER USER
Figure 3 is a plot of the per-customer power
consumption of each access technology as a
function of the headline access rate. Note here
that this access rate is the provisioned per-user
capacity multiplied by the oversubscription rate.
The technology used in Fig. 3 for all access rates
is the 2010-era technology described earlier.
From Fig. 3, it is clear that at low access rates
(< 1 Mb/s) PON, DSL, and HFC have similar
power consumption. At such rates, the overall
power consumption is dominated by the con-
sumption of the customer modem. In addition,
the power consumption of current networking
equipment does not typically scale with utiliza-
tion [8], resulting in very low efficiencies at low
utilization. At an access rate of 1 Mb/s, all five
wired access technologies are significantly under-
utilized. At these low rates, WiMAX and UMTS
can flexibly share capacity among a very large
number of users, and thus can achieve high effi-
ciency and utilization. Increasing the access rate
from 1 to 10 Mb/s increases the power consump-
tion of WiMAX and UMTS by a factor of two
and four, respectively, as fewer customers can
use a given radio channel or base station and
more resources must be provisioned to deliver
the service. The power consumption of HFC ser-
vices increases at a slower rate. At access rates
greater than 10 Mb/s, wired access technologies
are significantly more energy-efficient than wire-
less access technologies. HFC, DSL, and FTTN
all reach technology limits in the 10–100 Mb/s
range. For average service rates of a few to tens
of megabits per second envisaged for mass cus-
tomized streaming services such as high-defini-
tion video on demand, the PON has a clear
energy advantage.
COMPARISON OF POWER CONSUMPTION IN THE
HOME AND THE NETWORK
In ADSL, HFC, PON, and FTTN the customer
modem or ONU consumes over 65 percent of
the total power in the access network. These
units would normally operate continuously, but
the power consumption of the access network
could be significantly reduced through the use
of automated sleep modes in customer premis-
es network equipment [9]. Assuming the Inter-
net is used on average 8 h/day, automated
sleep modes in customer premises equipment
could reduce the energy consumption of the
access network by up to 40 percent. Additional
savings in power consumption could be real-
ized by fast/micro sleep modes, where cus-
tomer premises equipment enters a sleep mode
during periods of inactivity that are shorter
than a second.
Figure 3. Power consumption of DSL, HFC, PON, FTTN, PtP, WiMAX, and
UMTS as a function of access rate with an oversubscription rate of 20. The
technology used is fixed at 2010 vintage for all access rates.
Access rate (Mb/s)
101
UMTS
DSL PON
PtP
HFC
FTTN
WiMAX
2010 technology
5
0
Power per user (W)
10
15
20
25
30
102103
100
BALIGA LAYOUT 5/19/11 8:31 AM Page 75
IEEE Communications Magazine • June 2011
76
IMPROVEMENT IN
ENERGY EFFICIENCY WITH TIME
Improvements in complementary metal oxide
semiconductor (CMOS) and optical technology
should lead to energy efficiency improvements in
future generations of network equipment. For
example, the energy efficiency improvement rate
of Ethernet switches and OLTs is approximately
10 percent per annum [2, 10]. In this section we
estimate the overall rate of improvement of each
access network technology over time. To esti-
mate this improvement rate we first break down
the total power consumption of each item of
network equipment into four subsystems: elec-
tronic, optical, power amplification, and AC/DC
power conversion. We then apply standard esti-
mates of improvement rates (given below) to
these subsystems. We calculate the rate of
improvement of each item of network equip-
ment as the sum of the improvement rates of its
subsystems, weighted by the proportion of total
power consumed by that subsystem. This tech-
nique of estimating improvement rates is similar
to the analysis performed in [10]. Table 3 lists
estimates of the breakdown of total power con-
sumption of each item of network equipment
into these four subsystems. The per-annum
“business as usual” improvement rates for these
subsystems are:
• Electronics (26 percent)
• Optical interfaces (5 percent)
• Power conversion (0 percent)
• Power amplifiers (0 percent)
Figure 4 is a plot of the per-customer power
consumption for each access technology as a
function of time (bottom horizontal axis) and
access rate (top horizontal axis). This plot is one
scenario for future power consumption of each
of the seven access network technologies, using
the established “business as usual” efficiency
improvement trends outlined in the previous
paragraph. For this plot, the access rate is set at
5 Mb/s in 2010 and increases by 42 percent per
annum (double every two years), reaching 167
Mb/s in 2020. As before, the oversubscription
rate is 20. Although some ISPs today advertise
access rates of 100 Mb/s or more, the oversub-
scription rate used in those networks is typically
much greater than 20. The power consumption
curves for DSL, HFC, and FTTN cease prior to
2020 because we believe these technologies have
a limited ability to scale and meet future increas-
es in bandwidth requirements.
As shown in Fig. 4, we forecast that, if elec-
tronics and optics continue to improve at current
rates, a lack of improvement in power amplifiers
and power conversion subsystems will result in
an overall diminishing rate of improvement in all
access technologies. The power consumption of
HFC and UMTS falls by only 50 percent because
the majority of power consumption in these
access networks is in power amplifiers, which
have limited scope to improve in the future. The
results in Fig. 4 suggest that the per-user power
consumption of most high-speed access tech-
nologies (PON, PtP, FTTN, and WiMAX)
should fall by around 70 percent from 2010 to
2020. Wireless technologies will continue to con-
sume at least 10 times more power than wired
technologies when providing comparable access
rates and traffic volumes. PON will continue to
be the most energy-efficient access technology.
CONCLUSION
We have presented a model of energy consump-
tion of current and future access networks using
published specifications of representative com-
mercial equipment. We analyzed the energy con-
sumption of DSL, HFC, PONs, FTTN,
point-to-point optical systems, UMTS (W-
CDMA), and WiMAX. Passive optical networks
and point-to-point optical networks are the most
energy-efficient access solutions at high access
rates.
ACKNOWLEDGEMENT
The authors gratefully acknowledge the valuable
discussions with B. Gathercole of the University
of Melbourne and thank him for his assistance
with assembling the data for the energy con-
sumption of wireless networks.
REFERENCES
[1] M. Webb et al., “Smart 2020: Enabling the Low Carbon
Economy in the Information Age,” London, 2008.
[2] J. Baliga et al., “Energy Consumption in Optical IP Net-
works,” J. Lightwave Tech., vol. 27, no. 13, July 2009,
pp. 2391–403.
[3] R. S. Tucker et al., “Evolution of WDM Optical IP Net-
works: A Cost and Energy Perspective,” J. Lightwave
Tech., vol. 27, no. 3, Feb. 2009, pp. 243–52.
[4] P. Chanclou et al., “Overview of the Optical Broadband
Access Evolution: A Joint Article by Operators in the IST
Network of Excellence E-Photon/One,” IEEE Commun.
Mag., vol. 44, Aug. 2006, pp. 29–35.
[5] J. Baliga et al., “Energy Consumption in Access Net-
works,” Opt. Fiber Commun./National Fiber Optic Engi-
neers Conf. (OFC)/(NFOEC), San Diego, CA, Feb. 2008.
[6] U.S. Environmental Protection Agency, EPA Report on
Server and Data Center Energy Efficiency, Aug. 2007.
[7] J. Koomey et al., “Network Electricity Use Associated
with Wireless Personal Digital Assistants,” ASCE J. Infra-
structure Systems, vol. 10, Aug. 2004, pp. 131–37.
[8] L. Ceuppens, “Planning for Energy Efficiency — Net-
working in Numbers,” OFC)/NFOEC, San Diego, CA,
Mar. 2009.
Table 3. Breakdown of power consumption.
Electronics Optical
interfaces
Power
amplifiers
Power
conversion
Modem (DSL/HFC) 70% 0% 10% 20%
RF amplifiers 0% 0% 80% 20%
Node (HFC) 0% 20% 60% 20%
BNP 20% 60% 0% 20%
ONU 70% 10% 0% 20%
OMC 50% 30% 0% 20%
Modem
(WiMAX/UMTS) 60% 0% 40% 0%
BTS (WiMAX) 69% 0% 11% 20%
BTS (UMTS) 53% 0% 27% 20%
BALIGA LAYOUT 5/19/11 8:31 AM Page 76
IEEE Communications Magazine • June 2011 77
[9] M. Gupta, S. Grover, and S. Singh, “A Feasibility Study
for Power Management in LAN Switches,” Int’l. Conf.
Network Protocols, Oct. 2004, pp. 361–71.
[10] O. Tamm, C. Hermsmeyer, and A. M. Rush, “Eco-Sus-
tainable System and Network Architectures for Future
Transport Networks,” Bell Labs Tech. J., vol. 14, no. 4,
Feb. 2010, pp. 311–27.
BIOGRAPHIES
JAYANT BALIGA (jbaliga@ee.unimelb.edu.au) received his
B.Sc. degree in computer science and B.E. degree in electri-
cal and electronic engineering (with first class honors) in
2007 from the University of Melbourne, Australia. He is
currently working toward a Ph.D. degree in electrical engi-
neering at the same university. His research interests
include energy consumption, optical network architectures,
and wireless communications.
ROBERT W. A. AYRE received his B.Sc. degree in electronic
engineering from George Washington University, Washing-
ton, DC, in 1967, and B.E. and M.Eng.Sc. degrees from
Monash University, Melbourne, Australia, in 1970 and
1972, respectively. In 1972 he joined the Research Labora-
tories of Telstra Corporation, working in a number of roles
primarily in the areas of optical transmission for core and
access networks, and in broadband networking. In 2007 he
joined the ARC Special Centre for Ultra-Broadband Net-
works (CUBIN) at the University of Melbourne, continuing
work on networking and high-speed optical technologies.
KERRY HINTON received an Honors Bachelor of Engineering in
1978, an Honors Bachelor of Science in 1980, and a Master
of Science degree in mathematical sciences in 1982, all
from the University of Adelaide. He was awarded a Ph.D. in
theoretical physics from the University of Newcastle Upon
Tyne, United Kingdom, and a Diploma in industrial rela-
tions from the Newcastle Upon Tyne Polytechnic in 1984.
In the same year he joined Telstra Research Laboratories
(TRL), Victoria, Australia, and worked on analytical and
numerical modeling of optical systems and components.
His work has focused on optical communications devices
and architectures, physical layer issues for automatically
switched optical networks (ASONs), and monitoring in all-
optical networks. He was also a laser safety expert within
Telstra. In 2006 he joined the ARC Special Centre for Ultra-
Broadband Information Networks, Australia, at the Univer-
sity of Melbourne, where he is undertaking research into
the energy efficiency of the Internet and optical communi-
cations technologies.
RODNEY S. TUCKER [S’72, M’75, SM’85, F’90] received his B.E.
degree in electrical engineering and Ph.D. degree from the
University of Melbourne, Victoria, Australia, in 1969 and
1975, respectively. He is currently a Laureate Professor at
the University of Melbourne, where he is director of the
Institute for a Broadband-Enabled Society and the Centre
for Energy-Efficient Telecommunications. He is a Fellow of
the Australian Academy of Science, a Fellow of the Aus-
tralian Academy of Technological Sciences and Engineer-
ing, and a Fellow of the Optical Society of America. In
1975 he was the recipient of a Harkness Fellowship by the
Commonwealth Fund, New York. From 1988 to 1990 he
was Editor-in-Chief of IEEE Transactions on Microwave The-
ory and Techniques. From 1991 to 1993, he was with the
Management Committee of the Australian Telecommunica-
tions and Electronics Research Board, and a member of the
Australasian Council on Quantum Electronics. From 1995
to 1999 and from 2009 the present, he is a member of the
Board of Governors of the IEEE Lasers and Electrooptics
Society. In 1995 he was the recipient of the Institution of
Engineers, Australia, M. A. Sargent Medal for his contribu-
tions to electrical engineering and was named an IEEE
Lasers and Electro-optics Society Distinguished Lecturer for
the year 1995–1996. In 1997 he was the recipient of the
Australia Prize, Australia’s premier award for science and
technology, for his contributions to telecommunications.
From 1997 to 2006 he was an Associate Editor of IEEE
Photonics Technology Letters. He is currently Vice-Presi-
dent, Publications of the IEEE Photonics Society. In 2007 he
was the recipient of the IEEE Lasers and Electro-optics Soci-
ety Aron Kressel Award for his pioneering contributions to
high-speed semiconductor lasers.
Figure 4. Expected power consumption of latest generation DSL, HFC, PON,
FTTN, PtP, WiMAX and UMTS equipment as a function of the calendar
year. The base access rate in 2010 is taken as 5 Mb/s.
Time (year)
20122010
10
1
Power per user (W)
100
2014 2016 2018 2020
1052041
UMTS
WiMAX
FTTN
HFC
DSL PON
PtP
Access rate (Mb/s)
83 167
BALIGA LAYOUT 5/19/11 8:31 AM Page 77
... Passive Optical Networks (PONs) are widely utilized as an optimal technology in the access part of the network as well as the cloud domain due to their energy efficiency and high bandwidths that are particularly suited to high bit-rate applications such as streaming, Video on Demand (VoD) and cloud gaming, etc. [12]- [15]. The authors in [15] assessed wired and wireless network infrastructures in terms of their applicability to real-time applications. ...
... Passive Optical Networks (PONs) are widely utilized as an optimal technology in the access part of the network as well as the cloud domain due to their energy efficiency and high bandwidths that are particularly suited to high bit-rate applications such as streaming, Video on Demand (VoD) and cloud gaming, etc. [12]- [15]. The authors in [15] assessed wired and wireless network infrastructures in terms of their applicability to real-time applications. The study in [15] shows that PON is a power efficient access technology due to passive nodes, with the additional advantage of abundant bandwidth. ...
... The authors in [15] assessed wired and wireless network infrastructures in terms of their applicability to real-time applications. The study in [15] shows that PON is a power efficient access technology due to passive nodes, with the additional advantage of abundant bandwidth. ...
Conference Paper
Full-text available
The rapid growth of time-sensitive applications and services has driven enhancements to computing infrastructures. The main challenge that needs addressing for these applications is the optimal placement of the end-users’ demands to reduce the total power consumption and delay. One of the widely adopted paradigms to address such a challenge is fog computing. Placing fog units close to end-users at the edge of the network can help mitigate some of the latency and energy efficiency issues. Compared to the traditional hyperscale cloud data centres, fog computing units are constrained by computational power, hence, the capacity of fog units plays a critical role in meeting the stringent demands of the end-users due to intensive processing workloads. In this paper, we first propose a federated fog computing architecture where multiple distributed fog cells collaborate in serving users. These fog cells are connected through dedicated Passive Optical Network (PON) connections. We then aim to optimize the placement of virtual machines (VMs) demands originating from the end-users by formulating a Mixed Integer Linear Programming (MILP) model to minimize the total power consumption. The results show an increase in processing capacity and a reduction in the power consumption by up to 26% compared to a Non-Federated fogs computing architecture.
... The energy consumption of ACN compose of three parts i.e. the energy customer premises equipment consumed, the energy consumed by the base station (BS) or remote node and the energy consumed by the terminal unit (J. Baliga, 2011). To evaluate the per-customer power consumption Pa of the access technologies in fig. 1 (1) ...
... Where: PCPE = Power consumed by the customer premises equipment, PRN = Power consumed by the remote node or BS, and PTU = Power consumed by the terminal unit The additional factor of 1.5 is the provision for the extra power supplies, electricity distribution losses and cooling requirements (J. Baliga, 2011). Table 2 below depicts typical values of ACN parameters for the indicated standard. ...
... In WiMAX, power consumption of equipment rapidly increase at higher traffic volume and this is due to the fact that at higher average per-user traffic levels, only a limited number of customers could be properly served by a BS sector which will consequently result to either an increase in the number of BSs or increase in the number of sectors (J. Baliga, 2011). Now considering cellular networks such as UMTS which is a cellular mobile system that can also provide high-speed broadband access, radio access is the means of data and voice transmission. ...
Article
Full-text available
Over the years, the world of mobile communications system has brought about the rapid development experienced globally. Mobile communications technology has gone from 1G to 2G to 3G to 4G and now heading to 5G. In this paper, a review of architectures, protocols and technologies that underpin green migration from 4G to 5G has been presented. As the world is moving to green solutions, it is important to have the same solutions for our communication systems. The current systems have an undeniable amount of power consumption and dissipation as presented by figures in this paper, therefore there is a need for the communications system to be greener. In 5G when billions of devices will become connected to the internet then a question of the amount of energy these devices will consume and dissipate will arise. As a result, standards set by governing bodies that derive these systems to become greener have been presented. Finally, the technologies that will underpin the green migration from 4G to 5G will be discussed as well as challenges for the migration from 4G to 5G will be highlighted as well.
... The majority of the energy used by the Internet today is consumed in the access network. Thus access technologies should be a prime focus for energy use mitigation [87]. In Fig. 9, the proposed method's energy consumption is compared to HMMR-OS [43], Liu, et al. [45], and GA [46]. ...
Article
Full-text available
The Internet of Things (IoT) has emerged as a technology that enables information transmission and worldwide connectivity among devices. Fiber-Wireless (FiWi) network blends the high bandwidth capability of an optical access network with the simplicity of a wireless access network, making it a promising network structure for 5G communication's "lastmile" access. The evolution of the FiWi access network must be geared toward technical breakthroughs in daily traffic and evolving IoT traffic. The current investigation concentrates on the routing issue in IoT over the FiWi network. One of the main issues of newly generated networks is reducing efficiency and improving the quality of the requested services. If some network users are rapidly uploading data to the network, the buffer of the middle nodes of the network is filled very soon. The packet loss, reduced throughput, and delays result in lower service quality; therefore, the service quality needs to be controlled. Therefore, one of the commonly discussed issues in this type of network is the design of the efficient routing technique. In existing methods, control and management of the routing of these substrate types are often not considered in increasing the service quality in these types of networks. Due to this issue's NP-hard nature, in this paper, a hierarchical routing technique based on decision-making has been presented and evaluated using a cuckoo search algorithm. With the help of this algorithm, an update is improved, the best paths are found. As a result, this improvement will increase IoT efficiency over FiWi networks and provide a high quality of service. The simulation results on the OPNET software indicated that the proposed method improves packet loss and delay.
... Cloud resources, in particular, must not only be distributed to meet the user-specified QoS criteria via services level agreements (SLAs) but should also minimize energy consumption. [8] [40]. ...
... Therefore, the continuous growth in next generation of applications that require high bandwidth have led to a wide deployment of optical access networks. Besides the high bandwidth, Passive Optical Networks (PONs) are highly energy efficient due to unpowered components such as the Arrayed Waveguide grating Router (AWGR), coupler/splitter, and Fibre Bragg gratings used [12]- [16]. ...
Preprint
Full-text available
Recent years have witnessed a remarkable development in communication and computing systems, mainly driven by the increasing demands of data and processing intensive applications such as virtual reality, M2M, connected vehicles, IoT services, to name a few. Massive amounts of data will be collected by various mobile and fixed terminals that will need to be processed in order to extract knowledge from the data. Traditionally, a centralized approach is taken for processing the collected data using large data centers connected to a core network. However, due to the scale of the Internet-connected things, transporting raw data all the way to the core network is costly in terms of the power consumption, delay, and privacy. This has compelled researchers to propose different decentralized computing paradigms such as fog computing to process collected data at the network edge close to the terminals and users. In this paper, we study, in a Passive Optical Network (PON)-based collaborative-fog computing system, the impact of the heterogeneity of the fog units capacity and energy-efficiency on the overall energy-efficiency of the fog system. We optimized the virtual machine (VM) placement in this fog system with three fog cells and formulated the problem as a mixed integer linear programming (MILP) optimization model with the objective of minimizing the networking and processing power consumption of the fog system. The results indicate that in our proposed architecture, the processing power consumption is the crucial element to achieve energy efficient VMs placement.
... Passive Optical Networks have shown their ability to satisfy such enduser requirements as well as a potential to absorb future growth. PONs have now become a popular network technology in the access and cloud domains due to their energy efficiency and high bandwidths that are particularly suited to time-sensitive application demands [10]- [13]. Typically, the PON architecture includes passive devices such as Arrayed Waveguide grating Routers (AWGR), couplers/splitters, and possibly Fiber Bragg gratings [14]. ...
Conference Paper
Full-text available
There is a continuous growth in demand for time sensitive applications which has shifted the cloud paradigm from a centralized computing architecture towards distributed heterogeneous computing platforms where resources located at the edge of the network are used to provide cloud-like services. This paradigm is widely known as fog computing. Virtual machines (VMs) have been widely utilized in both paradigms to enhance the network scalability, improve resource utilization, and energy efficiency. Moreover, Passive Optical Networks (PON s) are a technology suited to handling the enormous volumes of data generated in the access network due to their energy efficiency and large bandwidth. In this paper, we utilize a PON to provide the connectivity between multiple distributed fog units to achieve federated (i.e., cooperative) computing units in the access network to serve intensive demands. We propose a mixed integer linear program (MILP) to optimize the VM placement in the federated fog computing units with the objective of minimizing the total power consumption while considering inter- Vmtraffic. The results show a significant power saving as a result of the proposed optimization model by up to 52%, in the VM -allocation compared to a baseline approach that allocates the VM requests while neglecting the power consumption and inter-VMs traffic in the optimization framework.
Article
This paper describes a set of network metrics are helpful to predict behavior of malicious node in wireless network. The Network and internet is the device in which multiple people can communicate with each other through the wired or wireless media. Nowadays, Internet of Things, Mobile, vehicular, and wireless ad hoc networks all merge into one shared network. These networks are often used to send receive confidential data and information. The unauthorized or malicious node misuse these secrete information. With a rise in rogue nodes, network performance will suffer. A rogue node in the network can cause variations in network metrics including the packet dropping percentage, throughput, latency, energy consumption, and average queue duration. This behavior used to identify the malicious node.
Article
Full-text available
We review technologies and architectures for WDM optical IP networks from the viewpoint of capital expenditure and network energy consumption. We show how requirements of low cost and low energy consumption can influence the choice of switching technologies as well as the overall network architecture.
Article
Full-text available
This article examines the widely cited claim that the network electricity use associated with a wireless personal digital assistant PDA is equal to the electricity consumed by a refrigerator. It compiles estimates of the data flows of wireless PDAs and related networks and allocates network and phone system electricity use based on these estimates. It also conducts sensitivity analyses to verify the robustness of these calculations. This analysis demonstrates that the network electricity use associated with a wireless PDA cannot equal that of a typical refrigerator, even under the most extreme assumptions. Our best-estimate case shows network electricity use for wireless PDAs of 0.5 kW • h/year, and therefore claims that wireless PDAs use as much electricity as a refrigerator are too high by more than a factor of 1,000. Even in our upper-limit assessment, the electricity used by a new U.S. refrigerator is about 100 times greater than the network electricity use associated with a wireless PDA.
Conference Paper
Full-text available
As community concerns about global energy consumption grow, the power consumption of the Internet is becoming an issue of increasing importance. In this paper, we present a network-based model of power consumption in optical IP networks and use this model to estimate the energy consumption of the Internet. The model includes the core, metro and edge, access and video distribution networks, and takes into account energy consumption in switching and transmission equipment. We include a number of access technologies, including digital subscriber line with ADSL2+, fiber to the home using passive optical networks, fiber to the node combined with very high-speed digital subscriber line and point-to-point optical systems. In addition to estimating the power consumption of today's Internet, we make predictions of power consumption in a future higher capacity Internet using estimates of improvements in efficiency in coming generations of network equipment. We estimate that the Internet currently consumes about 0.4% of electricity consumption in broadband-enabled countries. While the energy efficiency of network equipment will improve, and savings can be made by employing optical bypass and multicast, the power consumption of the Internet could approach 1% of electricity consumption as access rates increase. The energy consumption per bit of data on the Internet is around 75 mu J at low access rates and decreases to around 2-4 mu J at an access rate of 100 Mb/s.
Article
Full-text available
This article presents an operators' view of the evolution towards broadband optical access networks. First, we describe a possible evolution of the optical access solution for point-to-point and point-to-multipoint architectures. Subsequently, currently available optical access solutions are evaluated and compared. Finally, we consider regulatory issues inside and outside Europe, and conclude by offering a recommendation with respect to regulation
Conference Paper
We present a comparison of energy consumption of access networks. We consider passive optical networks, fiber to the node, point-to-point optical systems and WiMAX. Optical access technologies provide the most energy-efficient solutions.
Article
Technology analysis indicates that silicon technology evolution will not be able to catch up with future global Internet traffic growth rates. As a result, current network and system designs as well as network architectures may need to be revised significantly. This paper provides a detailed analysis of various network element types (e.g., IP routers, Ethernet switches, and synchronous optical network/synchronous digital hierarchy [SDH/SONET] switches) and their components with respect to power dissipation é 2010 Alcatel-Lucent.
Conference Paper
We examine the feasibility of introducing power management schemes in network devices in the LAN. Specifically, we investigate the possibility of putting various components on LAN switches to sleep during periods of low traffic activity. Traffic collected in our LAN indicates that there are significant periods of inactivity on specific switch interfaces. Using an abstract sleep model devised for LAN switches, we examine the potential energy savings possible for different times of day and different interfaces (e.g., interfaces connecting to hosts to switches, or interfaces connecting switches, or interfaces connecting switches and routers). Algorithms developed for sleeping, based on periodic protocol behavior as well as traffic estimation are shown to be capable of conserving significant amounts of energy. Our results show that sleeping is indeed feasible in the LAN and in some cases, with very little impact on other protocols. However, we note that in order to maximize energy savings while minimizing sleep-related losses, we need hardware that supports sleeping.
Planning for Energy Efficiency -Networking in Numbers
  • L Ceuppens
L. Ceuppens, "Planning for Energy Efficiency -Networking in Numbers," OFC)/NFOEC, San Diego, CA, Mar. 2009.
Energy Consumption in Optical IP Networks
  • J Baliga
J. Baliga et al., "Energy Consumption in Optical IP Networks," J. Lightwave Tech., vol. 27, no. 13, July 2009, pp. 2391-403.