![Figure 1. a) Power consumption of a typical wireless cellular network (source: Vodafone); b) CO2 emissions per subscriber per year as derived for the base station and mobile handset, after [1]. Embodied emissions arise from the manufacturing process rather than operation.](/figures/figure-1-a-power-consumption-of-a-typical-wireless-cellular-2ksuvpci.webp)
![Figure 5. a) Direct wireless link with average channel knowledge; b) direct wireless link with instantaneous feedback of channel conditions; c) relay link with average channel knowledge; d) relay link with instantaneous feedback of channel conditions; e) performance gains of relay links after [11].](/figures/figure-5-a-direct-wireless-link-with-average-channel-26wcjp67.webp)
![Figure 3. a) Simulated ECG of the frequency domain bandwidth expansion as a function of the required SNR at the receiver (after [5]); b) simulated ECG of various MIMO schemes, relative to SFBC all at 3 b/s/Hz spectral efficiency (after [6]).](/figures/figure-3-a-simulated-ecg-of-the-frequency-domain-bandwidth-1pskpfxe.webp)
INTRODUCTION
Given the worldwide growth in the number of
mobile subscribers, the move to higher-data-rate
mobile broadband, and the increasing contribu-
tion of information technology to the overall
energy consumption of the world, there is a need
on environmental grounds to reduce the energy
requirements of radio access networks. A typical
mobile phone network in the United Kingdom
may consume approximately 40–50 MW, even
excluding the power consumed by users’ hand-
sets. In developing countries direct electricity
connections are not readily available, so Voda-
fone, for example, use in excess of 1 million gal-
lons of diesel per day to power their network.
Mobile communications thus contributes a sig-
nificant proportion of the total energy consumed
by the information technology industry.
From an operator’s perspective, reducing ener-
gy consumption will also translate to lower operat-
ing expenditure (OPEX) costs. Reducing carbon
emissions and OPEX for wireless cellular net-
works are two key reasons behind the develop-
ment of the Mobile VCE Green Radio program.
For example, the U.K. operators Orange and
Vodafone both aim to achieve significant reduc-
tions in CO
2
emissions in the next 10 years. The
Green Radio program sets the aspiration of
achieving a hundredfold reduction in power con-
sumption over current designs for wireless com-
munication networks. This challenge is rendered
nontrivial by the requirement to achieve this
reduction without significantly compromising the
quality of service (QoS) experienced by the net-
work’s users. In order to meaningfully measure
success, appropriate measures of energy consump-
tion must be applied. For example, a reduction in
radiated power is not of benefit if it is achieved at
the expense of a greater increase in power con-
sumed in signal processing or vice versa.
The Green Radio project is pursuing energy
reduction from two different perspectives. The
first is to examine alternatives to the existing cel-
lular network structures to reduce energy con-
sumption. The second approach, discussed in
detail in the present article, is to study novel
techniques that can be used in base stations or
handsets to reduce energy consumption in the
network. We present the background to the pro-
ject. We move on to discuss base station model-
ing, which is a critical issue for the project. We
then present three case studies that describe the
energy savings obtainable from different tech-
niques that can be employed on wireless links.
Finally, we present conclusions to the article.
REDUCING ENERGY CONSUMPTION IN
WIRELESS NETWORKS
The specific objective of the Green Radio pro-
gram is to investigate and create innovative
methods for the reduction of the total energy
needed to operate a radio access network and to
identify appropriate radio architectures that
enable such a power reduction. The typical
power consumption of different elements of a
IEEE Communications Magazine • May 2011
46
0163-6804/11/$25.00 © 2011 IEEE
ABSTRACT
Recent analysis by manufacturers and net-
work operators has shown that current wireless
networks are not very energy efficient, particu-
larly the base stations by which terminals access
services from the network. In response to this
observation the Mobile Virtual Centre of Excel-
lence (VCE) Green Radio project was estab-
lished in 2009 to establish how significant energy
savings may be obtained in future wireless sys-
tems. This article discusses the technical back-
ground to the project and discusses models of
current energy consumption in base station
devices. It also describes some of the most
promising research directions in reducing the
energy consumption of future base stations.
ENERGY EFFICIENCY IN COMMUNICATIONS
Congzheng Han, Tim Harrold, and Simon Armour, University of Bristol
Ioannis Krikidis, Stefan Videv, Peter M. Grant, Harald Haas, and John S. Thompson, University of Edinburgh
Ivan Ku and Cheng-Xiang Wang, Heriot-Watt University
Tuan Anh Le and M. Reza Nakhai, Kings College London
Jiayi Zhang and Lajos Hanzo, University of Southampton
Green Radio: Radio Techniques to
Enable Energy-Efficient
Wireless Networks
THOMPSON LAYOUT 5/19/11 9:08 AM Page 46
IEEE Communications Magazine • June 2011
47
current wireless network is shown in Fig. 1a.
These results clearly show that reducing the
power consumption of the base station or access
point has to be an important element of this
research program.
Studies have indicated that the mobile hand-
set power drain per subscriber is much lower
than the base station component, Fig. 1b [1];
hence, the Green Radio project will mainly focus
on base station design issues. Figure 1b also
shows that the manufacturing or embodied ener-
gy is a much larger component in the mobile
handset than in the base station. This is because
the lifetime of a base station is typically 10–15
years, compared to a typical handset being used
for 2 years. In addition, the energy costs of a
base station are shared between many mobile
subscribers, leading to a large imbalance in the
contribution of embodied energy. From the
point of view of handsets, significant efforts need
to be put into reducing manufacturing energy
costs and increasing handset lifetime, through
recycling programs, for example. The Third
Generation Partnership Project (3GPP) Long
Term Evolution (LTE) system has been chosen
as the baseline technology for the research pro-
gram; its specifications have recently been com-
pleted with a view to rolling out networks in the
next two to three years [2].
The next section of this article discusses the
architecture of existing base stations and identi-
fies key parts of the system hardware where sig-
nificant energy savings can be obtained.
BASE STATION
POWER EFFICIENCY STUDIES
The overall efficiency of the base station, in
terms of the power drawn from its supply in
relation to its radio frequency (RF) power out-
put, is governed by the power consumption of its
various constituent parts, including the core
radio devices.
Radio transceivers: The equipment for gener-
ating transmit signals to and decoding signals
from mobile terminals.
Power amplifiers: These devices amplify the
transmit signals from the transceiver to a high
enough power level for transmission, typically
around 5–10 W.
Transmit antennas: The antennas are respon-
sible for physically radiating the signals, and are
typically highly directional to deliver the signal
to users without radiating the signal into the
ground or sky.
Base stations also contain other ancillary
equipment, providing facilities such as connec-
tion to the service provider’s network and cli-
mate control. A major opportunity to achieve
the power reduction targets of the program lies
in developing techniques to improve the efficien-
cy of base station hardware.
Analysis within the program has developed
models for various base station configurations
(macrocell, microcell, picocell, and femtocell) in
order to establish how improvements in the
hardware components will impact the overall
base station efficiency. The starting point for this
analysis has been the transmit chain. Near-mar-
ket power consumption figures have been used
in order to establish a benchmark efficiency
against which improvements made as part of the
project can be assessed. Target power consump-
tion figures allow future overall base station effi-
ciencies to be predicted.
REFERENCE BASE STATION ARCHITECTURE
The target system for the base station efficiency
analysis is the LTE system with support for four
transmit antennas. This system can exploit the
space domain to achieve high data throughputs
through multiple input multiple output (MIMO)
techniques [2]. The reference architecture under
investigation is shown in Fig. 2, this represents a
macrocellular base station with three sectors,
with an effective isotropic radiated power
(EIRP) of 27 dBW per sector. The four transmit
chains needed for the four antennas therefore
require 12 power amplifiers (PAs) and antennas
Figure 1. a) Power consumption of a typical wireless cellular network (source: Vodafone); b) CO
2
emissions per subscriber per year as
derived for the base station and mobile handset, after [1]. Embodied emissions arise from the manufacturing process rather than opera-
tion.
Power usage (%)
Cellular network power consumption
10%
Base station
Mobile switching
Core transmission
Data center
Retail
0% 20% 30% 40% 50% 60%
Operational
energy
(b)
(a)
Embodied
energy
9 kg
CO
2
4.3 kg
CO
2
8.1 kg
CO
2
2.6 kg
CO
2
Base Mobile
THOMPSON LAYOUT 5/19/11 9:08 AM Page 47
IEEE Communications Magazine • June 2011
48
per base station. For clarity, only one of the 12
transmit chains is shown in Fig. 2.
Estimated base station power consumption
figures for the target system, reflecting the state
of the art for the years 2010–2011, are given in
Table 1. These estimates have been produced
for reference purposes using efficiency figures
from [3]; however, to reflect recent innovations
[4], a power amplifier efficiency of 40 percent
has been used. Two efficiency figures are calcu-
lated in Table 1; the top of cabinet (TOC) effi-
ciency gives the ratio of the combined power
output of the PAs to the power supply unit
(PSU) power (which is used in studies such as
[3]), and the radiated efficiency, which refers to
the ratio of the efficiency to the total power
radiated by the antenna. This second figure
therefore includes antenna efficiency and feeder
losses.
TARGET CONSUMPTION
The vision for the project is to specify an LTE
compliant base station that is able to operate at
much lower overall consumption, possibly suffi-
ciently low to enable operation from renewable
sources locally generated (e.g., solar or wind).
Challenging power consumption targets have
been set by the Green Radio program in order
to achieve this aim; these target figures are given
in the right column of Table 1.
The project target figures show an improve-
ment in efficiency by reducing the power
required to operate the base station by at least
50 percent. One solution reduces inefficiencies
by locating the PA next to the antennas (typical-
ly both at the top of the mast) in order to mini-
mize the power lost in feeder cables. This
architecture also further reduces the need for
cooling, which could arise were the PAs to be
installed in cabinets in an equipment room.
Additional efficiency gains are expected to come
about by deactivating portions of hardware when
unused.
Analysis shows that the greatest potential
for increasing the overall base station efficiency
comes from improving the efficiency of the PA
and antenna, as well as optimizing the power
transfer between them. Work underway in the
program is seeking to achieve efficiency figures
of 85 and 90 percent for these components,
respectively. In the case of the PA, one possible
approach uses the Class J amplifier [5], which
relies on fundamental and second harmonic
tuning to achieve high efficiencies, while main-
taining the linearity required for LTE opera-
tion. In the case of the antenna, the 90 percent
efficiency target is to be achieved by exploiting
highly efficient dual-polarized patch antenna
elements.
CASE STUDIES FOR IMPROVING
ENERGY EFFICIENCY IN
WIRELESS BASE STATIONS
Earlier the power consumption of base stations
was discussed and strategies to minimize power
use in future base stations was described. In this
section, we will move on to consider approaches
which are designed around the signals that are
transmitted by the base stations. In this case, the
time dimension of these waveforms becomes
important. In such a case, measures of energy
(power × time) rather than just power become
important metrics to measure system perfor-
mance effectively. This section will therefore
begin by discussing suitable energy metrics and
then move on to discuss three case studies, based
around resource allocation, interference cancel-
lation, and the use of multihop relaying strate-
gies.
OVERVIEW OF ENERGY METRICS
The results in Fig. 1a of this article show the fact
that base stations account for a significant pro-
portion of the total power consumption of a
wireless network. If new techniques are pro-
posed to reduce the energy required in the net-
work, it is important to provide meaningful
metrics that identify what gains are achieved.
The metrics to be used in the Green Radio pro-
ject have been discussed extensively, and there
are two particularly important metrics that are
intended to be used during the project.
The first is an absolute measure of energy
and is closely related to the industry concept of
the energy consumption rating (ECR). This is
typically defined as a ratio of peak power divid-
ed by the maximum data throughput for a base
station transmitter. However, to be of practical
use, the ECR should measure the consumed
energy per information bit that is successfully
transported over the network and is measured in
units of joules per bit. This metric allows the
absolute performance of different wireless net-
works to be calibrated. As a simple example, a
typical LTE base station sector might operate
over a bandwidth of 10 MHz with an average
spectral efficiency of 1.5 b/s/Hz, thus achieving
an average data rate of 15 Mb/s. If a base station
antenna transmits 8 W of RF power (Table 1),
the RF ECR value for this system would be 0.53
μJ/b. However, if the total power budget of the
Figure 2. Reference base station architecture for a system with three sectors and
four transmit antennas per sector for MIMO capability. For clarity only one
transmit chain is shown.
Base-
band
Base-
band
Base-
band
Radio
Power supply unit
Feeder
Power
ampli-
fier
Switch/
duplexer
Cooling
system
Antenna
Interfacing
THOMPSON LAYOUT 5/19/11 9:08 AM Page 48
IEEE Communications Magazine • June 2011
49
base station (e.g., 450 W) is shared among 3 sec-
tors (i.e., 150 W/sector) the ECR value for one
sector would increase to 10 μJ/b.
The second metric is a relative measure
rather than an absolute one and is more useful
for comparing two different systems. Frequently,
one may wish to compare the energy perfor-
mance of a base station using a newly proposed
technique (system under test) and compare to a
baseline system where the approach is not
deployed. The energy consumption gain (ECG)
is simply the ratio (E
b
/E
t
), where E
b
is the ener-
gy consumed by the baseline system and E
t
is the
energy for the system under test. The larger the
value of the ECG, the more efficient the system
under test becomes. However, as with the ECR
metric, care needs to be taken to ensure that the
energy calculations are performed in a fair man-
ner. For example, if two base station designs are
being compared, it should be ensured that both
are serving the same number of users under the
same traffic load conditions, in order to provide
a fair comparison.
CASE STUDY 1:
RESOURCE ALLOCATION STRATEGIES
RF amplifiers were identified as a key contribu-
tor to the overall energy consumption of a typi-
cal base station. In this article we use the term
resource allocation to describe how the base sta-
tion transmitter make the decision of how and
when to transmit data to different users on the
downlink (base-mobile link) within the cell it is
serving. Resource allocation techniques that
make the most efficient use of the RF amplifier
have the potential to improve energy efficiency
significantly. Such energy reductions could lead
to further energy savings through switching off
transceiver equipment and base station cooling.
In addition, analysis of data traffic in wireless
networks show that the traffic load is typically
very uneven across the cells. In the analysis of
200 cells in [2, Ch. 9], it is shown that even in
peak hours, 90 percent of the data traffic is car-
ried by only 40 percent of the cells in the net-
work. Therefore, techniques that minimize
energy consumption across varying traffic load
conditions are an important research direction;
here we describe two complementary techniques
aimed at low and high traffic load conditions,
respectively.
Under low traffic load conditions, the base
station is likely to have more bandwidth avail-
able to transmit data to users than is actually
required at that time. One frequency domain
approach being studied in the project exploits
spare bandwidth resources to reduce energy con-
sumption. Due to the fact that channel capacity
scales linearly with the available bandwidth but
logarithmically with the radio transmission
power, it is possible to trade spectral for energy
efficiency, and achieve energy savings while
retaining quality of service [6]. Rather than use a
complex but spectrally efficient modulation
scheme (e.g., 16-quadrature amplitude modula-
tion [QAM]) with a narrow bandwidth, it is pos-
sible to use a simpler modulation scheme (e.g.,
quaternary phase shift keying [QPSK]) with a
wider bandwidth.
Figure 3a shows predicted ECG gain results
for this approach, as a function for the signal-to-
interference-plus-noise ratio (SINR) required at
the mobile receiver for a given data rate. Gener-
ally speaking, as the spectral efficiency of the
data rate increases, so does the required SINR.
The value of α specifies the permitted band-
width expansion factor, and curves are shown for
values of α in the range 2–6. For example, a
bandwidth expansion of α = 2 would permit 16-
Table 1. Estimated power consumption for base stations in 2010–2011 and target future power consump-
tion values for base stations.
Description Power In (W) Power Out (W) Efficiency Target Value
Radiated power (per sector) 8 501 (27dBW)
18dBi antenna
gain
18dBi antenna
gain
Antenna and Switch 12 8 65% efficient 85% efficient
Feeder 24 12 50% efficient 80% efficient
PA (total per sector) 60 24 40% efficient 85% efficient
PA (all sectors) 180 72
Transceiver (all sectors) 180 70% reduction
Free Air Cooling 40
Subtotal 400
PSU Input 450 400 88% efficient 88% efficient
TOC Efficiency 16% > 25%
Radiated Efficiency 5.3% > 20%
Resource allocation
techniques that
make the most effi-
cient use of the RF
amplifier have the
potential to improve
energy efficiency sig-
nificantly. Such ener-
gy reductions could
lead to further ener-
gy savings through
switching off
transceiver equip-
ment and base sta-
tion cooling.
THOMPSON LAYOUT 5/19/11 9:08 AM Page 49
IEEE Communications Magazine • June 2011
50
QAM modulation (4 b/s/Hz maximum data rate)
to be replaced by QPSK (2 b/s/Hz maximum
data rate), which would require a lower SINR
for reliable operation. The results show that as
the SINR increases, so does the potential
improvement in ECG from using the bandwidth
expansion technique. Increasing the value of α
beyond four is shown to provide diminishing
returns in terms of ECG, except at very high val-
ues of SINR where very spectrally efficient mod-
ulation schemes would be used.
When the traffic load is high, the base station
may be transmitting data to many users simulta-
neously, possibly using MIMO techniques. In
this case, it is usually possible to exploit multi-
user diversity to increase the overall multi-user
capacity achieved via an opportunistic resource
scheduling and allocation strategy. This is where
the scheduler assigns resources according to the
users’ instantaneous channel conditions in the
time, frequency, or/and space domains. The per-
formance gains can be translated to further
energy reduction at the transmitter. A link adap-
tation approach is also taken into consideration
to ensure the most energy saving transmission
mode is employed within the allocated resource
for a required QoS level. As an example from
[7], Fig. 3b shows the ECG performance of dif-
ferent MIMO precoding schemes compared to
using the single-user MIMO diversity scheme
space frequency block coding (SFBC) as the
baseline case. The multi-user MIMO schemes
exploiting a higher degree of diversity achieve
lower cost in terms of required transmitter ener-
gy for each information bit. When the number of
mobile users is large enough, performance evalu-
ation results show that a fivefold energy gain can
be achieved by multi-user MIMO through
employing appropriate link adaptation and
resource scheduling approaches compared to an
SFBC system.
Future work in this area will study the best
combination of scheduling techniques from an
energy efficiency perspective across the range of
traffic loads experienced in future LTE net-
works.
CASE STUDY 2:
INTERFERENCE MANAGEMENT AND MITIGATION
Interference cancellation schemes are indispens-
able to combat interference in any practical
communication systems where multiple base sta-
tions share the same spectrum. The impact of
interference is more severe as users move closer
to the boundary region between two cells, lead-
ing to significant SINR and hence data rate
reduction. Most existing interference cancella-
tion schemes have been designed to increase the
spectral efficiency and data rate, while overlook-
ing energy efficiency. However, research efforts
in the Green Radio program are focused on
developing energy-efficient interference cancel-
lation schemes. If the level of interference can
be reduced at mobile terminals, it will permit
base stations to reduce the wireless transmission
energy without compromising the SINR of the
wireless link. There are two complementary
strategies being considered, as shown in Fig. 4a:
distributed antenna systems and receiver inter-
ference cancellation.
One way to reduce interference in cellular
systems is to coordinate the multiple antennas of
the adjacent base stations to form a distributed
antenna system (DAS) [8]. For the resulting
coordinating DAS, each and every cell edge user
is collaboratively served by all of its surrounding
base stations rather than only by the single best
base station. This permits the interference to
users on the cell edge to be effectively controlled
and mitigated by coordinated transmit beam-
forming at all of the participating base stations.
The following three schemes can be used by
coordinating downlink beamforming:
Figure 3. a) Simulated ECG of the frequency domain bandwidth expansion as a function of the required SNR at the receiver (after [5]);
b) simulated ECG of various MIMO schemes, relative to SFBC all at 3 b/s/Hz spectral efficiency (after [6]).
Target SINR [dB]
(a) (b)
97
10
0
Energy consumption gain (ECG)
5
15
20
25
30
35
40
45
50
11 13 15 17 19 21 23 25 27
Number of users
LTE downlink simulation results
5
0
Energy consumption gain (ECG)
5
6
4
3
2
1
10 15 20 25
α=2
α=3
α=4
α=5
α=6
SU-MIMO
Full feedback MU-MIMO
Partial feedback MU-MIMO
SFBC
THOMPSON LAYOUT 5/19/11 9:08 AM Page 50