challenges and enabling technologies for energy aware mobile radio networks

IEEE Communications Magazine • November 201066 0163-6804/10/$25.00 © 2010 IEEE

ENERGY EFFICIENCY IN COMMUNICATIONS

Luis M. Correia, IST/IT Technical University of Lisbon

Dietrich Zeller, Oliver Blume, and Dieter Ferling, Alcatel-Lucent Bell Labs

Ylva Jading and István Gódor, Ericsson Research

Gunther Auer, DOCOMO Euro-Labs

Liesbet Van der Perre, IMEC

Challenges and Enabling Technologiesfor Energy Aware Mobile Radio Networks

INTRODUCTIONToday, the worldwide mobile communicationinfrastructure offers ubiquitous mobile connec-tivity and unprecedented service levels. Morethan 4 billion subscribers around the worlddepend on their mobile phones for their privateand professional lives; however, this has come atthe cost of increasing energy consumption. Thusfar, the mobile communications industry hasfocused on highly energy efficient terminals, soto make mobile phones reliable and attractivefor consumers, a factor that has strongly con-tributed to the global success of mobile radio.

The emerging trend towards energy-efficientnetwork operation shifts the focus towards theenergy consumption of the wireless access net-work infrastructure, which has triggered activi-ties in standardization and regulatory bodies,such as 3GPP, ITU, ETSI, and ATIS.

There are several reasons for the growingawareness of energy-efficient wireless networksin the telecommunication community. Increasingenergy prices imply that electricity bills havebecome a significant cost factor for mobile oper-ators. In addition, mobile telephony and mobilebroadband are entering new emerging markets,with an increasing share of base stations that arenot connected to the electricity grid. Such off-grid sites are typically diesel powered, where fuelis costly and distribution often unreliable for dis-tant sites that are difficult to access. For opera-tors with many off-grid sites, energy provisionmay contribute up to 50 percent of their totaloperational cost. There is also an increasingawareness in the society about manmade climateeffects, and the need to slow down global warm-ing. Moreover, political initiatives are beginningto put requirements on manufacturer and opera-tors to lower CO2 emissions of communicationnetworks. The European Commission researchproject EARTH [1], consolidates the energy effi-ciency activities from major vendors, operatorsand academia in Europe. The project concen-trates on energy efficiency in radio access net-works, covering a wide span of activities fromnear future component research and develop-ment to long term research for theoretical limitsof different concepts; the goal is to find solutionsand concepts that can reduce the energy con-sumption of mobile broadband systems by 50percent.

Unlike mobile phones, where the manufac-turing of the devices is the main contributor toCO2 emissions, on the network side the big sav-ing potential lies in the operation [2]. The main

ABSTRACT

Mobile communications are increasingly con-tributing to global energy consumption. In thisarticle, a holistic approach for energy efficientmobile radio networks is presented. The matterof having appropriate metrics and evaluationmethods that allow assessing the energy efficien-cy of the entire system is discussed. The mutualsupplementary saving concepts comprise compo-nent, link and network levels. At the componentlevel the power amplifier complemented by atransceiver and a digital platform supportingadvanced power management are key to efficientradio implementations. Discontinuous transmis-sion by base stations, where hardware compo-nents are switched off, facilitate energy efficientoperation at the link level. At the network level,the potential for reducing energy consumption isin the layout of networks and their management,that take into account slowly changing daily loadpatterns, as well as highly dynamic traffic fluctu-ations. Moreover, research has to analyze newdisruptive architectural approaches, includingmulti-hop transmission, ad-hoc meshed net-works, terminal-to-terminal communications,and cooperative multipoint architectures.

CORREIA LAYOUT 10/20/10 3:42 PM Page 66

IEEE Communications Magazine • November 2010 67

reason for this difference is the limited lifetimeof mobile phones, the effort for distribution ofthe devices, and the small recycling fraction ofmobile phones compared to radio networkequipment.

This article elaborates on all levels of thecommunication system, covering network levelaspects, including deployment, architecture andnetwork management; link level perspective; andthe component level, including hardware imple-mentation targeted for improved energy efficien-cy in radio access network operation.Furthermore, a framework for the evaluation ofthese concepts in a holistic way is presented,revealing the mutual dependencies, and trade-offs are discussed together with suitable metrics.

FRAMEWORK FOR ENERGYEFFICIENCY EVALUATION

The deployment of cellular networks is typicallyoptimized for ubiquitous radio access of mobileusers. This implies that a significant portion ofbase stations are primarily providing coverage,therefore not operating at full load, even at peaktraffic hours. During off-peak hours, virtually allbase stations operate at low load, or might noteven serve any user at all. Unfortunately, theenergy efficiency of base stations is particularlypoor in this condition, which can be attributedto:• Component level: The efficiency of the

power amplifier (PA) substantially degradesat lower output power.

• Link level: System information, synchroniza-tion, and reference signals (pilots) need tobe transmitted continuously, so that basestations are required to be always on.

• Network level: The cellular network deploy-ment paradigm with large macro-cellsrequires additional (micro- or pico-) cells tofulfill peak capacity demand, but this ratherstatic topology does not adapt to low loadsituations.This underlines the fact that energy efficient

network operation requires a holistic approachincluding all system aspects. The standardizationof an energy efficiency evaluation framework hasalready started in a number of bodies; forinstance, ETSI and 3GPP have initiated work onthe specification of energy efficiency metrics formobile cellular networks. However, today’s pro-posals exhibit three major shortcomings:• System performance is optimized for capaci-

ty and evaluated at full load, which impliesthat the system is poorly configured at lowloads;

• The output power at the antennas is takenas performance measure; clearly, in orderto evaluate the energy consumption, oneshould considered the total input powerinstead;

• Performance metrics that measure the ener-gy efficiency.In what follows, the necessary modifications

to existing performance evaluation frameworksare discussed, such that the actual power con-sumption of the entire network can be quanti-fied.

A POWER CONSUMPTION MODEL

To date, studies on energy efficiency for wirelessnetworks typically regard the output power asperformance measure. Often, such energy per-formance metrics only capture a small fractionof the overall power budget of wireless networks,and therefore may lead to incomplete and poten-tially misleading conclusions. In order to evalu-ate the energy efficiency of a wireless network,the power to operate the entire access networkis to be taken into account, comprising RF signalgeneration, digital and analogue signal process-ing, backhaul data link, AC/DC power conver-sion, and cooling.

The prohibitive complexity of the problemmakes a certain level of idealization unavoid-able. Already a fairly simple, yet powerful, linearpower model allows one to assess the energyefficiency of the entire network; Fig. 1 showssuch a model, which maps the consumed inputpower Pin to achieve a certain output power atthe antenna Pout, with power consumption valuesbased on [3, 4]. This model determines how thepower efficiency, denoted by Pout/Pin, degradeswith decreasing output power Pout; in particular,at zero output power a base station still con-sumes a non-negligible fraction of the maximuminput power. The DC power consumption of atypical 3-sector site at zero load is still 50 percentof the peak power [3]. The conventional modelwithout power supply and active cooling/air con-ditioning can be 400 W lower than the totalpower consumption of a site.

Moreover, the introduction of base stationsleep modes, where the required input power atzero load can be significantly reduced, are anenabler for discontinuous transmission (DTX)modes, and therefore need to be considered inthe linear power model as well.

The proposed model allows one to under-stand the system power consumption at networklevel, as a function of base station load, cell size,and system parameters. This allows one to opti-mize the radio resource management (RRM) forenergy efficient operation, given a certain PA.

Figure 1. Linear model approximating the relationship between input and out-put powers of a 3-sector macro base station.

Sleep modes

1800W

Pin

Pout

Pmax

(120W)

1400W

700W

Total Conventional Advanced


IEEE Communications Magazine • November 201068

On the other hand, the potential energy savingsof advanced PAs and other system componentsthat, e.g., improve the power efficiency at lowload, may also be assessed.

ENERGY EFFICIENCY METRICSOne of the most commonly used metric formeasuring the energy efficiency of a communi-cation link is the consumed energy over thenumber of information bits in [Joule/bit]. Forcellular networks, this metric relates the totalenergy consumed by the entire network to theaggregate network capacity. As the measure[Joule/bit] relates the cost (in terms of energy)to the generated utility (information bits), thismetric is appropriate to assess the energy effi-ciency at full loads. On the other hand, whenthe network is operated well below its capacity,the main objective is to minimize the powerconsumption to cover a certain area, in whichcase [W/m2] is deemed the most relevant energyefficiency metric.

There is a multitude of further figures ofmerit for optimising the different aspects andcomponents of the wireless system, e.g., qualityof service parameters, such as error rate, delay(jitter), as well as a fair distribution of the indi-vidual user throughputs throughout the network.Energy efficiency metrics complement the afore-mentioned figures of merit, rather than replacingthem; otherwise, optimisation may just improvetowards one direction, at the expense of otherimportant aspects.

COMPONENT LEVEL POTENTIALIn order to improve the total energy efficiencyof cellular networks, special attention needs tobe paid to the different base station compo-nents, as they are the main contributors to thetotal energy consumption. To this end, oneneeds to analyze the power consumption of dif-ferent radio components for relevant traffic sce-narios. As an example, Fig. 2 shows a typicalpower consumption of macro-cellular radioaccess equipment.

Figure 2 reveals that a significant portion ofthe energy is consumed in the power amplifica-tion process, since sufficient power is needed toreach distant terminals with high path losses.Unfortunately, power amplification has a ratherpoor efficiency, reaching up to 50 percent formaximum load, but degrading to much lowervalues in medium and low load situations. This isa major reason why the power consumption incellular networks is, to a large extent, indepen-dent of the traffic load (Fig. 1).

For this reason, in addition to the applicationof advanced power amplifiers, new power man-agement concepts are to be devised that adaptto varying traffic load. Introducing scalabilityinto hardware components, and supporting themby dynamic power management, enables theadaptation of energy consumption to actual per-formance requirements. Further power savingsare facilitated by the deactivation of componentsin time periods of no operation.

IMPROVING THE POWER AMPLIFIER EFFICIENCYModulation schemes used in WCDMA/HSPAand LTE are characterized by strongly varyingsignal envelopes, with peak-to-average powerratios (PAPR, also known as crest factor) exceed-ing 10 dB. On the other hand, to attain the quali-ty of transmitted radio signals requested by thestandards, a high linearity of the amplifiers isrequired, hence, PAs must operate well belowsaturation, leading to poor power efficiency.

To increase the power efficiency, signal con-ditioning algorithms, like crest factor reduction(CFR) for decreasing the PAPR and digital pre-distortion (DPD) for increasing the PA linearity,must be applied to enable the PA operationcloser to saturation. Furthermore, PAs based onspecial architectures must be used to achievehigh efficiency values. For example, DohertyPAs, which contain one main amplifier alwaysactive and an auxiliary one active only when sig-nal peaks occurs, show maximum efficiency atpower levels of 6 to 10 dB below the saturationpoint, matching with the PAPR values of inter-est. By using CFR and DPD with Doherty PAs,up to 50 percent efficiency can be achieved forfull load; at low loads, the power efficiencydecreases significantly to only around 5 percent,at 10 percent load.

BASE STATION POWER MANAGEMENTIn order to achieve energy efficiency, the con-sumed power in the base station front end shouldscale, as far as possible, to the amount of servedtraffic.

To this end, load adaptive CFR combinedwith adaptive power supply of the PA shows apromising solution in the analogue chain. Thishas to include energy efficiency optimization ofthe power supplies for variable input power.

Energy efficient power management requiresreconfigurable circuits as key elements. A lowpower front-end that supports different levels oftransmit power with adjustable performance(quantified as signal to distortion ratio, mainlydepending on the back-off) has been proposedin [6]. In this solution, both driver and poweramplifier are digitally controlled and flexible interms of output power, linearity, and DC power

Figure 2. Typical power consumption distribution in radio access technology(absolute values relate to the base station of Fig. 1) [5].

Power amplifierincl. feeder

50-80%(~1200W)

Air conditioning 50-80%

(~300W)

Signal processing (analogue and digital)

5-15% (~200W) Power supply

5-10% (~100W)



consumption. Measurements indicate that instan-taneous transmitter energy can be scaled by 30percent. Considering also adaptive scheduling,the average energy efficiency is even improvedby up to 40 percent, compared to non-scalablesystems.

It is highly desirable to extend this kind oftechniques to high power transceivers commonlyused in base stations serving macro-cells. Tech-niques like dynamic voltage and frequency scal-ing can adapt the voltage and clock frequency ofthe platform depending on the work load, whichallows for scaling the power consumed in thedigital chain with traffic.

DEACTIVATION OF COMPONENTSComponent deactivation is required to exploitenergy saving potential of downlink DTX andbase station sleep mode techniques managedfrom link or network levels.

Base station components must allow switch-ing off inactive digital and analogue circuits. Inthis case, inactive digital circuits should bepower gated in order to reduce the leakagepower on the platform. The analogue chainshould be totally or partly switched off duringthe inactive time periods, providing significantpower savings of the transceiver, especially bythe power amplifier.

The benefit of DTX techniques is highlydependent on how fast deactivation and reacti-vation can be supported by the functional unit.The shortest DTX, within a 1 ms subframe,requires timescales in the order of tenths ofmicroseconds, to suppress the power consump-tion during time periods of no transmission. Thisfunctional unit consists of PA, power supply, andsignal conditioning.

LINK LEVEL POTENTIALIn the last decade, quite a number of improve-ments of radio interfaces [7] have been achieved,boosting user data rates to the order of 100Mb/s. State-of-the-art radio link designs focus onachieving high peak data rates, optimal spectralefficiency, appropriate coverage, and low laten-cies. In addition, it now becomes necessary torethink the way radio interfaces are designedfrom an energy efficiency perspective.

Transceivers employing multiple transmit andreceive antennas boost spectral efficiency, interms of b/s/Hz, which potentially allows totransmit more bits for a given transmit power. Inpractice, however, these improvements hardlyreduce the power consumption of the network.The reasons for this are manifold, three exam-ples being discussed below.

REFERENCE SYMBOL ANDCONTROL SIGNAL OVERHEAD

Systems equipped with multiple transmit andreceive antennas are an integral part of beyond3G mobile communication systems, e.g., LTE.Combined with adaptive transmission techniquesthat exploit channel knowledge at the transmit-ter, improvements in spectral efficiency areobtained. On the other hand, accurate knowl-edge of the actual radio channel conditions is

required, both at the transmitter and the receiv-er, which is achieved by transmitting auxiliaryinformation in the form of reference symbols(pilots) known to the receiver; in particular, theamount of reference symbols grows strongly withthe number of transmit antennas. At full loadthe energy consumed to transmit this overheadis overcompensated by the enhanced spectralefficiency; in fact, in theory, up to 50 percent ofreference signal overhead is acceptable to stillsee capacity gains by adding transmit and receiveantennas at fixed total power [8]. However, theamount of the auxiliary information remains thesame in low load situations, where the radiointerface carries only a fraction of its maximumcapacity. As an example, the current LTE stan-dard requires transmission of cell specific refer-ence symbols over the whole bandwidth at alltime; for this, the overhead ranges between 5and 15 percent for 1 to 4 transmit antennas,respectively. Taking system information, controland synchronization channels into account, theoverhead energy consumption may exceed 25percent of its maximum even when no user datais transmitted. This fixed transmission of controlsignals contributes very significantly to the over-all power consumption of a network. With thetrend toward transmitters with multiple transmitantennas, the overhead in low load situationsincreasingly impacts on the overall energy con-sumed by a network.

BASEBAND SIGNAL PROCESSINGPower consumed by signal processing also con-stitutes a significant part of the overall powerconsumption of a network. Highly spectrally effi-cient transmission techniques have a tendency torequire more complex computations with a cor-responding increase of processing power. There-fore, the gains of such advanced transmissiontechniques on the energy efficiency may be outweighted by negative effects in other parts of thesystem. This is particularly true for deploymentswith small cell sizes, where the cell transmissionpower amounts only a few percent of that of amacro-cell, thus, baseband signal processing maydominate the overall energy consumption. How-ever, such difference between power consump-tion components of pico/micro- and macro-cellsdoes not determine in advance which cell typeshould dominate a particular network scenario.

DISCONTINUOUS TRANSMISSION ANDBASE STATION SLEEP MODES

The demand for very long standby and call timesof mobile terminals has resulted in a very effi-cient use of energy. This has been possible byemploying schemes like DTX (as mentionedabove), which periodically create periods in thetransmission protocol where power consumingcomponents can be switched off. Unfortunately,DTX at the base stations is not supported in theWCDMA/HSPA specifications as it requirescontinues pilot transmission; the situation hasalready improved somehow in LTE, since cellspecific reference signals are no longer transmit-ted continuously, although frequent transmissionof synchronization signals and broadcast channelremain. However, there is still potential for fur-

Highly spectrally

efficient transmission

techniques have a

tendency to require

more complex com-

putations with a cor-

responding increase

of processing power.

Therefore, the gains

of such advanced

transmission tech-

niques on the energy

efficiency may be

out weighed by neg-

ative effects in other

parts of the system.



ther exploiting downlink DTX schemes as wellas various kinds of base station sleep modes infuture standards:• Micro sleep modes, where base stations sus-

pend transmission in the order of millisec-onds

• Deep sleep modes, where base station trans-mitters are shut down for extended periodsof timeWhile in micro sleep mode base stations are

required to wake up almost immediately, in thedeep sleep some transmit circuits are completelyswitched off, which implies that wake up timesare substantially longer.

In order to utilize the potential of base sta-tion sleep modes, protocols need to be devel-oped that allow suspending the transmission ofreference symbols at low loads, hence, new pro-tocols for handover and initial access proceduresare needed that enable base stations to be onsleep mode.

POTENTIAL GAINSIn the development of radio link protocols, thefocus has been on low power consumption ofmobile terminals. Ignoring the power consump-tion of base stations in the design of the radiotransmission techniques inevitably leads to ener-gy wastage.

Current research on energy efficiency aspectsof cellular radio access technologies focuses onisolated hardware aspects, hence, an embracinganalysis is lacking. In order to achieve the largestenergy savings, a joint optimization of spectraland power efficiency for the radio link operationis necessary. Future research must instead takethe total energy consumption as optimizationcriterion into account, in both the radio inter-face and network nodes, and in different opera-tion and deployment situations, which isexpected to yield significant energy savings. Theresulting new radio interface and transmissiontechniques will exhibit a much higher flexibilityfor the dynamic adaptation of the radio trans-mission to the dynamic traffic load situationsand operation environment.

NETWORK LEVEL POTENTIALSOn network level, the potential for reducingenergy consumption is in the layout of networksand in their management.

The energy consumption for maintaining net-work operation has to be evaluated as a function

of cell size and deployment scenario, for giventraffic distributions and environments rangingfrom dense urban to rural areas [9]. Site specificdeployment options, like sectorization, antennatilting, and distributed antennas need to be con-sidered. Furthermore, heterogeneous networkswith a hierarchical mix of base stations with dif-ferent cell sizes have recently attracted increas-ing interest for 4G mobile communicationsystems [10], as they provide deployment strate-gies for tailored provisioning of capacity.

Despite the fact that the traffic load of mobilenetworks follows a daily profile of large ranges,networks are configured rather statically today.For an energy efficient operation, network man-agement needs to take slowly changing dailyload patterns into account, as well as dynamictraffic fluctuations. Moreover, research has toanalyze new architectures with more disruptiveapproaches, including multi-hop transmission,ad-hoc meshed networks, terminal-to-terminalcommunications, and cooperative multipointarchitectures.

NETWORK DEPLOYMENTIn heterogeneous networks, illustrated in Fig. 3,low power nodes complement the conventionalmacro-cellular network layout, micro-/pico-cellscover high traffic demand (hot spot) areas, whileindoor base stations constitute femto-cells thatprovide broadband coverage to indoor users.Furthermore, relay nodes are a cost efficientmeans to extend outdoor coverage, since expen-sive backhaul links are avoided. Such heteroge-neous networks potentially promise energysavings, since they shorten the propagation dis-tance between nodes, and thereby reduce therequired transmission power. On the other hand,a deployment with only small cells may be uneco-nomical, due to the prohibitive number of lowpower, and therefore short range, base stations.Moreover, this increases the number of base sta-tions operating at low loads, which may degradethe overall energy efficiency. Hence, for eachscenario (e.g., dense urban vs. rural) heteroge-neous deployments with an optimal balance ofmacro-, micro-, pico-, and femto-cells must befound for a most energy efficient network layout.

Cooperative transmission from several basestations to one mobile device is a further enablerfor enhanced energy efficient network operation;such inter-base station cooperation avoids inter-ference, or can even turn interference into a use-ful signal. However, the improvement in energyefficiency at the air interface may be cancelledout by more complex signal processing and/orincreased backhaul traffic. Fundamental trade-offs among these factors need to be studied, pro-viding the basis for designing energy efficientsite cooperation protocols.

NETWORK MANAGEMENTFigure 4 shows the daily traffic profile of a majoroperator, where the traffic in peak hours can be10 times higher than the one in off-peak periods.Realizing the potential in these slow traffic vari-ations, vendors and operators have recentlystarted to bundle traffic of certain transceivers,and switch others off in order to save energy.Beyond these basic techniques, self-organizing

Figure 3. Heterogeneous network deployment, consisting of micro-/pico-cellsserving hot spot areas, femto-cells for indoor users, and relays for coverageextension.

Relay

Micro-/pico-cell Backhaul Femto-cell

Macro-cell



networks can be exploited to adapt the networktopology to slow traffic variations. Furthermore,highly dynamic load variations should beaddressed by energy efficient radio resourcemanagement strategies.

Energy efficient adaptation to the daily andspatial variation of the traffic demand requiresspecific network management functionalities.The state-of-the-art solution is to simply reducethe number of active network elements when thetraffic demand decreases [12]; however, once dif-ferent energy efficient deployment scenarios areavailable for different traffic situations, newhorizons open for network management. That is,network management can provide a seamlesstransition from one network configuration andtopology setup to another, according to actualtraffic demands, considering both slowly chang-ing daily load patterns, as well as highly dynamictraffic fluctuations.

The transition process from one networksetup to another needs the reconfiguration ofparticular resources, or the coordinated recon-figuration of groups of resources, or even of thewhole network. When traffic demand decreases,the transition process may include, e.g., switch-ing off particular micro-/pico-cells, transferring asectorized base station to an omnidirectionalone, or increasing cell coverage by adapting theantenna tilt. When traffic demand increases, thetransition process may include, e.g., switchingbetween various MIMO functionalities, or direct-ing adaptive antenna arrays towards traffichotspots. Beyond the analysis of the aboveoptions, other issues are at stake, like developingnew network management concepts for dynamicreconfiguration of mobile networks, includingself-optimization, self-configuration and standby-operation.

When comparing different radio access tech-nologies, one can find that some are more ener-gy efficient for certain types of services thanothers. In order to take advantage of the coexis-tence of different radio access technologies, newnetwork management functions are needed tocontrol a common resource pool. That is, radioresource management has to involve cooperativescheduling, interference coordination, and jointpower and radio resource control. There is aneed to investigate how to define these manage-ment functionalities to leverage the capabilitiesof the different technologies, and to employmulti-radio access networks in a coordinatedfashion for energy savings.

CONCLUSIONSThis article addresses energy efficiency of mobilecellular networks, which is a concern of industry,operators, as well as of regulatory bodies and thesociety. It stresses that for fundamental improve-ments of the energy efficiency of wireless broad-band access, joint optimizations yielding anoptimum at system level need to be aimed atrather than optimizing single aspects or compo-nents. Initially, it presents a framework for ener-gy efficiency evaluation, identifying the levelsthat need to be addressed, i.e., component, linkand network. Metrics need to be taken for suchassessment, capturing, e.g., the backhaul power

consumption, and system parameters, e.g., quali-ty of service.

It is important to perform an assessment ofthe power consumption distribution of the dif-ferent radio components and subsystems in typi-cal communication scenarios. A key fraction ofthis consumption is observed in the power ampli-fication stage. Digital platforms should activelysupport advanced power management, togetherwith dynamic voltage and frequency scalingadapted to the workload. Energy-scalable recon-figurable transmitters and their control strategyare also key building blocks in this approach.

Radio interfaces and transmission techniquesplay also an important role. Aspects like spectralefficiency and information overhead need to beconsidered; however, more efficient techniquesusually require a much more complex computa-tion, which leads to an increase of the associatedprocessing power, therefore, reinforcing theneed for the global overview of the problem.Discontinuous transmission by base stations,where hardware components, like power ampli-fiers, can be switched off, can be seen as one ofthe directions to follow.

At the network level, the potential for reduc-ing energy consumption lies in the layout of net-works and their management. Networkmanagement needs to take slowly changing dailyload patterns into account, as well as highlydynamic traffic fluctuations. Moreover, researchhas to analyze new architectures with more dis-ruptive approaches, including multi-hop trans-mission, ad-hoc meshed networks,terminal-to-terminal communications and coop-erative multipoint architectures. The use of cel-lular heterogeneous structures, and cooperativetransmission from several base stations to onemobile device, are further enablers.

In conclusion, research on energy efficiencyaspects of mobile cellular radio network has aquite extensive list of topics to be addressed,requiring an embracing analysis. The EARTHproject is pursuing this vision, targeting to sub-due major energy savings by exploring theapproaches discussed within this article.

Figure 4. Measured weekly traffic variations of a mobile cellular network [11].

Volu

me

(GBy

te)

Monday

50

100

150

200

250

300

350

400

0 Tuesday Wednesday Thursday Friday Saturday Sunday

Application 1 Application 5






ACKNOWLEDGMENTS

The authors would like to acknowledge the sup-port of the European Commission by partiallyfunding this work, under project FP7-ICT-2009-4-247733-EARTH. We gratefully recognize thegreat contributions of many colleagues fromEARTH; in fruitful cooperation, many peoplefrom all partners contributed with valuableinsight, surveys and vision.

REFERENCES[1] EARTH (Energy Aware Radio and neTwork tecHnolo-

gies), “EU Funded Research Project FP7-ICT-2009-4-247733-EARTH,” Jan. 2010–June 2012;https://www.ict-earth.eu.

[2] J. Malmodin, “The Importance of Energy in LCA for theICT Sector,” Proc. 14th SETAC, Göteborg, Sweden, Dec.2007.

[3] D. Ferling et al., “Energy Efficiency Approaches forRadio Nodes,” Proc. Future Net. Mobile Summit ‘10,June 2010, Florence, Italy.

[4] G. Fischer, “Future Challenges in R&D with Respect toMobile Communication Base Stations,” Bell Labs Tech.Rep., Stuttgart, Germany, Jan. 2008.

[5] Alcatel-Lucent and Vodafone Chair on Mobile Commu-nication Systems, “Study on Energy Efficient RadioAccess Network (EERAN) Technologies,” UnpublishedProject Report, Technical University of Dresden, Dres-den, Germany, 2009.

[6] B. Debaillie et al., “Energy-Scalable OFDM TransmitterDesign and Control,” Proc. 43rd ACM/IEEE DAC ‘06,San Francisco, CA, July 2006.

[7] E. Dahlman et al., “3G Radio Access Evolution — HSPAand LTE for Mobile Broadband,” IEICE Trans. Commun.,vol. E92-B, no. 5, May 2009, pp.1432–40.

[8] B. Hassibi and B. M. Hochwald, “How much Training isNeeded in Multiple-Antenna Wireless Links?,” IEEETrans. Info. Theory, vol. 49, no. 4, Apr. 2003, pp.951–63.

[9] 3GPP TR 36.814 v2.0.1, “Further Advancements for E-UTRA. Physical Layer Aspects (Release 9),” Tech. Spec.Group Radio Access Net., Mar. 2010; http://www.3gpp.org.

[10] H. Claussen, L. Ho, and F. Pivit, “Effects of JointMacrocell and Residential Picocell Deployment on theNetwork Energy Efficiency,” Proc. 19th IEEE PIMRC ‘08,Cannes, France, Sept. 2008.

[11] Alcatel-Lucent, “9900 Wireless Network Guardian,”Tech. White Paper, 2008;http://www.alcatel-lucent.com.

[12] Ericsson, “Sustainable Energy use in Mobile Communi-cations,” White Paper, Aug. 2007;http://www.ericsson.com.

BIOGRAPHIESLUIS M. CORREIA [SM‘03] received his Ph.D. in Electrical andComputer Engineering from IST-TUL in 1991, where he iscurrently a professor in telecommunications, with his workfocused on wireless/mobile communications. He has beenactive in various projects within European frameworks. Hewas part of the COST Domain Committee on ICT and hasbeen involved in eMobility activities.

DIETRICH ZELLER received his university degree in Mathe-matics and Physics from the University of Tübingen in

Germany. In 1990 he joined Alcatel research where heworked on various fields of communications. His currentresearch interest is in energy efficient radio access sys-tems and he is leading a research team on energy effi-cient wireless access within Bel l Labs. He is projectcoordinator of the FP7 EARTH consortium funded by theEuropean Union. He is member of the Alcatel-LucentTechnical Academy.

OLIVER BLUME received a degree in Physics from the Uni-versity of Hamburg, Germany, and a Dr.-Ing. degree fromthe Technical University of Hamburg-Harburg in 2001. Heis a research engineer in the Wireless Access researchdomain of the Alcatel-Lucent Bell Labs in Stuttgart, Ger-many, where he is currently involved in research forGreen ICT. His research interests are in wireless communi-cations, radio research management, and IP mobility pro-tocols. He is member of the Alcatel-Lucent TechnicalAcademy.

YLVA JADING received her M.Sc. from Chalmers University ofTechnology, Gothenburg, Sweden, in 1992 and her Ph.D.in Nuclear Astrophysics from Johannes Gutenberg Univer-sität, Mainz, Germany, in 1996. She joined Ericsson in1999, being currently master researcher at EricssonResearch and active in concept and standardization workfor the LTE radio access technology both internally and in3GPP. In addition, she leads an internal research teamworking on energy efficiency and is the technical managerof the EARTH project.

GUNTHER AUER received the Dipl.-Ing. degree in ElectricalEngineering from the University of Ulm, Germany, in 1996,and the Ph.D. degree from The University of Edinburgh,UK, in 2000. Since 2001 he is with DOCOMO Euro-Labs,Munich, Germany, where he is team leader and researchmanager in wireless technologies research. His researchinterests include self-organized networks and multi-carrierbased communication systems, including medium access,channel estimation and synchronization techniques.

LIESBET VAN DER PERRE received the M.Sc. and the Ph.D.degrees in Electrical Engineering specializing in wirelesscommunication from the K.U. Leuven, Belgium, in 1992and 1997 respectively. She joined IMEC in 1997, takingresponsibilities as system architect, project leader, and pro-gram manager. Currently, she is the Director of the GreenRadio programs in IMEC. She’s an author and co-author ofover 200 scientific publications. Also, she’s a Professor atK.U. Leuven.

DIETER FERLING received the Dipl.-Ing. degree in ElectricalEngineering in 1990 at the University of Stuttgart, Ger-many. In 1990 he joined Alcatel-Lucent research, where heworked on mm wave transceivers, multichannel opticalmodules, and innovative interconnections and packages ofmm wave devices. His current research work is on base sta-tions for mobile communications at Bell Labs WirelessAccess Research Domain (Stuttgart), focusing on multibandand multi-standard capabilities of transceivers and on effi-cient power amplifiers.

ISTVA’N GO’ DOR received both his M.Sc. and Ph.D. degree inElectrical Engineering from Budapest University of Technol-ogy and Economics, Budapest, Hungary in 2000 and 2005,respectively. He is a research fellow at Ericsson Research,Traffic Analysis and Network Performance Laboratory ofEricsson Hungary. His research interests include networkdesign, combinatorial optimization, cross-layer optimiza-tion, and traffic modeling.

Research has to

analyze new

architectures with

more disruptive

approaches,

including multi-hop

transmission, ad-hoc

meshed networks,

terminal-to-terminal

communications and

cooperative

multipoint

architectures.

CORREIA LAYOUT 10/20/10 3:42 PM 2

challenges and enabling technologies for energy aware mobile radio networks

Documents