14 PERVASIVEcomputing Published by the IEEE CS and IEEE ComSoc ■ 1536-1268/05/$20.00 © 2005 IEEE

G U E S T E D I T O R S ’ I N T R O D U C T I O N

Pervasive computing aims to integratecomputation into our daily work prac-tice to enhance our activities withoutbeing noticed. In other words, comput-ing becomes truly invisible. Yet at the

heart of every pervasive computing system areelectronic components that consume energy. Man-aging the energy needs of mobile systems, or sys-tems for which reliable power isn’t guaranteed,can be a significant distraction for users. How canwe minimize user involvement in the energy man-agement process to make pervasive computingdevices more pervasive?

Power management techniques Most users can attest—from their experience

with mobile phones, portable media recorders, andportable media players—thatpower management can be dis-tracting. Of course, the magni-tude of distraction depends onthe specific application, the hard-ware, and the energy source’scharacteristics. Researchers aretrying to address this issue pre-dominantly by reducing powerconsumption and developing

improved energy sources. Unfortunately, bothapproaches have limitations.

Reducing power consumptionAttempts to reduce power consumption have

focused on hardware that consumes less power

and software that judiciously manages powerconsumption. Advances in VLSI fabrication andthe reduction of a transistor’s dynamic powerconsumption have led to dramatic decreases inthe power that electronic circuits consume. Thisin turn has allowed pervasive systems to delivergreater capabilities on a fixed power budget, orsimilar capabilities on a smaller power budget.However, in the future, it remains unclear to whatextent this trend will continue. Static power con-sumption is becoming a large fraction of totalpower consumption, because chips will consumemore power even when they are idle, yet we knowof no long-term technical solution. In addition,as users become more accustomed to the conve-nience of wireless capabilities and transfer largeamounts of data over wireless interfaces, physi-cal laws will dictate the minimum amount ofenergy that needs to be transmitted. And VLSIcircuitry can’t overcome these limitations.

Energy-aware software is a compatible tech-nique in which the software is designed to conserveenergy where possible. For example, the softwaremight judiciously power-down components not inuse, or engage power-saving operating modeswhen higher-power modes aren’t warranted todeliver a lower quality of service. The challenge forsuch software is to limit the degree to which theseoptimizations distract users and to limit theamount of user attention and configuration that’sneeded. This problem is further exacerbated whenmobile devices evolve from single-purpose devicesto multipurpose computers with operational char-

Roy Want Intel Research

Keith I. FarkasHewlett-Packard Labs

Chandra NarayanaswamiIBM TJ Watson Research Center

acteristics similar to those of desktopcomputers.

Improving energy sourcesToday, batteries represent the domi-

nant energy source for systems that can’tbe readily powered from a householdelectricity supply or for which powercables haven’t been laid—at a remoteweather station, for example. Althoughexponential improvements have occurredin hardware components (for example,in processor speed, memory density, andnetwork bandwidth), such improvementshaven’t occurred in battery technology,and we don’t anticipate any significantchanges in the future.

At the same time, owing to shrinkingcomponent sizes and the consumer elec-tronics industry’s persistent integrationof more features and capabilities, ourexpectations for mobile systems continueto outpace conventional batteries’ energystorage capabilities. This trend has givenrise to the development of alternativeenergy-storage solutions, such as directmethanol fuel cells. However, these cellsare no panacea because, like batteries,they require user management and hencedistract from the application at hand.Moreover, the energy density of batter-ies in some mobile devices is already veryhigh, and safety concerns arising fromrapid accidental discharge would becomemore serious with significant increases inenergy density.

Energy harvesting: A promising new approach

Fortunately, techniques exist that, giventhe appropriate applications and modesof use, can significantly reduce or simplifyuser involvement in the energy manage-ment process. One such technique isenergy harvesting (or energy scavenging),which aims to collect ambient energy tohelp power systems, possibly storingenergy when it isn’t required—typicallyin batteries, capacitors, or springs.

Many energy transducers exist, eachoffering differing degrees of usefulnessdepending on the application. Perhapsthe best known transducers are solarcells, which have long been used to powersimple hardware components such as cal-culators, emergency telephones along thehighway, and low-grade lighting forpathways at night (charged during theday). Another type of transducer convertsthe energy contained in a vibrating objectinto electrical energy—researchers haveused these to harvest energy from floors,stairs, and equipment housings.

A third type harvests mechanical energy,such as that produced by a person walk-ing and an object’s movement—for exam-ple, self-winding watches, which windthemselves from the swing of a person’sarm. Hybrid cars that transfer energyfrom the engine to the battery duringbraking are a more modern example ofthis technique, which has been used incommercial vehicles. Yet another type oftransducer is one that converts themomentum generated by radioactivereactions into electrical energy. Sometransducers can also exploit temperaturegradients, or pressure gradients, to pro-duce electrical energy.

Energy harvesting is most applicable toapplications that demand small amountsof continuous power or that have shortperiods of high-power use, which previ-ously harvested and stored energy canprovide for. Energy harvesting can alsosupplement more conventional energysources—for example, when a mobile

computer is in a low-power sleep stateor charging its battery. These supple-ments might offer significant contribu-tions to a computer’s energy needs, givena suitable usage pattern. For instance, ifa mobile user extensively uses a com-puter for short periods of time, an energyharvesting system might be able toensure the battery is always topped upduring the standby period when thecomputer isn’t in use.

Although energy harvesting techniquescan significantly benefit some systems andtheir associated applications, they’re lim-ited by the efficiency of the transducersand the raw energy available. As we dis-cussed earlier, a complementary techniquefor extending battery lifetime is to designsoftware that judiciously manages powerconsumption. Energy conservation sum-marizes the numerous techniques in thisarea—such as shutting down disk drivesafter a time-out period or slowing downa processor’s clock. Given that the energycontributions of energy harvesting will bemodest and will likely be serendipitousdepending on the location of use, conser-vation techniques must be used hand-in-hand with energy harvesting to make bestuse of its contribution.

Future trends The semiconductor industry believes

feature sizes for semiconductor deviceswill continue to decrease for severalyears, although the rate at which we canbuild them, owing to significant manu-facturing complexity, is slowing. Theresult will be continued benefits in

JANUARY–MARCH 2005 PERVASIVEcomputing 15

Although exponential improvements have

occurred in hardware components, such

improvements haven’t occurred in battery

technology, and we don’t anticipate any

significant changes in the future.

switching power, but with current leak-age also beginning to play a significantrole. This means that power dissipationrelates not only to the switching fre-quency but also to the area of a chip that’sactively connected to supply lines. Con-sequently, we might have to balance theneed for more capable electronic circuitsthat match user expectations against thepower cost of delivering the associatedincreased integration.

As more VLSI designs are built withpower characteristics that lend them-selves to low-power operation, the com-mercial opportunity for energy harvest-ing will become apparent. This will driveinterest in improving existing energytransducers, perhaps with more efficient

components, or in searching for funda-mentally new materials with improvedenergy conversion properties. Examplesinclude solar cells with greater than 20 to30 percent efficiency, perhaps based onnonsilicon solutions, or a better piezo-electric material than the commonlyused PZT (lead zirconate titanate).

Better understanding of solid-statephysics and refined manufacturing tech-niques will also play a role in improvingefficiency. As demand for the personalcomputer’s processing power continuesto increase, demand for improving thequality of the manufacturing process alsoincreases to guarantee the required highyields of silicon chips containing tens ofmillions of transistors. These improved

processes have a secondary beneficialeffect for other device types, includingthose used as energy transducers.

Although energy harvesting isn’t asolution for all energy needs, it’s a com-pelling concept that will increasinglyenhance or even replace batteries forlower-power applications as low-powercircuit design techniques and transducerenergy efficiencies improve. (Compare,for example, early battery-powered elec-tronic calculators and the solar-poweredproducts sold now.) If the economics areright, for some types of transducers itmight just be a question of applyingknown material characteristics in novelcombinations to improve existing energyharvesting techniques.

16 PERVASIVEcomputing www.computer.org/pervasive

G U E S T E D I T O R S ’ I N T R O D U C T I O N

T he News department’s feature story provides a brief survey

of the commercial uses of energy harvesting, offering insight

into the types of products that are both under development and

currently available. It examines energy harvesting in a broader

context, considering how it not only can replace batteries but also

can reduce energy costs and provide a source of energy when one

isn’t readily available, such as in underdeveloped regions or disas-

ter situations. It describes products ranging from new VLSI tech-

nology and thermoelectrically powered microprocessors to shake

flashlights, wind-up solar-powered radios, and solar-powered

refrigerators.

The first peer-reviewed article, “Energy Scavenging for Mobile

and Wireless Electronics,” is also a survey, but it focuses on the

amount of energy available from various ambient sources and the

technology you can use to harness it. The next article, “Improving

Power Output for Vibration-Based Energy Scavengers,” provides

an in-depth study of how you can harvest energy from mechanical

vibration—one of the previously less studied ambient-energy

sources.

“Battery-Free Wireless Identification and Sensing” demonstrates

how you can further use ultra-high-frequency electromagnetic

energy, generated to read RFID tags, to provide a remote-sens-

ing capability. Then, in “Inductive Telemetry of Multiple Sensor

Modules,” the authors analyze power-induction techniques in

the context of powering RF sensor modules.

Looking to novel forms of power, we can view the topic of the

next article, “Pervasive Power: A Radioisotope-Powered Piezoelec-

tric Generator,” as energy harvesting from low-grade radioactive

materials or as a new type of battery technology. The authors show

how concepts for a harvesting technology and a battery technol-

ogy can be blended together.

Finally, we turn our thoughts to energy conservation with “Ex-

periences in Managing Energy with ECOSystem.” The authors

explain how to effectively integrate energy management into an

operating system without individually customizing each applica-

tion for power efficiency.

Complementing the articles are four Works in Progress contri-

butions. The first reports on ongoing research at UCLA into the

challenges of building practical energy harvesting sensor net-

works. The second reports on ongoing research at ETH Zürich

into extracting user-level information from the rate of energy

harvesting, such as where the user is located or the orientation

of a mobile device. The third, from the University of Pittsburgh,

explores using piezoelectric materials to convert mechanical

energy (from pressure, force, and vibrations) into electricity. The

final WiP, from the Universitat Politècnica de Catalunya, reports

on efforts to harvest passive energy from a user’s daily activity

rather than from a specific action.

In This Issue

Energy harvesting is a nascenttechnology—currently, we canharvest and use only modestamounts of energy. However,

the increasing use of pervasive electron-ics will give this area more attention asresearchers try to overcome the difficultythat arises in managing power for nu-merous devices. Then, as we build newdevices with improved efficiency, theamount of energy we can harvest willincrease, and eventually we should beable to deploy these technologies in awide variety of settings.

We hope you find the articles in this issue(see the related sidebar) interesting andcontribute to further advances that leadto widespread use of this technology.

JANUARY–MARCH 2005 PERVASIVEcomputing 17

the AUTHORS

Roy Want is a principal engineer at Intel Research, where he leads the PersonalServer project. His research interests include ubiquitous computing, wireless proto-cols, hardware design, embedded systems, distributed systems, automatic identifica-tion and microelectromechanical systems. He received his PhD in computer sciencefrom Cambridge University. While at Olivetti Research and later Xerox PARC, he wasbest known for his work on the Active Badge, a system for automatically locatingpeople in a building, and the first context-aware computer system (PARCTab). Con-tact him at 2200 Mission College Blvd., Santa Clara, CA 95052; roy.want@intel.com.

Keith I. Farkas is a senior researcher at Hewlett-Packard Labs. His research interestsspan the areas of personal and enterprise computing and include such topics as soft-ware and hardware techniques for managing energy consumption, microprocessorarchitecture, and software-based techniques that enable the automated control andmanagement of utility computing environments. He is currently working in the lat-ter area, focusing on model infrastructure and increasing the utility of monitoringdata. He received his PhD in electrical and computer engineering from the Univer-sity of Toronto. He is a member of the IEEE and ACM. Contact him at HP Labs, 1501

Page Mill Rd. (MS 1177/3U), Palo Alto, CA 94304; keith.farkas@hp.com.

Chandra Narayanaswami manages the Wearable Computing group at the IBM T.J. Watson ResearchCenter. (His complete biography appears in “From the Editor in Chief,” on page 3.) Contact him atchandras@us.ibm.com.