IEEE Wireless Communications • October 200246 1070-9916/02/$17.00 © 2002 IEEE

ORL/STL active badge p

12.00 1st January 19

Accs 100%DVI Rm. 80%R302 Tue.R321 10:30R435 AwayR310 100%c. AwayF3 Mon.AH 100%AJ 90%Coffee 100%Meet. Rm. 11:20R311 100%

on Prob.

J MartinO MasonD MilwayB MinorsP MitalJ PorterB RobertsoC TurnerR WantM WilkesI WilsonS WrayK Zielinski

Name

An emerging themein pervasive comput-ing is the use of con-text to facilitate andmediate communica-tion among people.Along with theadvantages of ubiq-uitous communica-tion have come newproblems with “stay-ing in touch.”

CO N T E X T-AWARE CO M P U T I N G

INTRODUCTIONAn emerging theme in pervasive computing isthe use of context to facilitate and mediate com-munication among people. Along with the advan-tages of ubiquitous communication have comenew problems with “staying in touch.” Fortu-nately, the convergence of cellular telephony,palm-sized computers, location information, andother sensor data may well provide a basis forcontext-aware solutions to some of consumer’spervasive communication problems. This articlepresents a cross section of research that hasapplied context-aware concepts to reduce com-munication barriers. Our objective is not to pro-vide an exhaustive survey, but rather to give ahistorical perspective, as well as describe somerecent advances.

It is probably no coincidence that PARC’sEtherphone project in the late 1980s and Olivet-ti’s Active Badge location system in the early1990s both pursued call routing to a mobile useras a key application [1, 2]. At the time, beforecell phones were widespread, the notion ofphone calls that could follow people as theymoved was compelling. Even though mobilephones have lessened the need for call routing,many of us still look forward to integrating,coordinating, taming, and, in general, makingour communication technologies even smarter.The approach begun at Xerox PARC and Olivet-ti Research was to add context (i.e., location)into that process, and continuing this agenda

with more types of context will likely be impor-tant for future communication systems.

Along with the early work described above,context-aware communication has roots, inpart, in two other fields of computer scienceresearch, computer supported cooperattivework (CSCW) and human-computer interaction(HCI), and in particular media space researchand awareness systems. As Jonathan Grudinpoints out [3], early media space researchersrecognized the importance of shared context ingroup communication systems. Indeed, thefoundational abstraction “What You See IsWhat I See” (WYSIWIS) aims to support theperipheral context that makes face-to-faceinteraction work so well [4]. In recent yearsresearch on contextualizing collaborative sys-tems has generated an interest in awareness asan independent research focus. For example,the recent work of Hudson [5], Pedersen [6],and others apply abstract visual, or auditorymappings of people’s activities to provide situa-tional and social awareness for others, in partto help them construct communication chan-nels. The influence of CSCW and HCI can beseen in the systems described in this article.

In the next section we present a definition forcontext-aware communication and contrast itwith other forms of context-aware computing.This gives a simple set of dimensions by whichwe discuss a number of context-aware communi-cation systems. We conclude with some chal-lenges and open issues for further research.

DIMENSIONS OFCONTEXT-AWARE COMMUNICATION

Context-aware computing applications examineand react to a user’s changing context in orderto help promote and mediate people’s interac-tions with each other and their environment. Anearly overview paper on context-aware applica-tions from Xerox PARC’s Ubiquitous Comput-ing Initiative laid out the dimensions shown inTable 1 [7]. These dimensions encompass manytypes of context-aware applications, includingcontext-aware software to initiate and facilitatecommunication. Indeed, one of the applicationsfrom that paper, a contextual multi-user whiteboard, is presented later as an example of con-textual group communication. Over the lastdecade, it has become clear that there is a con-

BILL N. SCHILIT, INTEL CORPORATIONDAVID M. HILBERT AND JONATHAN TREVOR, FX PALO ALTO LABORATORY

ABSTRACTThis article describes how the changing infor-

mation about an individual’s location, environ-ment, and social situation can be used to initiateand facilitate people’s interactions with oneanother, individually and in groups. Context-aware communication is contrasted with otherforms of context-aware computing, and we char-acterize applications in terms of design decisionsalong two dimensions: the extent of autonomy incontext sensing and the extent of autonomy incommunication action. A number of context-aware communication applications from theresearch literature are presented in five applica-tion categories. Finally, a number of issues relat-ed to the design of context-aware communicationapplications are presented.

CONTEXT-AWARE COMMUNICATION

IEEE Wireless Communications • October 2002 47

tinuum from manual to automatic, instead ofdiscrete categories.

In this article we focus on context-aware com-munication, which is a subset of context-awarecomputing as it has been described in the litera-ture. However, much that we associate with con-text-aware computing does not involvecommunication. For example, researchers havebeen exploring how context can be used to man-age devices in our environment and to deliver andfilter all types of information from restaurantguides to operating instructions for a nearby copi-er. Neither of these topics is associated with com-munication in the sense we are considering in thisarticle. Nevertheless, the line between informa-tion and communication is not always clear. Forexample, is the Lovegety toy that facilitates con-versation by chirping when a “compatible” personis nearby an information or communicationdevice? This article takes a broad definition ofcommunication and includes these and otherawareness systems that aim to facilitate, in addi-tion to mediate, human-human communication.

We define context-aware communication asthe class of applications that apply knowledge ofpeople’s context to reduce communication barri-ers. We suggest a two-dimensional space forsuch applications based on a simple distinctionbetween “context acquisition” and “communica-tion actions.” Along the “acquisition” dimension,an application might ask people to manuallyenter their context, such as whether they are in ameeting or at lunch, or it may sense and infer aperson’s context with varying levels of autonomyand sophistication. Along the “action” dimen-sion communication might be manually con-trolled. For example, an answering machine thatsays “Lee has been motionless in a dim placewith [low] ambient sound for the last 45 minutes.Continue with call or leave a message [8]” relieson the caller to take manual action. In contrast,applications might act more autonomously, suchas automatically routing a voice call to a nearbyphone. As discussed later, it is not obvious thatapplication designers should simultaneously tryto maximize autonomy in both dimensions sincethis removes human common sense, a qualitythat Tom Erickson describes as “(at best) awful-ly hard to implement.”

The two dimensions in Fig. 1 provide one wayof categorizing various aspects of context-awarecommunication. The table is populated withexamples from the following section. It shouldbe noted that this categorization is only one ofmany possible ways to discuss context-awarecommunication. For example, Nagel et al. atGeorgia Institute of Technology [9] suggest thatstages of communication (initiating, mediating,and terminating) can categorize context-awarecommunication, which is a different yet usefulpoint of view.

CONTEXT-AWARE COMMUNICATION

In the following section we present a range ofcontext-aware communication systems organizedfunctionally. We include five application typesthat have been explored by the research commu-nity: routing, addressing, messaging, providingcaller awareness, and screening. We start with the

function of routing a message or call to an appro-priate nearby communication device, such as anoffice telephone. As we describe these applica-tions and systems we explain their position on thescale from manual to autonomous for contextacquisition and communication action.

ROUTING

Location information has long been used as away to route voice calls. Perhaps the first con-text-aware communications applications weredeveloped at nearly the same time at XeroxPARC and Olivetti Research Labs (ORL) forrouting telephone calls. As shown in Fig. 1 underlabels “Etherphone 1” and “Receptionist assis-tant 1,” these systems began at different designpoints. PARC’s Etherphone had the initialstrength of autonomous action, being able toautomatically route calls, but required manualentry of a person’s location. The Olivetti systemhad the initial strength of automatic person loca-tion, but required manual phone routing. Bothsystems converged on a fully automatedapproach in their second generation, and pro-vide lessons on the difficulty in adding autono-

■ Figure 1. Context-aware communication dimensions. Context (e.g., location)can be entered manually or sensed automatically, and the communication act(e.g., call routing) can be achieved manually, with user assistance, orautonomously. For example, receptionist assistant 1 automatically detects anddisplays user location, but requires a person to forward telephone calls. RegionI systems tend to automate sensing, and region II systems tend to automatecommunication acts.

Co

nte

xt a

cqu

isit

ion

Manual Communication action Autonomous

Autonomous

I

II

Receptionistassistant 1

Activemessenger

RoomotesAwareNexAudio Aura

Virtualwhiteboard

Context-Call andCalls.Calm

Mailinglist

UMD

Reminders

Ether-phone 2

Ether-phone 1

Receptionistassistant 2

■ Table 1. Context-aware software dimensions (adapted from [7]).

Manual Automatic

Information Seeing a selection list of Collaboration channels (e.g.,nearby devices and chat) established based oninformation regarding location and popup messagesnearby places triggered by context

Command “Print” routes to the nearest Mobile computers cache filesprinter onto nearest server

IEEE Wireless Communications • October 200248

my. We end this section on routing by describingubiquitous message delivery, another fullyautonomous approach employing intelligent soft-ware agents.

FOLLOWING CALLERS ONPARC’S ETHERPHONE SYSTEM

In the 1980s researchers at Xerox PARC devel-oped the Etherphone system that used an Ether-net office network, desktop computers, and officephones to provide enhanced functions for trans-mitting, storing, and manipulating digital voice [1].Around 50 Etherphones were deployed in officesat PARC, an environment where people tended tomove from office to office for impromptu meet-ings. When a person visited a colleague, theycould register as a “visitor” using the desktopcomputer interface, and phone calls for themwould ring at their own office as well as the visi-tied location. Alternatively, if an Etherphone userlogged into any Etherphone equipped desktopcomputer, the system would automatically registertheir new visitor location. An unusual aspect ofthe Etherphone system was that each user wasassigned a distinctive ring tone, or motif, such asthe first few bars of “Mary Had a Little Lamb,” sono matter where a call occurred, people were ableto recognize their calls before answering. This wasparticularly useful for the call routing functionbecause it meant that visitors could answer theirown calls and avoid any confusion to the callingparty. In terms of our dimensions in Fig. 1, theearly Etherphone system (Etherphone 1) providedautonomous phone routing (action) but tendedtoward manual location sensing, since visitors hadto manually enter themselves into the system.

Toward the latter part of the project, anOlivetti Active Badge system (see below) wasinstalled at PARC and provided automatic loca-tion information for the Etherphones. This latersystem (Etherphone 2 in Fig. 1) combined auton-omy in both sensing and action dimensions,reducing work for users, but also making the sys-tem more brittle when location sensing didn’t

work quite right. Swinehart tells a story of walk-ing down an active badge enhanced hallway andhearing his ring motif follow him in the officesalong the corridor [10]. Automating Etherphonelocation sensing had another consequence: thecase when call routing was not desired becameexceptions, requiring user action, rather than thedefault, requiring none.

OLIVETTI’S ACTIVE BADGE AIDING ATELEPHONE RECEPTIONIST

Olivetti Research Lab [2] designed and built anovel system for locating people within an officeutilizing infrared emitting “active badges” and anetwork of infrared receivers installed in offices,common areas, and major corridors. The origi-nal software application, an “aid for a telephonereceptionist,” produced a table of names along-side a constantly updating display of locationsand telephone extensions. The display is shownin Fig. 2. The column marked Prob. indicated aprobability that the badge wearer was still at thesighted location based on the number of sight-ings and the time of the last sighting.

In contrast to PARC’s use of Active Badgesfor automated phone routing, the purpose ofthis application was to provide a human recep-tionist with information useful for tracking downand manually routing incoming telephone calls.Even if people were not recently sighted by thesystem, the receptionist could call their lastsighted location to talk with colleagues in thearea and find out if they knew their where-abouts. Whereas PARC started with automationin telecommunications routing, Olivetti beganwith automation in location sensing (see “Recep-tionist assistant 1” in Fig. 1). Manual phonerouting had the advantage of human intelligenceto track down people missing from the badgenetwork, something that would be difficult tobuild into software systems.

Later on, a proof of concept interface wasbuilt to allow certain types of office phone sys-tems to automatically route calls. Olivetti’s sec-ond system (“Receptionist assistant 2”) is similar,in our dimensions, to the second Etherphonesystem. Although the original ORL system didnot take context other than location intoaccount, badge wearers expressed a desire forfiner control. For example, people wanted tocontrol call forwarding based on who they arewith, where they are, and what time of day it is.Personal control scripts were introduced toaddress this need.

It is interesting to note that as automationincreased, the “intelligence” of this systemdecreased, since there was no longer an operatorusing human judgment to track down people.Also, as the system became more autonomous,users wanted more control, but this came at theexpense of more work for users up front in spec-ifying rules and exceptions for call routing.

UBIQUITOUS MESSAGE DELIVERY

Another early example of message routing is theUbiquitous Message Delivery (UMD) applica-tion prototyped at Xerox PARC [11]. A maincontribution of this work was a system architec-

■ Figure 2. The Olivetti Active Badge displayed the locations of people in thelaboratory and was used as an aid for a telephone receptionist to forward callsfrom the main switchboard [2].

ORL/STL active badge project

12.00 1st January 1990

P Ainsworth X343 Accs 100%T Blackie X222 DVI Rm. 80%M Chopping X410 R302 Tue.D Clarke X316 R321 10:30V Falcao X218 R435 AwayD Garnett X232 R310 100%J Gibbons X0 Rec. AwayD Greaves X304 F3 Mon.A Hopper X434 AH 100%A Jackson X308 AJ 90%A Jones X210 Coffee 100%T King X309 Meet. Rm. 11:20D Lioupis X304 R311 100%

Name Location Prob.

J Martin X310 Mc Rm 100%O Mason X307 Lab 77%D Milway X307 Drill AwayB Minors X202 DVI Rm. 10:40P Mital X213 PM 11:20J Porter X396 Lab 100%B Robertson X307 Lab 100%C Turner X307 Lab Mon.R Want X309 Meet Rm. 77%M Wilkes X300 MW 100%I Wilson X307 Lab 100%S Wray X204 SW 11:20K Zielinski X402 Coffee 100%

Name Location Prob.

IEEE Wireless Communications • October 2002 49

ture that provided a level of security in the faceof compromised servers. Text messages sent to auser through UMD were delivered “at the soon-est acceptable time via the most appropriate ter-minal near the recipient.” The system employedactive badges, keyboard input activity, and explic-it commands as a means of detecting user loca-tion. The architecture consisted of user andterminal agents. User location and the user’spolicy regarding message delivery were main-tained by a user agent process. Similarly, sincedesktop terminals have owners, there are alsoterminal agents to manage the policy of out-putting messages on terminals.

Anyone wishing to send a message invokedSendMsg to submit the message to the person’suser agent. The user agent maintains informa-tion about which public and private terminalsthe user is currently accessing, as well as whatpeople are near the user’s location. The useragent can then check if the user’s current situa-tion allows delivery of the message. So, forexample, the user may specify that no low priori-ty messages should be delivered to public termi-nals, or when the user is in the presence of otherpeople. Terminal properties are exported by ter-minal agents so that user agents can make rea-sonable choices, such as delivering a message toa handheld device rather than a desktop displaywhen other people are present. When the user’scontext is suitable for delivery, and a suitableterminal agent exists, the user agent will sendthe message to the terminal agent for display;otherwise, it will wait until a suitable context orterminal agent exists.

The UMD system is primarily an autonomoussystem for both sensing and communicationacts. The architecture describes a user agentthat can encapsulate arbitrarily sophisticatedcomputations for deciding “acceptable” timesand “appropriate” terminals. It is likely that forreal-world use this system would require a set ofvery intelligent heuristics. We note, however,that the UMD design did not address the issueof a user-oriented mechanism for specifyingthese heuristics. Early context-aware systemsdesigners did not tend to focus on the difficultproblem of how autonomous behavior might beachieved in ways both reliable and comprehensi-ble to users.

ADDRESSING

A second function of context-aware communi-cation applications is addressing or determiningwhich people should be included in a communi-cation based on context. Whereas the previoussection on routing concerned contextuallyappropriate physical endpoints, addressing isabout contextually appropriate people. Belowwe describe two types of addressing: mailinglists and shared spaces. These applications aresituated somewhere between manual andautonomous communication action since peo-ple are involved in initiating the communica-tion. It is worth pointing out that communi-cation addressing applications appear “intelli-gent” but don’t seem to need as much work byusers in tuning heuristics as does communica-tion routing.

CONTEXT-AWARE MAILING LIST

In their 1993 paper on uses of location in ubiqui-tous computing [11], Spreitzer and Theimer pro-pose the “note distribution” application to “senda message to all people at a given location.” Asfar as we know, the context-addressed messagewas designed but never implemented by theauthors. However, more recently Dey and othersat the Georgia Institute of Technology developeda more practical version called “Context-AwareMailing List [12].” The Context-Aware MailingList is a dynamic distribution list that can be usedto deliver email messages to members of aresearch group who are currently in the building.The mailing list might be used, for example, tosend a “let’s get lunch” message without annoyingcolleagues not in the building. Dey’s dynamicallychanging mailing list is a new way of addressingemail-style communication to a group of peoplewho are selected by their location.

PARCTAB VIRTUAL WHITEBOARD

Synchronous group communication involves ashared channel or “space,” such as a chat room,where communication takes place. When mem-bership in the space is determined by the contextof the participants, a type of context-aware chatroom is created. The PARCTAB multi-userWhiteboard1 supported real-time communica-tion among a group of people selected by theircommon location and project membership [7,13]. The motivation for this application comesfrom the way people use real-world objects:when a group of people are together in a room,they can easily communicate over the physicalobjects in that place (e.g., a table with scatteredpapers or a whiteboard with diagrams). To pro-mote similar communication in the virtual world,this mobile palmtop computer application pro-vides a freeform ink workspace for each room.All users in the same room see an icon thatselects the room’s whiteboard. In addition, ifmultiple people from a project are collocated, aproject-oriented whiteboard can also be selected.Moving to a different room or into a group ofdifferent people brings up different drawing sur-faces. The contextual addressing of participantsmakes the PARCTAB Whiteboard more power-ful than the physical analogs since the virtualwhiteboard can persist from meeting to meeting,and follow participants from room to room.

MESSAGING

Providing the right information at the right timeis an often-stated goal of context-aware comput-ing applications. Researchers have looked atextending this notion to getting the right mes-sage from another person at the right time. Twotypes of these systems are described below.

CONTEXTUAL REMINDER MESSAGES INCOMMOTION AND CYBREMINDER

Contextual reminder applications that pop up amessage based on the receiver’s situation havebeen explored in a number of research projects.The comMotion [14] and CybreMinder [15] pro-

1 Also called “virtualpaper” and “tab draw” inother papers.

Synchronous groupcommunication

involves ashared channel or

“space,” such as achat room, where

communication takesplace. When

membership in thespace is determined

by the context of theparticipants, a type

of context-awarechat room is created.

IEEE Wireless Communications • October 200250

jects recognized that explicitly supporting anoth-er user to set reminders creates an unusual typeof communication. Both systems allow users toassociate to-do items with locations in the realworld. When the user is in the specified location,an audible cue is played and the user can inspectthe relevant text or audio item. The Cybre-Minder system goes beyond comMotion byallowing more elaborate specifications of contextincluding: “forecast is for rain and Bob is leavinghome”; “Bob and Sally are together”; and “Bobis alone and has free time and the stock price ofIBM is over $100.”

In our categorization, the two contextualreminder/messaging applications describedabove tend to automate both context acquisitionand communication action. CybreMinder allowsdescribing many sophisticated situations, andalthough there is significant overhead for speci-fying situations, the application is unusual in thatthe caller rather than the callee does the work.

MIT’S ACTIVE MESSENGERThe MIT Active Messenger (AM) monitors auser’s incoming email messages, prioritizesthem, and forwards them to phones, pagers,fax machines, and other communication chan-nels near users [16]. If a message is notreceived over one channel (e.g., a pager is outof range), it is resent, after a suitable delay, onanother channel. The AM determines whichchannels should be used based on a user pref-erence file along with the user’s current loca-tion. Although no devices directly reportlocation, the location information is sensedindirectly through caller ID when people phonein to retrieve voice mail or are logged into aworkstation. One unusual aspect of the system,and the reason it is included in this section, isthat AM doesn’t just route messages but alsoprioritizes the mail messages based on calen-dar, contact, and location information, anddecides how to give you a relevant message atthe right time. For example, an email messagesent from a person who works in the San Fran-cisco Bay Area may be deemed “timely” andforwarded to a user’s pager because a calendarentry includes a phone number within the“408” area code or the user has reviewed voicemail from a phone in the Bay Area.

The AM system provides contextual messagesin a compelling way because it takes advantageof the relationships of personal informationstored in calendars, contact lists, and mail, andlinks that to a simple technique for learning auser’s location (caller ID). In our categorizationactive messaging is mostly autonomous in itsaction, and somewhere between manual andautonomous in context acquisition.

PROVIDING AWARENESS

Sharing an awareness of one’s environment andthe context of friends, family, and colleagues canhelp determine if another person is available fora communication, and can also create serendipi-tous opportunities for communication. For exam-ple, Instant Messenger uses a status (Online,Away, On the Phone, etc.) that gives an indica-tion of a person’s availability and willingness to

chat. An example of serendipitous interactionsoccurred in the use of Watchdog [7], an applica-tion that allowed people to play an audio clip(e.g., a rooster crowing) when a sensor at theoffice coffee pot signaled fresh coffee. This con-textual reminder had the side-effect of synchro-nizing people’s visits to the kitchen and fosteringunplanned discussions and meetings. Users ofActive Badge systems are also familiar with thepractice of joining a group (meeting) of col-leagues they saw reported in the same room. Inthe next section we describe three systemsdesigned with awareness in mind.

AWARENESS WITH AWARENEXThe AwareNex system from Sun Laboratories[17] is a portable awareness and communicationtool for wireless PDAs, RIM pagers, and cellphones. As shown in Fig. 3, a contact list inAwareNex displays a list of colleagues and infor-mation about their locale and activities. This listmight show that Cathy was recently in her office,has been idle from her workstation for 23 min-utes, has a scheduled appointment in her calen-dar, and is on her phone. AwareNex goesbeyond mobile instant messaging applicationssince it connects into calendar information,desktop activity, and a telephone switch thatallows it to place calls as well as report if a useris on their phone. In terms of the dimensions,AwareNex does a good job of autonomouslyacquiring context, in part because AwareNexmediates telephone calls so that it can keeptrack of when users are on their cell or officephones. Similar to other awareness tools, thesystem expects users to manually make contact(the communication action), although this isfacilitated by the contact list that not only indi-cates people’s current activities, but also how toreach them.

■ Figure 3. AwareNex shows colleagues and theiractivities. For example, Nicole is in her office,idle for the last 23 minutes on her workstation,and in the middle of a meeting scheduled in hercalendar [17].

Sharing an aware-ness of one’senvironment and thecontext of friends,family, andcolleagues can helpdetermine if anotherperson is availablefor a communication,and can also createserendipitousopportunities forcommunication.

IEEE Wireless Communications • October 2002 51

AUDIO AURA

Audio Aura [18] is less about establishing com-munication and more about augmenting tradi-tional communication mechanisms. The systemuses active badges, wireless headphones, andvarious data sources (e.g., online calendars andemail) to create auditory cues as people walkaround a workplace. For instance, a person stop-ping by a colleague’s office and finding it emptywill hear an auditory cue indicating whether thecolleague has been in today and how long theyhave been away. The system can also produce a“group pulse,” indicating when a group ofcoworkers is currently editing shared groupresources, or whether some coworkers aretogether for a meeting. In terms of dimensions,Audio Aura provides little in the way of auto-mated communication action, but does use auto-mated sensing and casually presentedinformation so people can track down (or just beaware of) their colleague’s activities.

TRIGGERING REAL-WORLDMEETINGS WITH ROOMOTES

Roomotes gives people remote control of theirphysical surroundings through Web phones [19](Fig. 4). The system manages virtual rooms thatmirror physical rooms, and presents not only thedevices in a room, but also the people. Roomotesallows users to control the lighting and audio-video equipment in a conference room from anyWeb phone. An unusual aspect of Roomotes isnotification: users can request that text messagesbe sent to their phone whenever the people orstatus of the devices in a room changes. Thesetext messages sent by the system produce anawareness that can bring people together. Forexample, Bill and David are in the conferenceroom where they are using their Web phone tocontrol the projector and screen. This has theeffect of marking their presence in the confer-ence room. While Bill and David continue to setup for their meeting, Jonathan gets an alert onhis Web phone because he asked Roomotes tonotify him whenever presentation equipment isbeing used in the conference room. By acceptingthe message, Jonathan’s phone jumps to theroom’s page where he sees who is present in theroom, and can select a person and then dialthrough to their cell phone to let them know heis on his way. In terms of our dimensions thissystem provides context acquisition as a side-effect of remote controlling devices in an envi-ronment. The communication is facilitated byletting people connect to others with a singleclick on their Web phone.

SCREENING

Call screening concerns establishing communica-tion under appropriate conditions. This class ofapplication uses context to better inform bothcallers and callees. When initiating conversationsin person we usually pay attention to people’ssituation: who they are with, what they are doing,and where they are. Such context helps to bepolite, but also helps to have a productive con-versation. For example, it is unlikely you will get

an answer to a personal question during a busi-ness meeting, or an answer to a business ques-tion during a family dinner. The Context-Calland Calls.Calm projects described below sharethe similar aim of providing callers with contextinformation so that they can make reasonabledecisions about initiating conversations.

CONTEXT-CALL AND CALLS.CALMINFORM CALLER OF CONTEXT

People use many communication devices tofacilitate communication, but this can often leadto the phenomenon known as “phone tag” wherecallers go back and forth communicating witheach other’s devices but not actually with eachother. Both Context-Call [20] and Calls.calm[21] provide callers information about thecallee’s situation and then rely on callers tomake reasonable choices regarding the time andmechanisms for communication. In Calls.calm,the callee specifies the extent of a caller’s accessto situation information and communicationchannels in a database of relationships. Thespecifications from the relationship databasetogether with current data about the callee’s sit-uation are combined to create a custom interac-tion page that presents context information,communication options, and also short messages.Figure 5 shows an example interaction page.One advantage of Calls.calm is that it allowssmoother transition between synchronous andasynchronous communication, and allows peopleto coordinate on a suitable time to communicatesynchronously. In both these systems, the contextis entered manually by callees, and the commu-nication action is initiated by callers after receiv-ing contextual feedback.

CONCLUSIONS

Building communication applications that havean awareness of people’s context can help reducebarriers that routinely complicate communica-tion between individuals and groups. This articleprovided a sample of research that applies thenotion of context to route, address, message,provide awareness, and screen our communica-tions. We have presented a categorization ofthese systems and characterized them in termsof their level of autonomy with respect to con-text acquisition and communication action.

■ Figure 4. Roomotes is a Web-phone-based universal remote control that noti-fies friends and colleagues of interactions with the physical world. When a per-son enters a virtual room or a physical device in the room changes state, thesystem sends alerts to other people’s cell phones. This creates an awarenessthat brings people together in physical spaces [19].

IEEE Wireless Communications • October 200252

In viewing the systems from the manual-autonomous perspective, we noticed some ten-sions between the goal of autonomy and othersystem goals. On one hand, increasing autonomyof context acquisition is desirable since it reducesthe need for communication recipients to specifytheir context. Increasing autonomy of communi-cation actions is also desirable since it reducesthe need for third-party operators or communica-tion initiators to perform actions. On the otherhand, these benefits do not come without a cost.As context acquisition becomes moreautonomous, recipients may feel their privacy iseroding because systems will become more andmore aware of their day-to-day activities. Thus,reducing the work required from recipients mayalso reduce their sense of privacy. In addition, ascommunication actions become moreautonomous, recipients and initiators alike willnotice a reduction in “common sense” as humansare removed from the loop. Historically, reduc-ing work for third-party operators and initiatorshas shifted the burden back to recipients whomust manage complicated rules and exceptionsto model and deal with real-world complexity.

The systems in this area, while focusing on awide range of problems, suggest a general set ofdesign objectives that designers of context-awarecommunications system should consider:

Improving relevance. Deciding when a com-munication is relevant to the person’s current(or near future) situation; for example, gettingnotification about an email from your travelagent regarding itinerary changes while packingto leave for the airport.

Minimizing disruption. Deciding when andhow to notify people that they have a communi-cation. For example, your phone should vibrateand not ring when you are at the symphony(unless it is truly urgent).

Improving awareness. Deciding which infor-mation and mechanisms can help people makeintelligent communication decisions. For exam-ple, the caller should be told you are at the

movies before the call goes through.Reducing overload. Deciding how to reduce

the number of communications that don’t applygiven your context; for example, filtering outemails about going to lunch when you are awayfrom the office (or already at lunch).

Selecting channels. Deciding which communi-cation device should be used to get in touch withsomebody; for example, routing calls to yourhome phone instead of your cell phone whenyou are at home and cellular reception is poor.

There are many challenges in accomplishingthese objectives in terms of both automatingcontext acquisition and automating communica-tion actions. The research community must con-tinue to identify which context is useful formeeting the objectives and how sensors can reli-ably provide this information. However,automating context acquisition remains a diffi-cult problem because there is a considerable gapbetween what can be sensed and what is “actual-ly going on” in social interactions and people’sminds. For instance, is somebody quiet becausethey are deep in thought, about to make a math-ematical breakthrough, or are they daydreaming,waiting for a call for lunch? For automatingcommunication actions, we need to understandhow the burden of work can be placed appropri-ately (e.g., on the caller is some cases, on thecallee in others) so that the cost doesn’t exceedthe benefit for all parties.

As researchers we should question whethersystems that push on autonomy in both acquisi-tion and action are fundamentally brittle whenfaced with the real world. Is it possible to avoidthis brittleness by using machine learning andstatistical techniques that let people incremental-ly improve communication preferences? Thismay lessen the upfront cost for users, and alsoadapt to users’ changing patterns of behavior.An alternative approach to avoid brittleness is tobalance autonomous with manual mechanisms.The manually assisted sensing in AM (it usesCaller ID to obtain a person’s location) provides

■ Figure 5. Calls.Calm uses Web phones to mediate communication with subscribers. A person (a) selectswho to call and (b) is greeted by the callees contact page contextualized and customized for the caller; or,if the caller is unknown, (c) a generic page. For trusted callers, Calls.Calm reveals status, messages, and alist of preferred communication channels. The system supports negotiating a time to make a voice call byan exchange of short text messages ([21]).

a b c

Automating contextacquisition remainsa difficult problembecause there is aconsiderable gapbetween what canbe sensed and whatis “actually goingon” in socialinteractions andpeople’s minds.

IEEE Wireless Communications • October 2002 53

NOKIA 6100 CELL PHONE

The profile function on the Nokia6100 phones lets users adjust ringtones according to situations andcaller groups (Fig. 6). Up to sevenprofile settings are designed to suitthe different roles in people’s lives:General (Default), Meeting, Out-door Pager, Silent, Car Kit, andHeadset. The profile settings whencombined with priority grouping giveusers sophisticated control of whichcalls they choose to receive. Howev-er, since there are no sensors, theuser must manually set active pro-fi les, although researchers havedemonstrated how to connect a sen-sor unit to this phone [1].

PARENT PAGERCHILD SECURITY SYSTEM

The Parent Pager™ sounds an alertwhen children are about to wanderout of earshot of their parents (Fig.7). The adult’s unit can be slippedinto a pocket or clipped to a belt;the child’s unit is housed in a pouchon an adjustable belt, and can beworn in back to prevent tampering.The adult’s unit has a speciallydesigned short-range setting thatwill signal an alert if the childwanders out of a 10–15-ft radius.The unique sound on the adult’sunit will not be confused withother regular beepers and caneasily be heard in noisy areas.The second longer range on theadult’s unit is up to 50 ft . Inaddition, the Parent Pagerincorporates a pool alarm. Theadult’s unit immediately soundsif the child’s unit becomes sub-merged.

GARMIN RINOPEER-TO-PEER

POSITIONING SYSTEM

The Garmin Rino (Radios Integratedwith Navigation for the Outdoors) is aGPS receiver combined with an FRS(Family Radio Service) communicator(Fig. 8). The Rino has the ability to beamyour location to other Rino users withina 2-mi range using the FRS spectrum.Other users can then see your locationon a detailed map and navigate to yourlocation. The device can store and dis-play downloaded topographic, bathymet-ric, and street-level map information.

LOVEGETY

Lovegety devices help strangers meetand start conversations by alertingtheir owners that someone of theopposite sex, also carrying a Lovegety,is nearby (Fig. 9). When a maleLovegety and a female Lovegetydevice come within 4.5 m (15 ft) ofeach other, the device’s lights flashand a beeper goes off, alerting theowners of a possible rendezvous.Owners set their mode, whether theyare interested in talking, ready to singkaraoke, or ready to do anything (theGet2 mode). Upon encounteringanother Lovegety, the mode of thepotential partner is displayed.

FRIEND FINDERIn November 2001, the Swedishphone company Telia and SignalSoft,Inc., introduced a “friend finder”mobile phone service. FriendFinderuses automatic location via the cellu-lar phone network to bring “youthfulpeople with active social lives” togeth-er. People subscribing to the serviceset up group lists, similar to buddylists, which includes their friends,coworkers, or other lists. Then the

subscriber sends an SMS mes-sage to search for friends on theirlist. A return message informsthe subscriber of the location ofthe people on the list, and thensubscribers can communicateindividually or collectively withtheir friends.

REFERENCES[1] A. Schmidt et al., “Advanced Interac-

tion in Context,” Proc. 1st Int’l.Symp. Handheld and UbiquitousComputing (HUC’99), vol. 1707 ofLecture Notes in Computer Science,Springer-Verlag, pp. 89–101.

■ Figure 7. The Parent Pager is a child safety prod-uct that activates an alarm when a child wandersmore than 15 ft from the adult unit.

■ Figure 8. Garmin's Rino GPS Radio gives usersthe ability to beam their location to other Rinousers within a two-mile range and carry on con-versations at the same time.

■ Figure 9. Lovegety. A Japanese toy for meetingpeople, it beeps when a compatible partner isnearby.

■ Figure 6. TheNokia 6150 gives auser’s profile settings (e.g.,general, meeting,outdoor, car) thatcan be used in conjunction withcaller groups toprovidesophisticatednotification.

CONTEXT-AWARE COMMUNICATION PRODUCTS

IEEE Wireless Communications • October 200254

such a balance because people accessing voicemail are logically, at the same time, also interest-ed in other communication information.

Context-aware communication has made con-siderable progress in reducing barriers to com-munication, but there are still significantchallenges to be overcome. In the future, thegoal is a context-aware communication systemthat refuses to ring your phone at the operaunless it’s the babysitter calling to say your kidsjust set the house on fire.

REFERENCES[1] D. Swinehart, "Telephone Management in the Ether-

phone System," Proc. IEEE/IEICE Global Telecommun.Conf., Tokyo, Japan, Nov. 1987, pp. 1176–80.

[2] R. Want et al.,"The Active Badge Location System," ACMTrans. Info. Sys., vol. 10, no. 1, Jan. 1992, pp. 91–102.

[3] J. Grudin, "Desituating Action: Digital Representation ofContext," Special issue on Context-Aware Computing,Human-Comp. Interaction J., vol. 16, 2001.

[4] M. Stefik et al., "WYSIWIS Revised: Early Experienceswith Multiuser Interfaces," ACM Trans. Office Info. Sys.,vol. 5, no. 2, Apr. 1987, pp. 147–67.

[5] S. E. Hudson and I. Smith. "Techniques for AddressingFundamental Privacy and Disruption Tradeoffs inAwareness Support Systems," Proc. Comp. SupportedCooperative Work, 1996, pp. 248–57.

[6] E. R. Pedersen and T. Sokoler. "AROMA: Abstract Repre-sentation of Presence Supporting Mutual Awareness,"Proc. SIGCHI Conf. Human Factors in Comp. Sys. ,Atlanta, GA, Mar. 22–27, 1997, pp. 51–58.

[7] B. N. Schilit, N. Adams, and R. Want. "Context AwareComputing Applications," Proc. Wksp. Mobile Comp.Sys. and Apps., Santa Cruz, CA, Dec. 1994, pp. 85–90.

[8] T. Erickson, "Some Problems with the Notion of Con-text-Aware Computing," Commun. ACM, vol. 45, no. 2.Feb. 2002, pp. 102–4.

[9] K. Nagel et al., "The Family Intercom: Developing a Con-text-Aware Audio Communication System," Proc. Int'lConf. Ubiq. Comp., Atlanta, GA, Sept. 2001.

[10] D. Swinehart, Personal communication, Feb. 28, 2002.[11] M. Spreitzer and M. Theimer, "Providing Location

Information in a Ubiquitous Computing Environment,"Proc. 14th ACM Symp. Op. Sys. Principles, 1993, pp.270–83; also in Mobile Computing, T. Imielinski and H.Korth, Eds. Dordrecht, The Netherlands: Kluwer, 1996,pp. 397–423.

[12] A. K. Dey, G. D. Abowd, and D. Salber, "A ConceptualFramework and a Toolkit for Supporting the Rapid Pro-totyping of Context-Aware Applications," Special issueon Context-Aware Computing, Human-Comp. Interac-tion J., vol. 16, 2001, pp. 97–166.

[13] B. N. Schilit, "A Context-Aware System Architecture forMobile Distributed Computing," Ph.D. thesis, ColumbiaUniv., May 1995.

[14] N. Marmasse and C. Schmandt, "Location-aware Infor-mation Delivery with ComMotion," Proc. 2nd Int'l.Symp. Handheld and Ubiq. Comp., Bristol, U.K., Sept.2000. Springer Verlag, pp. 157–71.

[15] A. K. Dey and G. D. Abowd, "CybreMinder: A Context-Aware System for Supporting Reminders," Proc. 2ndInt'l. Symp. Handheld and Ubiq. Comp., Bristol, U.K.,Sept. 2000, pp. 172–86.

[16] C. Schmandt et al., "Everywhere Messaging," IBM Sys.J. 2000, vol. 39, nos. 3 and 4, pp. 660–77.

[17] J. Tang, N. Yankelovich et al., "ConNexus to Awarenex:Extending Awareness to Mobile Users," Proc. SIGCHIConf. Human Factors in Comp. Sys., 2001, Seattle, WA,Mar. 31–Apr. 5, 2001, pp. 221–28.

[18] E. D. Mynatt et al., "Designing Audio Aura," Proc.SIGCHI Conf. Human Factors in Comp. Sys., Apr. 1998,pp. 566–73.

[19] R. Wakikawa et al., "Roomotes: Ubiquitous Room-based Remote Control from Cell Phones," ExtendedAbstracts of the SIGCHI Conf. Human Factors in Comp.Sys., Seattle, WA, Mar. 31–Apr. 5, 2001, pp. 239–40.

[20] A. Schmidt, A. Takaluoma and J. Mäntyjärvi, "Context-Aware Telephony over WAP," Personal Technologies,vol. 4, no. 4, Sept. 2000, pp. 225–29.

[21] E. R. Pedersen, "Calls.calm: Enabling Caller and Calleeto Collaborate," Proc. SIGCHI Conf. Human Factors inComp. Sys., Apr. 1–4, 2001, Seattle, WA, pp. 235–36.

ADDITIONAL READING[1] R. Want et al., “The PARCTAB Ubiquitous Computing

Experiment,” Mobile Computing, H. F. Korth and T.Imielinski, Eds., Kluwer, 1996.

BIOGRAPHIESBILL SCHILIT (bill.schilit@intel.com) is a principal researcherand co-director of Intel Research Seattle. His research focus-es on ubiquitous and proactive computing applications. Hiswork is positioned at the intersection of networking andhuman-computer interaction. Prior to joining Intel, he man-aged the Personal and Mobile Computing Group at FX PaloAlto Laboratory, a Fuji Xerox Company. He was also aresearcher at AT&T Bell Labs and Xerox Palo Alto ResearchCenter where he championed the notion of location-awarecomputing and was part of the team that built the PARCTAB,the first context-aware handheld computer. He serves on theeditorial boards of IEEE Computer and IEEE Wireless Com-munications.

DAVID HILBERT [M] (hilbert@fxpal.com) is a research scientist atFX Palo Alto Laboratory (FXPAL). His research interests lie inthe design and evaluation of practical HCI, CSCW, and ubiq-uitous computing applications. His Ph.D. work at the Univer-sity of California at Irvine explored techniques for capturingapplication usage data on a large scale to help improve thefit between application design and use. Before joining FXPAL,he also worked as a software engineer at NASA’s Jet Propul-sion Laboratory and as a program manager at Microsoft. Heis a member of the ACM and Phi Beta Kappa.

JONATHAN TREVOR (trevor@fxpal.com) is a senior research sci-entist at FX Palo Alto Laboratory (FXPAL) working in theareas of ubiquitous systems and computer-supported coop-erative work. For his Ph.D. at the CSCW Research Center,Lancaster University, he developed an infrastructure fordeveloping cooperative systems. While at GMD, Germany, heco-developed BSCW, a Web-based shared workspace systemthat gained the European Software Innovation Award in1996. He returned to Lancaster to work on a number ofEurope-wide projects, including eSCAPE, an electronic land-scape providing interconnections to other virtual environ-ments. His current research at FXPAL continues within aninterdisciplinary team, and focuses on the development ofreadily accessible groupware and HCI applications across awide-range of technologies and platforms.

In the future, thegoal is a context-aware communica-tion system thatrefuses to ring yourphone at the opera,unless it’s thebabysitter calling tosay your kids just setthe house on fire.