Special Theme: Ambient Intelligence

European Research Consortium for Informatics and Mathematicswww.ercim.org

Number 47 October 2001

Special Theme:

Next Issue: January 2002

Special Theme:e-government

AmbientIntelligence

CONTENTS

Cover image: The Innovation Center for the Intelligent House, a testbed for future forms of living and working. See article on page 17.

KEYNOTE

3 by Commissioner Erkki Liikanen

JOINT ERCIM ACTIONS

4 DELOS Network of Excellence on Digital Libraries

7 ERCIM Fellowship Programme 2002/2003 launched

SPECIAL THEME

8 Ambient Intelligenceby Jari Ahola

9 Ambient Intelligence: a UK Perspectiveby Ronan Sleep

10 Ambient Intelligence in the Context of UniversalAccessby Constantine Stephanidis

12 Ambient Intelligence within a Home Environment by Mária Bieliková and Tibor Krajcovic

14 A Flood of Intelligence - the Living Room Project by Jukka Vanhala

16 Ambient Intelligence and Home Networking forWellness Management and Home Automationby Ilkka Korhonen

17 Wired and Smart: from the Fridge to the Bathtubby Franz Miller

18 Enabling Ambient Intelligence Research with SoapBoxPlatform by Esa Tuulari

20 Designing for Ludic Aspects of Everyday Lifeby Bill Gaver

21 Ubiquitous Service Environmentsby Carl Gustaf Jansson and Peter Lönnqvist

23 Offering a Consumer-Oriented Ambient IntelligenceEnvironmentby Valerie Issarny

24 Global Ambient Intelligence: The Hostess Conceptby András Lörincz

25 Ubiquitous Computing Infrastructures by Friedemann Mattern

26 Ubiquitous Computing and Embedded OperatingSystems Designby Michel Banâtre

28 Ambient Interfaces for Distributed Work Groups by Tom Gross

29 GeoNotes: Social Enhancement of Physical Space by Per Persson and Fredrik Espinoza

30 Contextualisation in Nomadic Computingby Markus Eisenhauer and Roland Klemke

31 Scene Interpretation for Video Communicationby Monique Thonnat

32 Mining Multimedia Documents by Visual Content by Nozha Boujemaa and Marin Ferecatu

34 How can I be sure that my DVD player understands my TV? by Wan Fokkink, Izak van Langevelde, Bas Luttikand Yaroslav Usenko

36 Fault Diagnosis in Wireless Sensor Networksby Stefano Chessa and Paolo Santi

TECHNOLOGY TRANSFER

37 Domain Specific Languagesby Arie van Deursen and Jan Heering

38 QuantifiCare: A new INRIA spin-off in Medical ImageProcessingby Jean-Philippe Thirion

39 MIA — Querying Multimedia Databasesby Arjen de Vries

EVENTS

40 IEA/AIE-2001 Conference Reportby László Monostori and József Váncza

40 Sixth ERCIM FMICS International Workshopby Hubert Garavel

41 Workshop on User Interfaces to Data Intensive Systems by Epaminondas Kapetanios

42 1st International Conference on Universal Access inHuman-Computer Interactionby Constantine Stephanidis

42 ECOOP 2001 - 15th European Conference on Object-Oriented Programmingby László Kozma

43 First DELOS Summer School on Digital LibraryTechnologiesby Maristella Agosti and Vittore Casarosa

44 Joint DELOS-NSF Workshop on Personalisation and Recommender Systems in Digital Libraries by Alan Smeaton and Jamie Callan

45 CLEF2001: Workshop of the Cross-LanguageEvaluation Forumby Carol Peters

46 Announcements

47 IN BRIEF

The EuropeanCommission’sproposal for

the next FrameworkProgramme (2003-2006) for Researchand Development(FP6) was submitted

to the Council of ministers and European Parliament inFebruary 2001 and the related Specific programmes in May2001. Information Society Technologies (IST) are one of themain priorities in the Commission proposal given their funda-mental role in realising Europe’s objectives for the knowledgesociety as agreed at the Lisbon European Council of 2000, theStockholm Council of 2001, and reflected in the e-EuropeAction Plan. This renewed effort in RTD in IST is essential toensure European leadership in the generic and applied tech-nologies at the heart of the knowledge society, to improveEuropean competitiveness and to enable all European citizensto benefit from the knowledge society.

An important challenge for the academic and industrialresearch community in IST is to maintain the focus on mediumto long-term objectives amid changing socio-economic con-texts. We know that today only a small part of the world popu-lation, between 5 and 10 %, have access to IST applications andservices. Costs, complexity, unavailability and unreliability areoften preventing the further development and broader deploy-ment of the knowledge and information society and the digitaldivide is widening. Even in developed countries, only a minorfraction of the possibilities that IST can offer, is actually useddespite recent progress in the uptake of internet and mobiletechnology.

As we start to understand the advantages and limitations ofcurrent technologies, research in IST should aim at newavenues that will not only extend the scope, functionality andefficiency of IST applications and services, but that will alsomake them available, in the most natural and trustful way, tocitizens whoever they are, whatever their age is, anywhere andanytime. It must lead to a world where every child has access topersonalised learning resources, where every patient can betreated within the comfort of his own home, where every engi-neer has the power of global computing resources at their fin-gertips, where every business is plugged into world-widetrading communities; and where every citizen is able to accesspublic services anywhere, anytime.

This is the world of ‘Ambient Intelligence’ that will, graduallybut surely, emerge from research in IST. This is the vision thatthe IST priority in FP6 is following. It puts people at the centre

of the development of future IST, ie, ‘design technologies forpeople and not make people adapt to technologies’. It aims atmaking technology invisible, embedded in our natural sur-rounding and present whenever we need it (eg, electricity) andat making interaction with the technology simple, effortlessand using all our senses. This vision sustains and extends theobjectives of the European Commission’s eEurope 2002Action Plan of bringing IST applications and services toeveryone, every home, school and to all businesses.

The trend towards an Ambient Intelligence landscape providesa clear opportunity for European industry to build on, andstrengthen its leading position in areas such as mobile commu-nications, consumer electronics and home appliances,embedded software and microelectronics. It will help reinforcethe competitiveness of all industrial sectors.

Europe can build on pioneering work in the field such as theEU funded initiative on the ‘disappearing computer’ and onindustrial and technological achievements. A main stepforward is already brought in mobile/wireless technologieswith the 2.5 and 3G mobile systems. These provide anywhereaccess to applications and services from a non-PC platform andfor the normal ‘man and woman on the street’. Wearablemobile devices that incorporate interfaces making use of oursenses such as speech and gesture are currently under develop-ment. Cars already encompass an expanse of IST devices thatwe use and to which we have access without knowing.

A sustained research effort is needed to accelerate the progressin key areas such as advanced interfaces, broadband mobile,wireless and optical communications, distributed andembedded computing technologies and knowledge handlingtechniques. It is also required in order to push the limits ofminiaturisation and minimise the costs and power consumptionof microelectronic components and to explore new materials,such as organic flexible material for displays and sensors sothat they can be placed anywhere and take any shape.

Europe can not afford to miss the unique opportunities offeredby this next generation of IST, either in social or economicterms. The stake is even higher given the role of IST in all otherresearch and engineering fields and their impact on the realisa-tion of the European Research Area. We hope that the supportto IST in FP6 will provide further impetus to the members ofERCIM and the whole research community, from industry andacademia, to pull together their effort and build coherentapproaches to realise the vision.

3ERCIM News No. 47, October 2001

Ambient Intelligence in future EU Research Efforts

Erkki LiikanenMember of the European Commission,responsible for Enterprise and Information Society

KEYNOTE

4 ERCIM News 47, October 2001

JOINT ERCIM ACTIONS

This is the second of a series of articles reporting on past andupcoming activities of the DELOS Network of Excellence for Digital

Libraries. Further information on the objectives, activities andmembership of DELOS can be accessed from the DELOS home page:

http://www.ercim.org/delos/

Digital Libraries: FutureResearch Directions for a European ResearchProgrammeA brainstorming meeting took place on 13-15 June 2001 in San Cassiano(Dolomites), Italy.

The objective of the meeting was tooutline the main research directions of afuture European research programme inthe field of digital libraries. To this end,prominent members of the Europeanresearch community were invited to par-ticipate in and contribute to the defini-tion of this new vision of digitallibraries, and to translate it into concreteresearch actions which will become anintegral part of the next 6th FP. Someprominent members from the USresearch community were also invited,in collaboration with the NationalScience Foundation. 15 EU and 5 UShigh-level researchers attended. Themeeting was also attended by a repre-sentative from the Commission.

The meeting was divided into three ses-sions. On the first day a plenary sessionwas held where the main objective ofthe meeting was presented. All partici-pants briefly presented short positionstatements reflecting their views aboutthe main advances expected over thenext few years in their field of activityand expertise, and how these develop-ments can contribute to the generationof new systems and services. The aim ofthis session was to provide a commonbackground to facilitate the drafting ofthe final results of the meeting. On thesecond day three parallel sessions wereheld to discuss and draft a set of recom-mendations for a future research agendarelevant to Digital Libraries. A rappor-

teur was assigned for each group. Onthe third half-day a final plenary sessionwas held, where the rapporteur of eachgroup presented the proposals and rec-ommendations of his/her group. A dis-cussion was then held to try to consoli-date all the proposals in order to draft areport on the outcome of the meeting.

The agenda (including slide presenta-tions), position statements and list ofparticipants are available on the delosweb site. A report on the outcome of themeeting providing recommendations onthe main research directions of a futureEuropean research programme in thefield of digital libraries will be availableat the site in October 2001.

DELOS / IPI-RAS / IITEWorkshop on European-Russian Cooperation in Digital Libraries A DELOS-Russia CollaborationWorkshop took place on 7-8 June 2001in Moscow, Russia.

The Workshop was organized in theframework of the DELOS ‘Inter-national Cooperation’ Forum accordingto the DELOS-Russian DL Programme.The objective of the 2001 Workshop inRussia was to organize a meetingbetween European and Russian repre-sentatives of active DL projects to iden-tify areas of common interest and poten-tial cooperation. In May 2001 the list ofparticipating projects from both sideswas fixed. The list was based on theEuropean projects that in Luxembourg,at the EU-DL All Projects Coordinationmeeting in February 2001, hadexpressed interest in possiblecooperation with Russian DL projects.

At that time, the UNESCO Institute forInformation Technologies in Education(IITE) proposed to host the workshop inMoscow, including in the agenda aUNESCO Expert Meeting on DL inEducation.

The following European projects wererepresented at the Workshop:ARTISTE, SCHOLNET, TEL, MIND,ARION, RENARDUS, ECHO, GLOBALINFO (now DL-Forum). The followingRussian projects from Moscow, St.Petersburg and Novosibirsk were repre-sented at the Workshop: RSL project,Integration of Infosources of RAS,Subject Domain Mediation, GENEEX-PRESS, Cultural heritage network,Cultivate - Russia, Russian ArchivesOnline, AEROSPACE MEDIAT-ECHA, Open Education: Models. Thetopics discussed helped to identify sig-nificant overlap in the interests ofEuropean and Russian DL profes-sionals.

After the presentations and discussions,the Workshop formulated and approvedseveral recommendations for concreteactions, including:• Forming a DELOS-Russia network

which, on the one hand, will be anode in the DELOS Network ofExcellence and, on the other hand,will be a network of Russian(sub)nodes with activities similar tothose of DELOS proper. DELOS-Russia may be analogous to theCULTIVATE-Russia network.

• Improving and intensifying theexchange of research results anddevelopment efforts in the area ofDL between existing European andRussian projects, building onexisting DELOS and UNESCO IITEframeworks to establish cooperationbetween existing projects or to form

ERCIM News No. 47, October 2001 5

JOINT ERCIM ACTIONS

new joint projects within the currentand future funding programs.

• Identifying joint activity areas thatdeserve to be generalized andextended to the level of DELOSworking groups. DL frameworks forscientific collection support can besuggested as one such potentialdirection.

Tutorial on MultilingualAccess for InformationSystems

A DELOS tutorial on Multilingual Accessfor Information Systems was held atIFLA2001 (International Federation ofLibrary Associations and Institutions) on 23 August 2001 in Boston,Massachusetts.

40 participants attended the tutorialwhich provided an overview of themain issues of interest in the followingtwo sectors:• multiple language recognition,

manipulation and display • multilingual or cross-language

search and retrieval.

The topics covered included: the spe-cific requirements of particular lan-guages and scripts, processing of multi-lingual document collections, multilin-gual metadata, techniques for cross-lan-guage retrieval, presentation issues,system evaluation. The focus of thetutorial was on the needs of the user -whether information seeker, provider,or administrator - and on how today'smultilingual retrieval systems canimpact on each of these user types. Thetutorialists were Páraic Sheridan(Director of MNIS-TextWise Labs, theR&D division of Manning & NapierInformation Services based in Syracuse,New York) and Carol Peters(Researcher at the IEI-CNR).

Workshop on Audio-VisualMetadata Modelling

A DELOS workshop on Audio-VisualMetadata Modelling will be held on 23September 2001 in London, UK in con-junction with the FIAT/IFTA AnnualConference (London, 22-25September). The workshop will bedivided into two sessions. In the firstsession, research results and on-goingstandardization activities will bereported. In the second session, a panelof leading representatives from applica-tion domains (eg broadcasting,archives, movie industry) will presentand discuss real-life experiences withmetadata models and editors in order tostimulate a discussion and an exchangeof opinions with the audience.

Workshop on Collaborationbetween DELOS and Non-EU MediterraneanCountries A DELOS workshop on the Design andImplementation of Digital Libraries willbe held at the Alakhawayn University inIfrane (AUI), Morocco on November 8-9, 2001. This practical workshopaims at bringing together DigitalLibraries’ researchers from both sidesof the Mediterranean to present anddiscuss their experiences in designing,implementing, and maintaining digitallibraries with respect to today’s techno-logical and economic environment. Themain objective of the workshop is tofacilitate an exchange of experiencesamong researchers involved in tech-nologies of interest to digital libraries,to explore and discuss subjects ofcommon interest for possible interna-tional cooperation, and to allow the AUIlibrary to learn from other practicalexperiences and to have a good andclear idea on how to monitor the designand the implementation of its owndigital library.

Setting up of DELOS-NSFJoint Working Groups

DELOS and the NSF have agreed to setup seven joint working groups whoseaim is to work together on researchareas of important relevance to thedigital library community. Eachworking group will be co-chaired by aprominent member of the European andthe US digital library research commu-nity, and will consist of approximately10 members. The joint working groupspresently being set up are:• Spoken-Word Digital Audio

Collections• Information Seeking, Searching

and Querying in Digital Libraries• Personalization and Recommender

Systems in Digital Libraries• Emerging Language Technologies

and Cultural Heritage• Digital Imagery for Significant

Cultural and Historical Materials• Preservation and Archiving• Evaluation and Test Suites.

NSF / EU DL All Projects MeetingDELOS is planning to hold a NSF/EUAll Projects meeting in Rome, Italy 10-11 January 2002 (tentative date). TheEuropean Commission’s InformationSociety Technologies Programme andthe National Science Foundation aresupporting substantial research con-cerned with digital libraries. Bothexpect to develop the enabling tech-nologies as well as principles for DLdesign and operation, to study the eco-nomic and legal issues surroundingelectronic distribution of IR, and tobetter understand the social context - forexample, human information needs - inwhich DLs will operate. By and large,scientists involved in the EU and NSFprojects have not had the opportunity tomeet regularly. A coordination effort inthe DL field can help avoid duplicationof effort, prevent the development offragmented digital systems, andencourage productive interchange ofscientific knowledge and scholarly dataaround the world.

JOINT ERCIM ACTIONS

In this context, DELOS is planning tohold the first of a series of annual meet-ings to bring together scientistsinvolved in European and NSF pro-jects. The objectives are to exchangeworking experiences and researchresults in the field of digital libraries,and to establish collaborative linksbetween on-going projects. From theUS side, participants will be represen-tatives from the Digital LibraryInitiative (phase2) projects. From theEuropean side, representatives will befrom EU funded Digital Library pro-jects and from DL projects funded byEuropean National Initiatives.

The second DELOS workshop onPersonalization and RecommenderSystems in Digital Libraries was heldon 18-20 June 2001 in Dublin, Ireland(see article on page 44).

The first DELOS InternationalSummer School on Digital LibraryTechnologies was held on 9-13 July 2001in Pisa, Italy (see article on page 43).

A CLEF 2001 Workshop of theCross-Language Evaluation Forumwas held on 3-4 September 2001 inDarmstadt, Germany (see article onpage 45).

The Fifth European Conference onDigital Libraries (ECDL2001) washeld in Darmstadt, Germany on 5-7September 2001. An article on the con-ference will appear in the next issue ofERCIM News.

The Third DELOS Workshop onInteroperability and Mediation inHeterogeneous Digital Libraries washeld on 8-9 September 2001 inDarmstadt, Germany, in conjunctionwith ECDL2001. An article on theworkshop will appear in the next issueof ERCIM News.

The DELOS Network of Excellence onDigital Libraries, funded by the 5thFramework Programme of theEuropean Commission, provides aframework to foster national andinternational research activities in theDigital Libraries domain and todisseminate the research results in therelevant user communities.

The activities of the DELOS Network ofExcellence are organized into fiveForums: the Research Forum, theEvaluation Forum, the StandardizationForum, the Technology Transfer Forumand the International CooperationForum.

The activities of the DELOS Network ofExcellence are open to all thoseworking or interested in DL-relatedresearch or applications.

Delos website:http://www.ercim.org/delos

Please contact: Costantino Thanos — IEI-CNRNetwork DirectorTel: +39 050 315 2910/2912E-mail: [email protected]

Order Form

If you wish to subscribe to ERCIM News

free of chargeor if you know of a colleague who would

like to receive regular copies of ERCIM

News, please fill in this form and we will

add you/them to the mailing list.

send, fax or email this form to:

ERCIM NEWSDomaine de Voluceau

RocquencourtBP 105

F-78153 Le Chesnay CedexFax: +33 1 3963 5052

E-mail: [email protected]

Data from this form will be held on a computer database. By giving your email address,

you allow ERCIM to send you email

Name:

Organisation/Company:

Address:

Post Code:

City:

Country

E-mail:

✄

You can also subscribe to ERCIM News and order back copies by filling out the form at the ERCIM web site at http://www.ercim.org/publication/Ercim_News/


JOINT ERCIM ACTIONS

The objective of the FellowshipProgramme is to enable bright young sci-entists from all over the world to workcollectively on a challenging problem asfellows of leading European researchcenters. In addition, an ERCIM fellow-ship helps widen and intensify thenetwork of personal relations and under-standing among scientists. The pro-gramme offers the opportunity toERCIM fellows: • to improve their knowledge about

European research structures andnetworks

• to familiarize them with workingconditions in leading Europeanresearch centres

• to promote cross-fertilization andcooperation between research groupsworking in similar areas in differentlaboratories, through the fellowships.

Conditions Aside from researchers from academicinstitutions, scientists working inindustry are strongly encouraged toapply. Applicants must: • have a PhD degree (or equivalent) or

be in the last year of the thesis work• be fluent in English • be discharged or get deferment from

military service • start the grant before October 2002

Fellowships are of 18 months duration,generally spent in two institutes. For theentire period of 18 months, the fellowwill receive a monthly allowance whichmay vary depending on the country.Costs for travelling to and from the insti-tutes will be paid. In order to encouragethe mobility, a member institution willnot be eligible to host a candidate of thesame nationality.

Deadlines for ApplicationThe Call is divided into two stages withthe deadlines: • 31 October 2001• 30 April 2002.

How to applyApplications must be submitted elec-tronically:http://www.ercim.org/fellowship/

Selection ProcedureEach application is reviewed by one ormore senior scientists in each ERCIMinstitute. ERCIM representatives con-sider the results and decide to which can-didates fellowships should be offered,taking into account: • the qualification of the applicant • the overlap of interest between

applicant and the hosting institution • available funding.

Research InfrastructureAt the ERCIM institutes, state-of-the-artcomputer hardware and software is available to support the major areas ofresearch listed below. Library facilitiesand access to scientific databases are seenas another key component in the researchenvironment. All ERCIM institutesprovide access to excellent library facili-ties in IT and mathematics. All ERCIMinstitutes have extensive seminar andresearch training programmes.

Research Areas of InterestFellowships are available in the scien-tific fields covered by the ERCIMWorking Groups the following areas: • Database Research• Constraints Technology and

Applications• Control and Systems Theory• Formal Methods• Electronic Commerce• User Interfaces for All• Environmental Modelling• Health and Information Technology• Digital Libraries• E-Learning• Matrix Computations and Statistics.

ERCIM institutes covering a broad rangeof research topics, fellowships are alsoavailable in areas not covered by the

above topics. ERCIM is particularlyinterested in candidates with skills in:• Mathematics and Foundations of

Computer Science: combinatorics,graph theory, theoretical computerscience, application of number theoryto security, computer algebra,stochastics, theory of computation

• Software: real time and highperformance programming, softwarespecification, analysis and testing,operating systems, distributed andparallel systems, interactive softwareand systems

• Computing methodologies: speech,visualisation and virtual reality,image analysis

• Communication technology:communication networks, networkarchitecture and management.

A detailed list of topics in which ERCIMinstitutes are active is available athttp://www.ercim.org/activity/expertise.html

Links: Detailed description and online application form: http://www.ercim.org/fellowhsip/

Please contact: ERCIM officeTel: +33 4 92 38 50 10E-mail: [email protected]

ERCIM offers 18-month fellowships in leading European informationtechnology research centres. This offer is available for PhD holders (or equivalent) from all over the world.

ERCIM Fellowship Programme 2002/2003 launched

8 ERCIM News No. 47, October 2001

Since the 1999 IST ProgrammeAdvisory Group (ISTAG) vision state-ment for Framework Programme 5 chal-lenging to create an AmbientIntelligence (AmI) landscape for seam-less delivery of services and applicationsin Europe, it rapidly became widelyembedded in the work programme foryears 2000-2001. AmI is also recognisedas one of the key concepts related toInformation Society in the FrameworkProgramme 6 and as we can see from themultitude of articles in this issue,ERCIM members are already well onthis track. For a more detailed intro, ourfirst contact with AmI is envisioned in areport by ISTAG (ISTAG. Scenarios forAmbient Intelligence in 2010. FinalReport, Feb 2001, EC 2001. Availableat: http://www.cordis.lu/ ist/istag.htm)using a set of scenarios depicting dif-ferent potential futures with four ficti-tious users.

Ambient Intelligence builds on threerecent key technologies: UbiquitousComputing, Ubiquitous Communicationand Intelligent User Interfaces – some ofthese concepts are barely a decade oldand this reflects on the focus of currentimplementations of AmI (more on thislater on). Ubiquitous Computing meansintegration of microprocessors intoeveryday objects like furniture, clothing,white goods, toys, even paint.Ubiquitous Communication enablesthese objects to communicate with eachother and the user by means of ad-hocand wireless networking. An IntelligentUser Interface enables the inhabitants ofthe AmI environment to control andinteract with the environment in a natural(voice, gestures) and personalised way(preferences, context).

Making AmI real is no easy task: as itcommonly takes place with a new tech-nology, soon after high-flying visions weare demonstrated with the first pieces ofhardware for the intelligent environ-ment. However, making a door knobable to compute and communicate doesnot make it intelligent: the key (and chal-lenge) to really adding wit to the envi-ronment lies in the way how the systemlearns and keeps up to date with theneeds of the user by itself. A thinkingmachine, you might conclude – not quitebut close: if you rely on the intelligentenvironment you expect it to operate cor-rectly every time without tedioustraining or updates and management.You might be willing to do it once butnot constantly even in the case of fre-quent changes of objects, inhabitants orpreferences in the environment. Alearning machine, I'll say.

The following articles in this specialtheme issue showcase the variousaspects of AmI research in Europe. Inaddition to background information onAmI related activities within the ERCIMmembers we have a number of articleson the infrastructure for AmI environ-ments followed with algorithms addingsome of the intelligence required toreach our goal for 2010.

Link:ISTAG; Scenarios for Ambient Intelligence in 2010; Final Report, Feb 2001, EC 2001: http://www.cordis.lu/ist/istag.htm

Please contact: Jari Ahola — VTT Information TechnologyE-mail: [email protected]

Defined by the EC Information Scociety Technologies Advisory Group in a vision of theInformation Society, Ambient Intelligence emphasises on greater user-friendliness,more efficient services support, user-empowerment, and support for humaninteractions. In this vision, people will be surrounded by intelligent and intuitiveinterfaces embedded in everyday objects around us and an environment recognisingand responding to the presence of individuals in an invisible way by year 2010.

Ambient Intelligenceby Jari Ahola

Articles in this section9 Ambient Intelligence:

a UK Perspectiveby Ronan Sleep

10 Ambient Intelligence in the Contextof Universal Accessby Constantine Stephanidis

12 Ambient Intelligence within a Home Environment by Mária Bieliková and Tibor Krajcovic

14 A Flood of Intelligence - the Living Room Project by Jukka Vanhala

16 Ambient Intelligence and HomeNetworking for WellnessManagement and Home Automationby Ilkka Korhonen

17 Wired and Smart: from the Fridge to the Bathtubby Franz Miller

18 Enabling Ambient IntelligenceResearch with SoapBox Platform by Esa Tuulari

20 Designing for Ludic Aspects of Everyday Lifeby Bill Gaver

21 Ubiquitous Service Environmentsby Carl Gustaf Jansson and Peter Lönnqvist

23 Offering a Consumer-OrientedAmbient Intelligence Environmentby Valerie Issarny

24 Global Ambient Intelligence: The Hostess Conceptby András Lörincz

25 Ubiquitous ComputingInfrastructures by Friedemann Mattern

26 Ubiquitous Computing and Embedded Operating Systems Designby Michel Banâtre

28 Ambient Interfaces for Distributed Work Groups by Tom Gross

29 GeoNotes: Social Enhancement of Physical Space by Per Persson and Fredrik Espinoza

30 Contextualisation in NomadicComputingby Markus Eisenhauer and Roland Klemke

31 Scene Interpretation for Video Communicationby Monique Thonnat

32 Mining Multimedia Documents by Visual Content by Nozha Boujemaa and MarinFerecatu

34 How can I be sure that my DVDplayer understands my TV? by Wan Fokkink, Izak vanLangevelde, Bas Luttik and Yaroslav Usenko

36 Fault Diagnosis in Wireless SensorNetworksby Stefano Chessa and Paolo Santi

SPECIAL THEME: AMBIENT INTELLIGENCE



The ISTAG meetings culminated in anIPTS (Institute for ProspectiveTechnology Studies) document whichdevelops a number of scenarios illus-trating Ambient Intelligence: seehttp://www.cordis.lu/ist/istag.htm. Thesescenarios illustrate the breadth possibleapplications. Some are relatively nearterm and are based on applying intelli-gent agent technology via an extensionof mobile communications systems, forexample to support a busy executiveperpetually travelling from meeting tomeeting as in the ‘Maria, Road Warrior’scenario. In contrast the ‘Ambient forSocial learning’ is much more techno-logically demanding, assuming forexample holographic projection and theability to manipulate an individual’ssound field within a room.

As the scenarios meetings progressed,participants became increasingly con-cerned with possible impacts of the tech-nology on both individuals and societies.At one extreme, access to the ‘ambientintelligence landscape’ could berestricted to those engaged in highpowered decision making involvingmuch travel and many meetings. At theother extreme, the technology could beused in a much more diluted fashion tomake small, but ubiquitous improve-ments in the lives of most citizens. By nomeans all uses of an ambient intelligencewhich are possible are desirable: issuesof privacy, choice, and trust of human,organisational and artificial actorsbecome paramount: an ill-consideredrush into exploiting the technologybefore it is both desired and trusted is alltoo likely to foster the sort of reactionfaced by the promoters of GM foods.

Although Ambient Intelligence covers alarge range of concerns, both human andtechnical, there are some technologies

which might be excluded. They are char-acterised by Mark Weiser’s statementabout Ubiquitous computing:“Ubiquitous computing is roughly theopposite of virtual reality. Where virtualreality puts people inside a computer-generated world, ubiquitous computingforces the computer to live out here inthe world with people.”. Seen in thislight, Ambient Intelligence is the limit ofa process which introduces the tech-nology into people’s lives in such a waythat the introduction never feels like aconscious learning curve: no specialinterface is needed because human expe-rience is already a rich ‘Manual’ of waysof interfacing to changing systems andservices. Somehow, we need to createtechnology that leverages this powerfulhuman resource rather than trying to sup-press it by requiring humans to partici-pate in inflexible interaction protocols ofthe sort supported by current call centretechnology.

From this perspective, realising AmbientIntelligence need not wait until we areliterally surrounded by the technology: itcan begin by studying normal life, usingethnographic and other techniques, andthen exploring acceptable ways of usingthe technology to enhance everydayexperience. As people become increas-ingly empowered, a point is reachedwhere as with the original telephone andits recent mobile offspring it becomespossible to do things that were not pos-sible without the technology. Here at theUniversity of East Anglia in Norwich wehave a number of projects exploring thisgradualist route to Ambient Intelligence,eg ViSiCAST see http://www.visicast.sys.uea.ac.uk/Publications.html#Presentations.

More generally within the UK the termspervasive and ubiquitous computing aremore prevalent than Ambient

Intelligence, but the vision featuresstrongly in a number of recent reports onthe future of ICT in the UK. For example,an influential UK government reporthttp://les1.man.ac.uk/cric/dgrc_reports.htmenvisages the following developmentsduring the next 5 years:• UK-based firms taking leading

positions in the development ofmobile computing, and contributingto new global standards in the area

• A further improvement in theavailability of venture capital, and inexperimentation with new businessmodels

The notion of Ambient Intelligence was developed during a series of ISTAG (InformationSocieties Technology Advisory Group) and other meetings as a guiding vision to givean overall direction to Europe’s Information Societies Technology programme. AmbientIntelligence is essentially an elaboration of Mark Weiser’s vision of Ubiquitous - butCalm Computing which stresses the importance of social and human factors as well asdeveloping the base technologies on which aspects of the vision are founded.

Ambient Intelligence: a UK Perspectiveby Ronan Sleep

Towards unobtrusive technology. Here aprototype TESSA system for assisting thehard of hearing is being trialled at a PostOffice Counter. Later generations of suchtechnology should merge entirely into thebackground.




• Strong UK ‘content’ companiesemerging in fields such as E-education and E-health

• Cheap access combined withdemanding consumers for ICTservices which creates a populationwhich is also skilled and innovativein ICT innovation

• Government agencies leading theway in releasing the commercialvalue of under-utilised informationthrough the Internet

• Rapid growth of B2B and B2C e-commerce.

UK Basic and Applied research in ICT isalso being increasingly guided by thegeneral vision of Ambient Intelligence:indeed ‘Pervasive Computing’ is one offive key research challenges identified ina recent report on UK Computing

Research (see http://www.cs.ncl.ac.uk/people/cliff.jones/home.formal/UKCRCtoIR.pdf).

Other key challenges identified includeSystem Design, the InformationPowered Society, Devices and Theory,and many of the issues covered withinthese topics have a close match withresearch issues relevant to AmbientIntelligence.

A flagship UK research activity whichhas similar motivations to the AmbientIntelligence Vision (although withperhaps more of a Virtual Reality flavourthan Mark Weiser would have liked) isthe EPSRC funded EQUATORInterdisciplinary Research Challenge (seehttp://www.epsrc.ac.uk/documents/about_epsrc/corporate_publications/newsline_journal/newslin16/smarter.htm for a briefdescription). EQUATOR spans eightsites, including the Universities ofLancaster, Bristol, Nottingham,University College London, Sussex,Glasgow, Southampton and the RoyalCollege of Art, and will involve aca-demics from traditionally disparate disci-plines such as computing, sociology anddesign. To quote Tom Rodden (projectleader, now at Nottingham University):“Rather than leap on every small techno-logical step forward, the project will thinkabout how the world might be in five orten years time, and aim to design andbuild devices for that moment. We have

the opportunity to realise some of thethings which technology futurists talkabout, but to do it in a very reasoned way.The Equator IRC will investigate allaspects of the development of futuredevices: the fundamental computationalinfrastructure required, the theories andconcepts that underpin it, the productdesign that will most enhance the device’sfunction, and the likely ways in which itwill be used. We’re looking at discov-ering and meeting future needs: whatshould be developed and how it might fitwith the way people live now, and howthey might wish to live in the future.We’re looking to see what may or maynot be the future.” (The author partici-pated in several of the ISTAG meetingswhich developed the AmbientIntelligence Landscape vision).

Links: ISTAG: http://www.cordis.lu/ist/istag.htm

VISICAST: http://www.visicast.sys.uea.ac.uk/Publications.html#Presentations

DGRC Reports:http://les1.man.ac.uk/cric/dgrc_reports.htm

EQUATOR: http://www.epsrc.ac.uk/documents/about_epsrc/corporate_publications/newsline_journal/newslin16/smarter.htm

Please contact: Ronan Sleep — University Of East Anglia, Norwich, UKTel: +44 1 603 592 693E-mail: [email protected]

A signing avatar from the prototypeTESSA systems.

The on-going paradigm shift towards aknowledge-intensive InformationSociety has brought about radicalchanges in the way people work andinteract with each other and with infor-mation. Computer-mediated humanactivities undergo fundamental changes,and new ones appear continuously, asnovel, intelligent, distributed, and highly

interactive technological environmentsemerge, making available concurrentaccess to heterogeneous informationsources and interpersonal communica-tion. This dynamic evolution is charac-terized by several dimensions of diver-sity that become evident when consid-ering the broad range of user characteris-tics, the changing nature of human activ-

ities, the variety of contexts of use, theincreasing availability and diversifica-tion of information, knowledge sourcesand services, the proliferation of diversetechnological platforms, etc.

In this context, the ‘typical’ computeruser can no longer be identified: infor-mation artifacts are used by diverse user

This article discusses Ambient Intelligence under a Universal Access perspective, aswell as the Unified User Interface Development framework as a design and engi-neering methodology addressing, through a proactive and generic approach, thedimensions of diversity arising in the context of Ambient Intelligence.

Ambient Intelligence in the Context of Universal Accessby Constantine Stephanidis



groups, including people with differentcultural, educational, training andemployment background, novice andexperienced computer users, the veryyoung and the elderly, and people withdifferent types of disabilities. Existingcomputer-mediated human activitiesundergo fundamental changes, and awide variety of new ones appear, such asaccess to on-line information, e-commu-nication, digital libraries, e-business, on-line health services, e-learning, on-linecommunities, on-line public and admin-istrative services, e-democracy, tele-work and tele-presence, on-line enter-tainment, etc. Similarly, the context ofuse is changing. The ‘traditional’ use ofcomputers (ie, scientific use by the spe-cialist, business use for productivityenhancement) is increasingly being com-plemented by residential and nomadicuse, thus penetrating a wider range ofhuman activities in a broader variety ofenvironments, such as the school, thehome, the market place, and other civiland social contexts. Finally, technolog-ical proliferation contributes with anincreased range of systems or devices tofacilitate access to the community-widepool of information resources. Thesedevices include computers, standardtelephones, cellular telephones withbuilt-in displays, television sets, infor-mation kiosks, information appliances,and various other ‘network-attachable’devices.

The notion of Universal Access hasbecome critically important for ensuringsocial acceptability of the emergingInformation Society. Universal accessimplies the accessibility and usability ofInformation Society Technologies (IST)by anyone, anywhere, anytime. Its aim isto enable equitable access and active par-ticipation of potentially all citizens inexisting and emerging computer-medi-ated human activities. Therefore, it isimportant to develop universally acces-sible and usable technological land-scapes, capable of accommodating allusers in all potential contexts of use,independently of location, user’sprimary task, target machine, run-timeenvironment, or the current physicalconditions of the external environment.

Universal Access is directly related tothe emerging concept of Ambient

Intelligence, that calls for the develop-ment of multi-sensorial interfaces sup-ported by computing and networkingtechnologies present everywhere andembedded in everyday objects. In thiscontext, the technological environmentis seen as an all-encompassing com-puting platform, composed of multipledistributed processing and interactiveunits, and exhibiting various levels ofintelligence. Ambient Intelligence incor-porates properties of distributed interac-tivity (eg, multiple interactive devices,remote interaction capabilities), ubiqui-tous computing (the ‘invisible’ computerconcept, ie, user - environment directinteraction), and nomadic or mobilecomputing (eg, location awareness,interface migration). AmbientIntelligence has the potential to providethe user with a virtual space enablingflexible and natural communication withthe computing environment or with otherusers, providing input and perceivingfeedback by utilising proportionally allthe available senses and communicationchannels, while optimising human andsystem resources.

In the context of Universal Access,ambient intelligence is addressedthrough a proactive and genericapproach, which accounts for all dimen-sions of variation (ie, the abilities, skills,requirements and preferences of users,the characteristics of technological plat-forms, the relevant aspects of the contextof use), and supports intelligent interfacerun-time adaptation, ie, the capability ofautomatically adapting to individualusers’ characteristics and contexts of usethrough the realization of alternative pat-terns of interactive behaviour.

The Unified User Interface developmentframework (see [Stephanidis, C. (Ed.).(2001). User Interfaces for All -Concepts, Methods, and Tools. Mahwah,NJ: Lawrence Erlbaum Associates(ISBN 0-8058-2967-9, 760 pages)]) pro-vides a design and engineering method-ology supporting the development ofself-adapting interfaces exhibiting bothuser awareness (ie, the interface iscapable of user-adapted behaviour byautomatically selecting an interactionpatterns appropriate to the particularend-user), and usage context awareness(ie, the interface is capable of usage-

context adapted behaviour by automati-cally selecting interaction patternsappropriate to the particular physical andtechnological environment). A UnifiedUser Interface comprises a single(unified) interface specification definedby the following key properties: • it encompasses user- and context-

specific information• it contains alternative dialogue

artifacts in an implemented form,each appropriate for different user-and context-specific parameters

• it applies adaptation decisions,activates dialogue artifacts and iscapable of interaction monitoring.

Recent and ongoing applications of theUnified User Interface developmentframework include:• the AVANTI web browser, providing

web-based user-adapted interaction(see article in ERCIM News 41)

• the PALIO platform for anyone andanywhere access to community-oriented services, supporting locationawareness and diverse platforms suchas PCs, kiosks and mobile phones(see article in ERCIM News 46)

• the 2WEAR platform for migratoryinterfaces to wearable devices,supporting distributed dynamic I/Ocontrol (IST-2000-25286 ‘A Runtimefor Adaptive and ExtensibleWirelessWearables’).

Please contact:Constantine Stephanidis — FORTH-ICSTel: +30 81 391741E-mail: [email protected]



Rastislav Habala, Jaroslav Kuruc,Vladimir Marko, Dalibor Rak and AntonWeissensteiner are all undergraduatestudents at the Slovak University ofTechnology and worked towards avision of a household where appliancesserve a user according her/his currentneeds. The students designed and devel-oped a prototype of the system calledEUNICA, which is intended to delivervarious home-related services to theusers. It is sensitive and responsive to thepresence of people (the occupants or vis-itors of a household). The user in thehousehold is surrounded by a multitudeof interconnected appliances. Thenetwork of appliances is invisible.Appliances serve the user according toher/his customs and preferences. Thewhole system exhibits some kind ofintelligence, eg, it is able to recognizeeach individual in the household andadapt behaviour according their needs, itis able to recognize specific events (suchas time or movement of a user) and actupon various situations.

An important feature of EUNICA is pro-viding mobility: mobile access to appli-ances on the one side and possiblemobile connection of appliances on theother. The wireless technology, namelyBluetooth, which presents the primaryway of communication, is grantingEUNICA’s great features to supersedeconventional infrared remote controls.However, EUNICA is not limited to asingle set of communication methods ora fixed set of appliances. It uses the openconcept of loadable drivers to be readyfor future appliances and technologies.The system covers features allowingaccess to appliances from outside of auser’s household via phone lines or theInternet. EUNICA also provides strongappliance management functions that arekept hidden for the standard use. It pro-vides a secure way of appliance installa-tion.

The heart of the EUNICA is the controlunit. Any appliance can be connected tothe control unit using various types ofconnections (see Figure 1). Monitoringand controlling appliances are visualizedon simple mobile Java based user inter-face devices connected to the controlunit using the Bluetooth wireless com-munications technology. All kinds ofinexpensive, physically compactdevices, such as PDAs are allowed. Wecall these devices ‘eurecos’ (EUNICAremote controls). They display and allowbrowsing information received from the

control unit (represented in the EUNICAMarkup Language that is based onXML) and send user’s requests back tothe control unit.

To fit the system to a house of any sizethe access to mobile components of thesystem in a big house is proposedthrough a network of access points (real-ized by Bluetooth modules connected tothe control unit). The appliances them-selves are not part of EUNICA; howeverthey should include support forEUNICA.

From the user perspective, EUNICA is asystem, which enables them to monitorand control appliances in a household byeurecos. Several eurecos can exist in thehousehold. Any user can use any eureco,as each eureco is always adapted to thecurrent user. To make the eureco able toeasily and comfortably identify itscurrent user, the eureco prototype isequipped with a fingerprint scanner.

To simplify the monitoring and controlof appliances for various individuals andgroups (eg, children, adult, elder), the

controls have to adapt and customise.The adaptability considers the number ofcontrol elements, its layout and appear-ance. In addition, EUNICA monitorsuser’s actions and adapts to their prefer-ences (eg, EUNICA can detect the levelof light brightness the user uses mostoften and later automatically sets to theobserved level).

EUNICA provides sophisticated accessrights control. Parents can restrictwatching particular TV channel for onechild, or forbid switching on the washmachine for all children. EUNICA

EUNICA (Extensible Universal Control of Appliances) is a prototype of an intelligenthousehold system, developed by a team of five students from the Slovak University ofTechnology in Bratislava. With this system, the team won the fourth price at the annualIEEE Computer Society International Design Competition (CSIDC) World Finals.

Ambient Intelligence within a Home Environment by Mária Bieliková and Tibor Krajcovic

Figure 1: The overview of Eunica.



brings new view to control overappliances. It allows:• a common way of appliances

control (remote control of TV-sets, Hi-Fi)

• to monitor state of appliances(temperature on thermometer,remaining washing time,surveillance camera in thechildren’s room or electronicdoorman)

• to control appliances usingmenus (automatic programselections based on the amountand kind of clothes insertedusing the wash-machine menu)

• to program various embeddedappliances (heating system,alarm), and help in automaticdata acquisition (eg, fromvarious kinds energy meters).

Measured data is sent through thenetwork to the energy providerswithout a possibility of user inter-vention, so the user cannot abuse it.

The eurecos can also be used to accessInternet resources such as email, weatherforecast or electronic newspaper. Thesystem can support wearable computingdevices for tracking physiologicalparameters, locating the users, etc.

In the view of manufacturers EUNICArepresents a straightforward extensionfor appliances they produce. In order tosupport EUNICA they should extendexisting appliances with the Bluetooth oranother connection technology(EUNICA provides versatility of appli-ance connection - by serial or parallellink, by net, AC power lines, or wirelessBluetooth). Manufacturers shoulddevelop a driver that will be dynamicallyadded to the control unit. It will enrichthe appliance with required logic, whichwill perform the commands of thecontrol unit and send the state of theappliance to the control unit.

Drivers and appliances are not passivecomponents which only respond to auser requests. A wristlet on a child’shand can monitor her/his temperatureand, if it increases, the EUNICA auto-matically notifies the parents. Driversand appliances have to satisfy certainrequirements for installing and running.

These requirements ensure co-operationbetween drivers in the control unit, butalso security. Finally, manufacturershave to design control pages in EUNICAMarkup Language for their appliances.

The control unit is implemented usingSun Java2 JDK 1.3 Standard Edition,because of reducing the platform depen-dencies and taking advantages of exten-sibility including security architectureand dynamic loading of new classes intothe running code. These features stood inthe good stead when outfitting thesystem with runtime module addition.

The eureco is implemented in Sun Java2JDK 1.3 Micro Edition (J2ME). It con-sists of Connected Limited DeviceConfiguration 1.0.2 (CLDC) specifica-tion and reference implementation aswell as Kilobyte Virtual Machine(KVM) reference implementation forvarious platforms including MicrosoftWindows, which we used. CLDC andKVM implementations are contained inmany portable personal appliances suchas PDA. Target platform for KVM isdevices with low memory (512KB),battery operated and with limited com-munication capabilities, so it perfectlyfits the needs for remote control.

For testing and demonstration pur-poses two sample appliance proto-types were created: • radio as an example of appliance

with adaptable and adaptivesupport (as real hardwareappliance)

• heating as an example of highlydesired appliance accordingmarket research provided duringthe project, seehttp://eunica.host.sk/ (as an appliance simulated oncomputer).

EUNICA is designed consideringsecurity issues on several levels:security in communication betweenthe control unit and appliances,security in communication amongmodules inside the control unit andavoiding unconscious and con-scious misuse of EUNICA andappliances.

AcknowledgementWe would like to thank our undergrad-uate students Rastislav Habala, JaroslavKuruc, Vladimir Marko, Dalibor Rakand Anton Weissensteiner for theirexcellent project work and presentationat the CSIDC World Finals inWashington, DC, and to ALEF, s.r.o.and APIS, s.r.o. for providing us withfingerprint scanner hardware and soft-ware development kit. The workreported here was partially supported bySlovak Science Grant Agency, grant No.G1/7611/20 and by IEEE ComputerSociety via Computer Society DesignInternational Competition (CSIDC)2001.

Links:SUT CSIDC web pagehttp://www.dcs.elf.stuba.sk/csidcCS-IEEE CSIDC web pagehttp://www.computer.org/csidc

Please contact:Maria Bielikova — Slovak University of Technology / SRCIMTel: +421 2 602 91 473E-mail: [email protected]

Eureco design of radio control for children


14

Today we are able to add intelligenceinto an object with about 20 Euros. Weare interested in what happens when thecost is eventually closer to 20 cents. Theresearch is conducted at the Institute ofElectronics of the Tampere University ofTechnology in Finland by a researchgroup called Personal ElectronicsGroup, PEG. The project started insummer 1999 and is continuing.

Let’s get PhysicalThe concept of digital convergence isexpanding. We share the vision that theambient intelligence arises from theconvergence of three key technologies:Ubiquitous Computing, Ubiquitouscommunication, and Intelligent UserFriendly Interfaces. But there is also thefourth component, which is the real worldphysical objects.

The boundary between real world andthe information world will vanish.Traditionally things have existed on theirown sides. Food and tools clearly arematerial, books and email are mostlyinformation. The new technology willconnect the two worlds. A new car hasdozens of micro controllers and a com-puter and a mobile phone are primarilyinformation appliances. It becomes pos-sible to control the real world trough thenetwork and make use of informationcontents by manipulating real worldobjects.

In addition to information processing,communication, and contents theeveryday living environment gets adigital form. Eg imagine a tee shirt suit-able for children’s games. There wouldbe a computer with a game contentsdownloaded from the network andsensors measuring the actions of theplayer. The shirt communicates with theother players’ shirts. Clearly the tee shirtis a real physical object, which has beenaugmented with digital contents.

The everyday environment at the homeor at the office will still be a real physicalenvironment. The physical objects willget new digital content and functions,but they still have their traditional func-tions. How people will see the effect ofthe Moore’s law is not in the high-endcomputers doing spectacular things, butrather in the very small processorsembedded in every imaginable objectforming a very fine grain parallel multi-computer with a user interface spreadacross the whole environment.

the Living Room project studies thesmart home as a place where intelligenceis embedded in practically all physicalobjects ranging from kitchen appliancesto toothpicks. A home of the futuremight have thousands of active smartobjects. Today the cost of embedding

some simple sensors, a microcontroller,a short-range radio link, and a batteryinto an object is about 20 euros. The sizeof the device is about the size of amatchbox. In about ten years it is pos-sible to integrate all electronics onto asingle chip and the cost and the size willdrop dramatically. We reach an inter-esting point when the intelligence addsless than 20 cents to the cost of theproduct. Then it is possible to have smartthings costing about 2 euros total, a pricetag on a say toothbrush or on a cone ofice cream.

Natural and Control InterfacesA smart object has a natural user inter-face and a control user interface, whichcan be physically separated. All objectshave a physical appearance and func-tionality that represents its natural

The Living Room project studies the smart home environment. The project is unique inits approach of developing user interface paradigms and communication techniques forlarge collections of intelligent small objects.

A Flood of Intelligence - the Living Room Projectby Jukka Vanhala

The interior of theLiving Roomshowingspotlights, thefloor with pressuresensors, and thesmart flowerpot.

ERCIM News No. 47, October 2001



purpose. Tools have handles and arefrigerator has a door and shelves. Wecall this the natural user interface. Thedesign of it has traditionally been the jobof the industrial designers and themechanical engineers. Now they toohave to learn new skills, as the embeddedelectronics will give new functionality toold objects. The enhanced natural userinterfaces utilizes sensors to measure theuser action and the object reacts withactuators typical to each object.

As electronics is embedded into objects,they will get also information contentand ability to process information. Theinformation is in symbolic form andtherefore a smart object needs also aninterface capable to represent and inputsymbols. Without the communicationcapability each of the objects would needits own small control interface withprobably a few buttons and a couple oftext lines on an LCD-display. It is diffi-cult to design a good interface with sucha limited set of controls. It is also clearthat most of these interfaces would havedifferent logic and different appearance.Thus the user must learn to use dozens ofbadly designed interfaces.

When the ability to communicate withthe home network is an integral part ofthe devices the situation improves. Allcontrol interfaces can be detached fromthe actual device much in the samemanner as a TV has a remote controller.Here we run into another well-knownproblem. It is no better to have dozens ofloose remote controllers instead of thecontrol interfaces on each device. Thesolution is to have a universal remotecontroller, which can control all possibledevices. There are already such deviceson the market although they are prettylimited in their design. A future controldevice, the universal information appli-ance, UIA, should have a large screen, apointing device, speech input and a stan-dard method to describe the appearanceand functionality of the interface, whichcan be eg loaded from the web page ofthe appliance manufacturer. Designingsets of ‘skins’ to control interfaces forsmart objects could become a new valueadded service. Service provider could beindependent of appliance manufacturersand provide matching designs for alldevices. There will also be a need for dif-

ferent kinds of terminals, thus the inter-faces must scale to meet the abilities ofthe underlying hardware.

The Living RoomLiving Room is besides the name of theprojects also the name of the actual labo-ratory where the new systems can beevaluated. The 40 square meter labora-tory resembles a flat with a living roomand a bar kitchen. During the summer2002 a new laboratory will be openedwith a two room fully equipped flatincluding a full kitchen and a sauna.

At the moment the infrastructure is readycomprising of a short range RF-network,the design for the controller to beembedded into smart objects, the mastercontrol computer and the prototypesystem software with a control userinterface. There are also a number ofsmart objects including a flowerpot, adoor with locks, an array of spotlights,shades, refrigerator, pressure sensitivefloor, and even a smart wine bottle rack.

Communication is done wirelessly witha narrow band control network, whichhas been designed just for this purpose. Itis based on the Nordic VLSI transceiverchip, which operates on the license-free433 MHz IMS band. The throughput is10 kbits/s and the reliable range is over30 meters, which is enough for thenormal rooms at home. Of course it isnot possible to transmit images or speechover the control network. For thatpurpose there is the IEEE 802.11 wire-less LAN connecting the control com-puter and the user interface terminals,which are both PC-class computers.

Many legacy products such as TVs andaudio equipment use infra red transmis-sion for communication. Since thosedevices will be in use for at least anothertwenty years, there must be a way ofcontrolling them. An IR-receiver/trans-mitter is connected to the control com-puter via a serial link so that the sameunified user interface as is used tocontrol devices via the RF-network canbe used.

The controller to be embedded into smartobjects consists of an Atmel AVR-seriesmicrocontroller, sensor interface elec-tronics, the RF-link, and a battery.

Software is easily customizable for dif-ferent requirements. The size of the con-troller is about 2 times 4 times 6 centime-ters.

At the moment we are using a normaloffice-PC with the Windows-98 oper-ating system as the control computer.Eventually the controller will be a homegateway type standalone computer butthe prototyping work is much easier withthe good development tools available forWindows-98. The user control interfaceis implemented on an industrial PC witha touch screen. The computer isembedded into the surface of a coffeetable. Similar control terminals can beinstalled on walls and other surfaces.

Conclusions and Future DevelopmentTraditional user interface research hasconcentrated on the design of the controlinterface. When all objects in a smartspace can be used to collect and dis-tribute information the collection of thenatural user interfaces will dominate thebehavior. We call the collection of thenatural user interfaces a collective userinterface. There is not much empiricalknowledge or experience on this becauseit would require a full implementation ofa smart home, which has not yet beenbuilt anywhere. The Living Room–project is working towards imple-menting a prototype environment of acollective user interface. Our approach isvery practical. We are interested both inthe overall architectural design and in theimplementation details. Real worldexperiments and prototype installationsare used to test all designs.

Link: http://www.ele.tut.fi/research/personalelectronics/

Please contact:Jukka Vanhala — Tampere University of Technology, Institute of Electronics, FinlandTel: +358 3 365 3383E-mail: [email protected]



Self-monitoring is an efficient tool inpersonal health management. Self-monitoring of eg weight means dailymeasurements of weight, storage of theresults, and using the recorded weighthistory to follow up possible changes inweight. This would allow the user to learnabout the behavioural factors affecting theweight, and thereby to manage his/herweight. This is called behavioural feedbackmodel (see ERCIM News No 46, pp. 60-61;http://www.ercim.org/publication/Ercim_News/enw46/parkaa.html).Efficient self-monitoring requires dailyactions (weighing, stroring of the result,browsing the feedback), which shouldrequire minimal efforts by the user for itto be successfully applied in the long-term.Ideally, the measurements should be ableto be made wherever and wheneverconvenient (mobility), the storage of theresults should occur automatically(wireless communications, automatic useridentification, existence of ambientdatabase services), and the user should beable to get the personalised feedback rightwhen needed, independent of time, place

or user interface s/he has at hand(persolisation, adaptation of userinterfaces, etc.). This implies applicationof ambient intelligence (or, homenetworking, ubiquitous computing).

The first part of the Wireless WellnessMonitor project (WWM I) was carried outbetween 1.8.1999 and 31.12.2000 (seeERCIM News No 46, pp. 60-61;http://www.ercim.org/publication/Ercim_News/enw46/parkaa.html). In that part abehavioural feedback model for healthmanagement was presented, and a lab-prototype of ambient intelligence basedweight management was developed anddemonstrated. In the current continuationproject (WWM II) we extent the conceptto include home automation and otherfunctionalities. This is done in order todemonstrate that different ambientintelligence based services at home shouldbe running as different services on thesame infrastructure, including a homenetwork, a home gateway, and differentuser interfaces such as personal digitalassistant, mobile phone, or TV. Especially

the project studies how simple devices(such as a personal scale or coffee maker)may be networked at home and how smarthealth management and home automationapplications may actually be implementedin the home networking infrastructure.The project will produce two kinds ofresults: 1) a home networking and ambientintelligence concept, and 2) ademonstration system, in which theconcept is implemented, including theinfrastructure and wellness managementand home automation applications.

Actual ambient intelligence enabled homerequires two levels of implementation.The first level provides the infrastructureand makes the technology ambient,communicative. The second level buildsthe actual smart applications; theintelligence. One goal of the project is tostudy and demonstrate this two-layeredapplication structure in case of healthmanagement and home automation. Aschematic illustration of the physicalnetwork to be built in the project isillustrated in the figure. The maincharacteristics of the demonstrationsystem are: 1) the intelligence is mostlylocated at a home server (OSGi server);2) from the service point of view, thehome network is IP-based; however,simple non-IP devices are represented bya proxy to the network thus allowingheterogenous physical connections to thenetwork. The demonstration system is tobe completed by the end of 2001, and usertrials are to be carried out early 2002.Partners: VTT Information Technology,VTT Electronics, Nokia Research Center,IST Oy, Celotron Oy and the NationalTechnology Agency of Finland (TEKES).

Link:http://www.vtt.fi/tte/samba/projects/wwm/

Contact: Ilkka Korhonen — VTT Information TechnologyTel: +358 3 316 3352E-mail: [email protected]

Different ambient intelligence based services at home using the same infrastructure arewithin the scope of the the Wireless Wellness Monitor project carried out at VTTInformation Technology.

Ambient Intelligence and Home Networking for Wellness Management and Home Automationby Ilkka Korhonen

Physical network of the Wireless Wellness Monitor II demonstration system. Thearchitecture is based on home server (OSGi server) and IP-based networking. Simplenon-IP devices, such as health monitors or coffee machine, join the network physicallyby eg RS232 and are represented as IP-devices by a device proxy.



On the Internet, you can even take a lookat the far side of the moon whenever youlike, but you can’t look to see whetheryou’ve left the hot plate switched on inthe kitchen at home. The Internet andmobile communications have penetratedevery last remote spot of the globe, butstrangely enough, the places closest toour daily lives, our homes, are still cutoff from the communication networks.Household appliances and domesticsystems are isolated from one another,silently dumb. Wouldn’t it be a goodidea to enable all of these appliances tocommunicate, and link them together inan ‘intelligent house’? If we set up anetwork to connect heating systems,security alarms, TVs, fridges, ther-mostats, movement sensors, lighting andwindow shutters, we would be able tocreate completely new functions andsave money too. The wired, partiallyautomated and remotely controlledhome, the ‘smart house’, is one of thebiggest innovation projects of thisdecade.

The aim of smart-house technology is tosave energy, increase security andcomfort, and to provide the basis for arange of new services. Researchers allover the world are working on the visionof the ‘intelligent house’ – experimentalhouses are already in operation in theUSA, Japan, Holland, Sweden andSwitzerland. But a great number of ques-tions still remain open: Which data trans-mission standards are likely to be mostwidely used? How can we make the bestuse of what is technically feasible? Andabove all: What do people want in theirhomes?

Now Germany, too, can join the ranks ofthe innovators, with its own large-scaleexperimental project. The InnovationCenter for the Intelligent House, or‘inHaus’ for short, opened in April inDuisburg. In a project initially scheduled

to run for five years, this otherwise ordi-nary semi-detached house will be used todevelop, test and demonstrate many ofthe ways in which modern progress ininformation technology and communica-tions can be integrated into our everydayprivate and working lives. Two featuresmake inHaus stand out from othersimilar projects: It is both an experi-mental laboratory for living and a work-shop for innovation, and will thereforecontinue to change in the light of newdevelopments. And it is an open system,ready to accommodate all suppliers andconcepts. This innovation center forfuture forms of living and working in a‘smart environment’ is being supported

by a broadly based consortium of firms,headed by the Fraunhofer Institute forMicroelectronic Circuits and SystemsIMS in Duisburg, who have all chosen tocoordinate their components so that theywork together.

The beginnings of inHaus date back fiveyears, to the time when the first conceptsfor data networks in domestic buildingswere formed. So far, such networks havenot made the breakthrough to the massmarket, because there were no standardsor integration concepts. The existingsolutions tend to be isolated projects.Over the next five years, the Duisburg

There’s a smart wired house in Duisburg; even the car and the gardenare integrated in the network. The ‘Innovation Center for the IntelligentHouse’ will be used by researchers and industrial firms as a workshopand testing ground for future forms of living and working.

Wired and Smart: from the Fridge to the Bathtubby Franz Miller

The residential half of inHaus is fully furnished and equipped, enabling the presentstate of the art to be tested. © Fraunhofer IMS / Axel Struwe.

Controlling inHaus from the car.© Fraunhofer IMS / Axel Struwe;Volkswagen AG.

The inHaus complex integrates all of thefunctions of an experimental, networkeddwelling house and a workshop of thefuture. Practically all of the installationsare hidden behind the walls and tiles© Fraunhofer IMS / Axel Struwe.



innovation center hopes to pave the wayto series production.

The inHaus complex is housed in a pairof connected buildings. One is anauthentic home, with the laboratory forliving and a home office, supplementedby a car (the connect-Passat) and an intel-ligent garden. The other is a research anddevelopment workshop, containing abathroom laboratory, a kitchen labora-tory, the future craftsman’s workshop,the central building utilities room and aplanning and advice office, which alsohandles teleservices.

Klaus Scherer of the IMS, initiator of theproject and now project manager forstrategy and marketing, defines theobjectives as follows: “On the one handwe want to find out how various appli-ances, components and infra-structures,which often operate on the basis ofextremely different standards applicablein different professional fields, can bemade to work together in a singlesystem, in a useful and efficient way. Inother words: find ways for technologiesto understand each other better. And onthe other hand we want to find ways fortechnology and people to understandeach other better. The aim is to providepeople with genuine support in theirliving environment, in their world ofwork, in communications and in

mobility. This means, for instance, thatthe technology must as simple as pos-sible to operate.”

The operators of inHaus are testingseveral networks simultaneously, bothcabled and wireless. The system integra-tion solutions are largely based onInternet technology. The IMS hasembedded this technology in the appli-ances and system components(‘embedded Internet’). The multimediahome cabling system HomeWay linksTVs, telephones and personal com-puters, and provides any required servicein any room of the house. Hometronicand EIB networks use wireless and cableconnections to control all electrical ele-ments, including heating – fully auto-matically or by manual keypads, viaremote control or telephone and Internet.To allow the Internet messages to leavethe house, the IMS and its partners havedeveloped a Residential Gateway, as thecentral element linking the external andinternal data networks. It works bothways, so inHaus residents can connectwith the appliances in their home whilethey are at work – to check their status orissue remote-control commands.Another major element, providing linkswith services in the outside world, is‘Smart Home’, a novel Internet serviceplatform that is connected to theHometronic wireless network in the

inHaus complex. Special emphasis hasbeen placed on new functions for savingenergy: Sensors in the rooms measurethe temperature, humidity and airquality, and automatically open andclose the motor-driven windows. If athunderstorm approaches, the houseautomatically makes sure that everythingis weatherproof. When the occupantsleave the house, the heating switchesitself down to economy level.

“In ten years or so,” predicts KlausScherer, “the intelligent house will havebecome a normal part of everybody’slife.” Already today, it is possible tocatch a glimpse of the way of life andworking to come.

The inHaus partners are: Ackermann,BurgWächter/Secu, Deutsche Telekom/T-Systems, Fraunhofer IMS, Geberit,Grothe, Henkel, Honeywell, Kaldewei,Liebherr, Merten, Miele, Siedle, Sony,Stadtwerke Duisburg, Viessmann,Volkswagen, Winkhaus and ca. 60 com-ponent suppliers and sponsors

Links: http://www.inhaus-nrw.dehttp://www.ims.fhg.de/englisch/index.htm

Please contact: Klaus Scherer — Fraunhofer Institute for Microelectronic Circuits and SystemsTel: +49 203 3783 2 11E-mail: [email protected]

SoapBox is put together from commer-cial components. It possesses small size,low power consumption, RF-communi-cation in unlicenced radio-band and a setof in-built sensors. It also has IO capabil-ities to connect to user devices. As userdevices we use mainly PDA’s whichnowadays offer decent support for appli-cation development. However, using aPC or laptop is also possible.

By installing SoapBoxes in everydayobjects we really can have computerseverywhere. With the wireless commu-nication link all these computers are con-tinuously connected to the user device,enhancing user-real world communica-tion to entirely new level compared tosituation with no such system or systemsbased solely on existing commercialdevices (see Figure 2).

One approach for ambient intelligence research is to study what happens as sensing,computing and wireless communication capabilities are added to everyday objects. AtVTT Electronics scientists have designed and implemented SoapBox, Sensing,Operating and Activating Peripheral Box, which makes this approach possible.

Enabling Ambient Intelligence Research with SoapBox Platformby Esa Tuulari

Figure 1. SoapBox architecture consistingof one central Box attached to the userdevice (in the back of the Palm, notshown) and several remote Boxeslocating in the environment.



From the devices point-of-view also theuser is part of the environment. In theSoapBox architecture we can easilyinstrument the user with SoapBoxes. Forexample measuring and recognicing useractivity and user behaviour is mucheasier if the sensors are distributed in thelimbs than concentrated in the belt orback, as is the situation with many wear-able computers.

Application ExamplesIn general it is very easy to invent andimplement new applications based onSoapBoxes. For example sensing ifdoors are open or closed, if someoneschair has moved lately, or if there are lotof traffic in the corridor offer a basic setof applications that can be implementedwith SoapBoxes.

By installing a SoapBox in the ceiling ofa mailbox we can use the proximitysensor for detecting letters (or anythingsimilar) inserted in the mailbox. TheSoapBox then transmits an RF-messageindicating “new mail available”. If theowner of the mailbox is nearby with hisPDA and central SoapBox attached to it,the message is received and the user isinformed about new mail. In a factory ora store the possibility to monitor theusage of for example fork-lifts couldprovide valuable information. With theSoapBox this can be done withoutinstalling any tags or cameras in thebuilding. Instrumenting the movingobject is enough to start with.

TechnologyShort range wireless communication isone of the key features of the SoapBox.The challenge has been to combine lowpower consumption and such perfor-mance parameters as communicationrange, bit rate and message delay.

Luckily in many applications high bitrate is not needed since only short mes-sages are used and time between mes-sages is rather long. SoapBox electronicsconsists of two printed circuit boards(see Figure 3), one containing thesensors and the other containing all theother electronics. This makes it easy todevelop further versions in a modularfashion.

SoapBox software has been imple-mented mainly in C language. A fewlines of assembly language has beenused mainly in timing critical parts. Allthe application software can be written inC, utilizing the API offering services forRF- ans RS-communication as well asfor reading sensors. A single channel868 MHz radio is used for two way, halfduplex wireless communication. A smallhelical antenna fits inside the SoapBoxencapsulation. Maximum data rate is 10kbps and maximum transmit power is 1mW. A real-time clock and calendarcircuit is also included. It can be utilizedfor example for timing control functionsand for recording timestamps fordetected events. The sensor board ofSoapBox ver. 1.0 includes four differentkinds of sensors:• Three axis acceleration sensors. These

sensors can be used to measureacceleration or tilt of the device

• Illumination sensor. The intensity ofvisible light can be measured. Thedynamic range of the sensor enablesboth indoor and outdoor measurements

• Electronic compass can be used forsensing the direction of the Box andalso for detecting magnetic objects

• Optical proximity sensor formeasuring relative distances. Thissensor is based on sending IR pulsesand measuring the intensity ofreflected light.

AcknowledgementThis work has been partially conductedin European joint research project‘Beyond the GUI’ under ITEA clusterproject of the EUREKA network, andfinancially supported by TEKES(Technology Development CentreFinland), to which organisations theauthor wish to express his gratitude.

Link:http://www.ele.vtt.fi/projects/iie/

Please contact:Esa Tuulari — VTT ElectronicsTel: +358 8 551 2334E-mail: [email protected]

Figure 2: Enhancing user - real world interaction. Left: Original situation. Centre: Cellular phone alone doesn’t help much. Right: Adding SoapBoxes in both ends solves the problem.

Figure 3: Picture of SoapBox boards,illustrating the size and shape of theelectronics.



If technology is to become ‘ambient’ inthe sense of surrounding us throughoutour lives, the question is not only whatform it should take – web pages, ubiqui-tous computing, tangible media, infor-mation appliances, etc. – but what rolesit might play in our lives. What is tech-nology good for? How can it support ourvalues, individually, socially, and cultur-ally?

There is a danger that as technologymoves from the office into our homes, itwill bring along with it workplace valuessuch as efficiency and productivity at theexpense of other possibilities. People donot just pursue tasks and solve problems,they also explore, wonder, love,worship, and waste time. These activi-ties, captured by the notion of ‘HomoLudens,’ or people defined as playfulcreatures, are meaningful and valuable,but difficult to handle from traditionalperspectives on Human ComputerInteraction. Thus we are developingdesign-driven techniques to explorepeople’s lives, create evocative designproposals, and develop prototypes ofnew systems that focus on the emotionaland sociocultural effects.

What sort of technologies might be appro-priate for Homo Ludens? One explorationtook the form of a palette of design pro-posals called Alternatives. Over a periodof about three months, we conceptualisedabout 20 products in the domains ofhome, street, intimacy, and wonder (seeFigure 1). Presenting each using collagesthat suggested but did not specify theirforms, and short descriptions that contex-tualised them and explained how theymight be experienced, these proposalswere meant to spur the imagination andopen new spaces for design. The DataLamp, for instance, is about the size of afloor lamp and allows images to be dis-played on its façade or released to paintsurfaces in a room. Dawn Chorus is anartificially intelligent bird feeder that

trains local birds one’s favourite songs.The Intimate View camera allows sepa-rated lovers to transmit close-up picturesof their surroundings for moments ofintense shared focus, while the Telegotchiis a virtual creature with no controls,allowing people to practice their psychicpowers as they enter into a relationshipwith it. These proposals are what TonyDunne has termed ‘value fictions,’exploring unusual values using plausibletechnologies in order to question therestricted scope currently assumed.In

order to develop possibilities for ludictechnologies further, we are currentlyengaged in a long-term research pro-gramme called Equator in collaborationwith eight UK academic institutions. TheEquator Project is focusing on mergingthe virtual with the real in situations suchas performance, play, museums, andcities. Our group is focusing on technolo-gies for the home, seeking to explore thevalues people have in their lives awayfrom work. To do this, we are expandingthe Cultural Probes method originallydeveloped for another project. Havingrecruited about 20 households throughadvertisements in the popular press andon shop windows, we have distributed‘Domestic Probe’ packages, with avariety of rovocative tasks to which theyare asked to respond (see Figure 2). Thesematerials include disposable cameraswith requests for pictures (eg ‘the mostuncomfortable place in your home,’ ‘theview from your kitchen window,’ ‘some-thing red’), a device for recording a vivid

Researchers at Royal College of Art, UK, are developing design-driven techniques toexplore people’s lives, create evocative design proposals, and develop prototypes ofnew systems that focus on emotional and sociocultural effects.

Designing for Ludic Aspects of Everyday Lifeby Bill Gaver

Figure 2: Domestic Probe package.

Figure 1: Alternatives proposals: Data Lamp, Dawn Chorus, Telegotchi and IntimateView camera.



dream when they awake, a glass for lis-tening to interesting sounds around theirhome, and a collection of bizarre newsarticles for them to annotate. The Probesare, in some ways, like the projective testsused by psychoanalysts: suggestive butambiguous, they elicit revealing frag-ments from participants which inform andinspire our designs.

With most of the Probe returns in, we arebeginning to develop new product con-cepts based on the replies. For instance,one woman revealed rigid notions of aproperly ordered house, shows a stronginterest in horoscopes and fortune-telling,and seeks ‘positive energy’ in her life. Forher, we are developing a system of dis-

tributed sensors so that the state of herhome can be translated into propheciesand prognoses. One man enjoys gazing athis tank of tropical fish and identifies acertain chair as the most uncomfortablespot in his home, so we are thinking abouttransforming the chair into a controlcentre from which he can guide a smalltelerobotic camera to watch his fish fromwithin the tank. Finally, a musician wholives in a penthouse flat may be offered asystem in which sounds transmitted to hishome from wireless microphones aresampled and ordered rhythmically toallow him to make music from the noisesin his environment. These proposals are intheir early stages, and no doubt willdevelop and change before implementa-

tion, but they capture some of the possi-bilities we see for technology aimed atsupporting people’s domestic lives. Theymay seem whimsical, but it would be amistake to dismiss them on that ground:for unless we start to respect the full rangeof values that make us human, the tech-nologies we build are likely to be dull anduninteresting at best, and de-humanisingat worst.

Link: http://www.crd.rca.ac.uk

Please contact: Bill Gaver — Royal College of Art, UKTel: +44 20 7590 4296E-mail: [email protected]

Service Environments are dynamic con-figurations of services available in phys-ical spaces for both collaborative andindividual use. An important objective isto achieve service environments that areopen, transparent, adaptive and degradegracefully with respect to capabilities ofthe interactive devices available in theroom. The term ubiquitous refers to theobjective that device technology prefer-ably should be designed in such a waythat it melts into the periphery when notneeded (ambient or calm technology).

Inhibiting factors for the realisation ofubiquitous service environments aredevice technologies of today charac-terised by heavy, bulky, focusdemanding devices, weak integrationbetween IT- non IT devices, strong divi-sion between private and public devices,weak local communication amongdevices, too many wires for energy andcommunication and static individualbehaviours and configurations ofdevices.

The situation of today can be remediedin two ways:• Establishing a richer set of channels

for interaction, by disassemblinginteractive devices (order of 10 ->100)accommodating non-conventionalappearance of interactive devices,mix the use of private and publicdevices, the use of mobile andstationary devices as well as the useof several modalities (sound, visualand tactile modalities)

• Enforcing synchronised behaviour ofservices and devices, by configuringdevices in a wireless, dynamic andad-hoc fashion, creating adequatemechanisms for context sensitivityboth for physical and organisationalcontexts and finally allowing forcollaboration and communicationbetween services.

The FUSE group focuses on the fol-lowing issues:• design of software architectures for

ubiquitous service environments

• development and test of prototypes ofinteractive services that supportdistributed as well as local work

• design of ubiquitous hardwareenvironments with the two foci above,to diversify and to synchronise

• HMI issues such as design ofinterfaces for small devices, tangibleinterfaces, multi-modal transparency,adaptability as well as properties ofconscious/non conscious interactions

• AI issues like negotiation mechanismsfor intelligent software agents forsynchronisation of services andimplementation of autonomous andproactive services in a form called‘active documents’

• Software Security and privacy issuesfocussed on client authentication andaccess control in dynamic and ad-hocenvironment

• systems complexity issues, likeproblems raised by scaling up in termsof number of devices, modalities andservices as well as emergentbehaviours in the system.

The FUSE (Future Ubiquitous Service Environments) group focuses on transparentlyaccessible configurations of services made available on assemblies of personal / public/ mobile / stationary devices that melt into the periphery. FUSE is affiliated with KTH inStockholm and funded by both national and EU sources, eg the IST FET DisappearingComputer project FEEL (2001-2003).

Ubiquitous Service Environments by Carl Gustaf Jansson and Peter Lönnqvist



Current FUSE projects include:• FEEL, an EU IST FET Disappearing

Computer Program project, aiming atmechanisms that ensures lessintrusions on collaborative work fromparallel electronic individual anddistributed work. The project attacksthe problem by employing a numberof the above mentioned principles.The work involves both prototypedesign and simulations.

• I-space, funded by the SwedishWallenberg Foundation, aiming atinteractive spaces where the‘technology level’ can be easilyadjusted and where specific ‘workingmodes’ can be easily restored. In acollaborative project with CS dept. atStanford University, project basedcourses at both locations will utiliseco-designed interactive spaces. Asubstantial grant has also beenreceived for establishing aninteractive space with ambientcharacteristics in the educationalfacilities at KTH in Kista.

• MEETPCC, a project funded by theSwedish Foundation for strategicresearch (SSF. The project focuses onactive documents that supports workprocesses and which are context

sensitive and able to detect specificevents as well as these activedocuments strategies for materialisingthemselves on the technologyavailable in the space where the eventoccurs. This project involves bothaspects of human machine interaction,computer security and artificialintelligence (see figure).

• Agent Based Technology, a projectfunded by the Swedish fundingagencies NUTEK/VINNOVA,focussing on negotiation mechanismsfor software agents. Negotiationamong software agents is used as theprimary mechanism for synchroni-sation of services. Both rationalistic,market-based and social paradigmsfor negotiation have been utilised.

Future extensions of the work as outlinedin current project proposals include:• handling mobile robots as active

participants in the interactive spaces• handling seamless and graceful

degradation of functionality ofservices when moving betweeninteractive spaces with differentcharacteristics

• exploring social paradigms fornegotiation among multiple services.

FUSE is a research group at the Dept. forComputer and Systems Sciences at KTHin Stockholm, Sweden. FUSE is alsoaffiliated with the new centre forWireless Systems at KTH. FUSE co-operates directly with SICS and theUniversity of Southampton within theFEEL project and more loosely withERCIM members such as GMD andINRIA through related projects withinthe Disappearing Computer Initiative.

Link:http://www.dsv.su.se/fuse/

Please contact:Carl Gustaf Jansson — Royal Institute of Technology (KTH), Sweden. Tel: +46 8 16 1605 E-mail: [email protected]

Active documents supporting teamwork.

1. A person enters the meeting room carrying her laptop and an identification button. A document is waiting somewhere for events (like meetings) to occur.

2. The person announces her presence by pushing her identification button into a receptor. The document will now detect that two persons, members of the same project, are in the same room and therefore it assumes that there is a meeting going on.

3. The document moves itself to the room and tries to find a public display where it will present itself.

AN ACTIVE DOCUMENT is a daemon supervising all activities in a work process. As long as the active document's conditions for appearing are not met, it is in an idle mode.

IN THE SIMPLEST case the active document tries, through different means, to detect who is present in a room.

WHEN THE GIVEN conditions are met, the active document can appear using the most appropriate modality.



The objective of the OZONE project is todevelop novel concepts, techniques andtools to provide invisible computing forthe domestic and nomadic personal useof information technology. The devel-oped concepts aim at improving theacceptability and usability for theaverage customer. One of the importantconcepts is the application of advancedtechnologies to support the user centricretrieval and consumption of informationcompared to the, current practice, com-puter centric approach. This requiresspecial emphasis on natural interfacesthat put the user in the foreground, andthe system in the background. Securityand privacy are prerequisites for con-sumer acceptance of these systems andare covered in the project’s objectives. Afinal objective deals with the provision ofa strong technology base, enabling pow-erful, but energy-efficient, computing.

The OZONE project targets the provi-sioning of a generic framework that isnot tailored to a specific environment butthat is rather able to integrate new tech-nologies as they appear. The frameworkarchitecture is based on a layeredapproach ranging from ‘serviceenabling’ functionality over the ‘soft-ware environment’ towards the under-lying ‘platform architecture’. An incre-mental approach of specification, designand prototype implementation will beapplied to arrive at enabling technolo-gies, providing a generic invisible com-puting framework driven by demon-strator scenarios from the extendedhome environment area.

The main focus of INRIA in OZONE ison the service enabling and software envi-ronment layers. INRIA also contributes tosome specific aspects of the platformarchitecture layer, relating to silicon com-pilation technology and resource manage-ment. INRIA will further develop a

demonstrator of a new transportationsystem based on fully automated vehicles.The OZONE research will concern thehuman interface with the transportationsystem: reservation and call of vehicles,on-board control and information, cityinformation, etc. The information supportwill be based on fixed terminals (home oroffice or city) as well as portable and on-board devices.

The service enabling layer comprisesenabling technologies for seamlessservice discovery based on the user’slocation and context, and for multimodaland secure access to available and person-alized services. Core of the work willfocus on the design and implementationof a linguistic knowledge managementsupport environment to supply the situa-tion sensitive information to the servicesoperating in the framework, so that theycan adapt to the context in a dynamic way.Next to this functionality, functionalitieswill be developed to support the applica-tion of multimodal user interfaces fromthe services as well as measures to protectthe privacy of the user. For the multi-modal interfaces, the combination ofspeech input with pointing devices willbe the starting point; the application ofvision will be investigated as an exten-sion to this idea.

The software environment layer providesthe necessary support for achievingservice delivery, independent of connec-tivity, device in use and power supplymeans. Component-oriented softwaredevelopment methodology, and currentlyavailable tooling to help the developmentand deployment of distributed softwaresystems will be taken as a starting pointfor the design of the software environ-ment. However, these technologies arenot sufficient to seamlessly offer servicesto users from the existing services that areavailable worldwide. Any requested

service should be deployed dynamicallyin a way that guarantees both the func-tional and the non-functional propertiesthat are expected from the service by theend-user. The approach undertaken tomeet the aforementioned requirementslies in the combination of two technolo-gies:• mobile agent technology will be

exploited for the specification of thecomplex services to be deployeddynamically with respect to both thebase services to be used and the userenvironment

• component-based technology will beexploited for the integration ofexisting services, hence enabling theiruse by mobile agents.

The OZONE IST project will start inNovember 2001. The project will firstinvestigate the specification of the genericOZONE framework, based on require-ments elicitation and extensive inventoryof the relevant technology domains. Theproject will then elaborate the genericframework enabling consumer-orientedambient intelligence applications, whichwill be assessed using applications fromthe extended home environment area. TheINRIA research teams contributing toOZONE are: Langue & Dialogue, Maiaand Parole at INRIA-Lorraine, Cosi andTemics at INRIA-Rennes, Sirac atINRIA-Rhone Alpes, and Imara,Moscova, and Solidor at INRIA-Rocquencourt, The other partners ofOZONE are: Philips (Project coordinator,The Netherlands), Interuniversity MicroElectronics Center (Belgium), Labo-ratoire d’Electronique de Philips (France),EPICTOID (The Netherlands), TechnicalUniversity Eindhoven (The Netherlands),and Thomson Multimedia (France).

Please contact:Valerie Issarny — INRIATel: +33 1 39 63 57 17E-mail: [email protected]

OZONE is an IST project that aims at specifying and implementing a generic frameworkthat will support the effective use and acceptance of ambient intelligence in theconsumer domain. Nine INRIA research teams are contributing to OZONE, focusingmainly on the service enabling and software environment layers of the OZONEframework.

Offering a Consumer-Oriented Ambient Intelligence Environmentby Valerie Issarny



The research has originated by noticingthe need for adaptive asynchronousinformation collection and controlmethods for distributed decision-making. Several experiences (see, eghttp://www.eds.com/case_studies/case_arkansas.shtml) point towards this direc-tion. Our model example is the complexcase of Parkinson disease patients(PDPs) using Deep Brain Stimulators(DBS). Simpler systems can be derivedas analogies of this case.

For PDPs, histories of individual patientsshow that sporadic experiences at dif-ferent hospitals need to be collected toimprove evaluation. Remote setting ofthe DBS, augmented reality with infor-mation collection and data managementtogether, could form global ambientintelligence (GAI) for full support of thepatients. Data management should meetconstraints derived from Good ClinicalPractice and could be seen as an exten-sion of drug administration procedures.GAI can serve the patient, can help thedoctor and can estimate risk. Adaptivetools for such (safety critical) applica-tions have become available by our latestachievements in reinforcement learning(RL). Optimization of decisions on con-tinuing data collection or executing anaction can be termed as optimization ofperception-action loops. Such chal-lenging optimizations concern partiallyobserved environments, being in thefocus of RL research. We are not awareof any similar global approach on thefield.

The project aims at defining the‘hostess’, ‘who’ is a goal-oriented agentwith possibly adaptive and distributedsubsystems. The hostess ‘serves’someone, eg a patient, a visitor, theowner of a home, or groups of those. Forthe interactions with people served, thehostess makes use of signals that are

important for efficient human communi-cation, including facial expressions,prosody, body talk, and behavioral pat-terns. In case of Parkinson diseasepatients global ambient intelligenceneeds to exploit:• the medical history of the patient

(available locally or over the internet)• the medical history of other patients• mobile communication outside of

safe environments (PDPs with DBSoften engage in activities outdoor)

• optimization of the neural prostheticdevice, the DBS, with option forremote control

• visual, acoustic and haptic augmentedreality tools

• distributed data editing• (provisional) access control• decision-making in this safety critical

situation.

A particular aspect of the project is thatdecision is achieved by combining theassessments of experts who may belocated at different sites and may havedifferent roles. The access to (medical)data may improve the decisions at theexpense of real time, communicationtime, and computational costs. Risk-sen-sitive decision-making needs to considerpipeline operations, both in communica-

tion and in computation tasks. Theproject is ‘just-in-time-research’. Wedevelop components under the assump-tion that Quality of Service will be avail-able for internet networks in about twoyears.

State-of-the-art techniques have beenadapted, developed, and are ready to useon the following fields: connectionistparallel algorithms, reinforcementlearning for partially observable environ-ments, parallel techniques for satisfia-bility problems, adaptive and robustcontrol methods, tracking and proba-bilistic inference methods, internet tech-nologies. The components are developedin a modular fashion. Some of the com-ponents are in use or are ready to use.Examples include text clustering andclassification tools for search on theinternet and other databases, tree valida-tion, clustering and classification toolsfor XML documents, time series recogni-tion and time series prediction tools,internet crawler and hostess technologyin JAVA.

At each stage, the technologies we haveadopted or developed undergo functionaltesting. For example, our internetcrawler has been under testing. Medical

The Neural Information Processing Group of the Eötvös Loránd University, Budapest,launched a project 24 months ago to develop a general methodology for distributedcollaborating goal-oriented experts (including humans) without using any assumption onsynchronization. The expert community will have adaptive means for distributed computing,editing, and decision-making. Medical applications and home monitoring are in focus.

Global Ambient Intelligence: The Hostess Concept by András Lörincz

Crawlers: Left: breadth first crawler, middle CF crawler of the NEC Institute, right: ELU NIPGcrawler. The diameter of the red circles is proportional to the number of hits at a givensite. The color of lines shows recency of the search. Dark blue: early part of the search.Light blue: recent part of the search. The size of travel is inversely proportional to theline thickness.


25

Ubiquitous computing aims at makingcomputers available throughout the envi-ronment, while rendering them effec-tively invisible. Real-world objects thatcontain embedded processors andsensors provide novel ways of accessinginformation, but they may also react totheir environment and they may providenew emergent functionality when inter-acting with other smart things.

This vision of smart objects in particularand ubiquitous computing in general isgrounded in the belief that microproces-sors and advanced sensors will soonbecome so small and inexpensive thatthey can be embedded in almost every-thing. It is expected that billions of suchobjects will be interwoven and con-nected together by wireless networks,forming a world-wide distributed systemseveral orders of magnitude larger thantoday’s Internet. Infrastructures forcooperating smart objects have to copewith a highly dynamic environment andshould, among other things, providelocation information to mobile objects,

represent context information, andenable reliable and scalable service cre-ation.

One of the major ubiquitous computingprojects at ETH Zurich is the Smart-Itsproject. It is one of 16 projects conductedunder the European Union’sDisappearing Computer initiative withinthe Future and Emerging Technologiesprogramme. Its goal is to develop unob-trusive, deeply interconnected smartdevices (so-called ‘Smart-Its’) that can beattached to everyday items in order to

support new functionality, novel interac-tion patterns, and intelligent collaborativebehavior. Eventually, Smart-Its should beas cheap and as small as state-of-the-artradio tags (RFIDs), but in addition theywill also be able to communicate withpeers, and they will be customizable intheir behavior. In order to facilitate ameaningful integration in their environ-ment, Smart-Its are equipped with varioussensors providing context information.

Smart-Its only develop their inherentpotential when acting in a collaborativeenvironment together with other Smart-Its providing different sensor informa-tion and services. They require a back-ground infrastructure in order to accessdistributed services, connect to remotedevices, or exchange application-spe-cific information. Among others, suchservices include location and securityservices as well as services for the prop-agation of context information.

The Smart-Its project is conducted incooperation with the Perceptual

Incorporation of computing power into everyday objects gives rise to ‘smart things’. Toenable communication and cooperation among such smart objects, new informationinfrastructures are required. The Distributed Systems Group at ETH Zurich addressesthe challenges of designing and implementing such infrastructures.

Ubiquitous Computing Infrastructuresby Friedemann Mattern

doctors will help to improve graphicaluser interfaces. Two other componentsclose to testing are: • connectionist knowledge representation

(CFR) for individual (human) expertswhich/who are not present duringdecision-making

• facial expression recognition software.

Results on these components will beavailable in six months. CFR will betested by its distributed editing capabili-ties for expert teams. We see no bottle-neck for global ambient intelligence.Today, all essential technology compo-nents are within reach, including:• JavaSpaces, the flexible message-

ware between available/existingsoftware components at various sites

• XML access control (XACL) forprovisional authorization

• reconfigurable hardware components.

Sub-components of the project havepotentials for other immediate applica-tions.

Individual sub-projects have been andare supported by: • EOARD of US Air Force, London

(Structure based identification ofXML documents)

• Honda Future Technology Research,Germany (image processing forhumanoid robots)

• Panasonic, Osaka (Software-hardwareco-synthesis with pipeline operationusing intelligent properties).

The figure is about our internet crawlerthat combines support vector machineand RL. Three graphs show the searchpatterns of the traditional breadth firstcrawler, the state-of-the-art contextfocussed (CF) crawler (NEC Institute),and our crawler. It can be seen that ourcrawler travels further and collects moreinformation (given by the red circles).The efficiency of our crawler is almostan order of magnitude better than that ofthe state-of-the-art NEC crawler.

Link: http://www.inf.elte.hu/~lorincz

Please contact:András Lörincz — Eötvös Loránd University, HungaryTel: +36 1 463 3515E-mail: [email protected]

One of the first Smart-Its prototypes,based on an Atmel microcontroller.Bluetooth modules allow communicationbetween different devices.

ERCIM News No. 47, October 2001



One of the main purposes of an operatingsystem is to provide the basic mechanismsto manage hardware resources availableon architectures in order to run applica-tions in an efficient way. Obviously, itspurpose is not limited to hardwareresources management, it provides allfacilities for their virtualization to makeresources accesses easier.

Resource management is always theproblem that operating systems researchershave to solve. Obviously solutions pro-vided ten years ago, need to be stronglyreconsidered, mainly due to the evolutionof the hardware architectures and also tonew application domains. Some of theseapplication domains have already beencarefully considered. For example a lot ofresearch has been carried out concerningmultimedia applications. Here, one of themain problems is to guarantee a ‘quality ofservice’ (QoS) which can be defined interms of security, availability or efficiency:Today, the growing interest for mobiledevices (PDA, cell phones, CarInformation Systems, etc.) requiresadvanced researches in the emergence of‘Ubiquitous Computing’. Intuitively it can

be defined as making computer systemspervasive, enhancing user interaction withthe environment, as seamlessly as possible.Obviously context awareness plays a majorrole in this area. The UbiquitousComputing research area is not new, MarcWeiser identified it ten years ago.However, at that time the need of specificequipment in term of communication facili-ties or terminals made difficult to imaginewhat kind of applications could benefitfrom the new research area. Today, the situ-ation is totally different with the develop-ment of wireless communication facilities(eg GSM), the new generation of mobilephones and PDA, the short distance wire-less technology (wavelan, bluetooth, etc.).Some research projects are running, butthey are mainly concerned with the defini-tion of user interfaces or interaction sce-narios between the user and its PDA, whichleads to the notion of ‘wearable computers’.A great deal of research remains to bedone about operating systems to supportUbiquitous Computing. Reasons for thisare:• these systems could be considered as

embedded systems as they have toensure some minimal running facilities

in a strictly autonomous way during amission for which we do not know, apriori, the tasks which will be run

• these systems may have to ensure softreal time constraints due to possiblemultimedia applications

• these systems have to take into accountthe specificity of some kind ofresources, in particular the energy, thephysical context and the mobility of theuser, and the limited amount of someresources such as memory or CPU

• these systems have to be designedtaking into account the instability ofcellular networks (bandwidth,connection, etc).

To summarize, we can advocate that oper-ating systems have to be reconsideredtaking into account constraints related tothe physical environment at large. To dothat, we need to examine the impact ofubiquitous computing in areas such thatresource allocation, load balancing tech-niques, scheduling algorithms, etc. It isclear that a lot of results exist concerningthese activities, but the challenge is toprovide global solutions, which take intoaccount simultaneously, for example, soft

A research activity at INRIA concerns the impact of Ubiquitous Computing on operatingsystem design, particularly the aspects related to Java for appliances, SpontaneousInformation Systems and context awareness.

Ubiquitous Computing and Embedded Operating Systems Design by Michel Banâtre

Computing and Computer Vision Group(ETH Zurich), the Cooperative SystemsEngineering Group (Lancaster University,U.K.), TecO (University of Karlsruhe,Germany), PLAY (Viktoria Institute,Sweden), and VTT. The research areasbeing addressed include: embeddeddevice development, perceptual com-puting methods for collective perception,infrastructures for smart devices, architec-tures for context-awareness, and evalua-tion of application scenarios.

At ETH Zurich, we are engaged indeveloping and evaluating appropriatecommunication schemes and adequateinfrastructures for cooperating Smart-Its.However, we also interested in securityand privacy issues, since this will be of

prime concern in a world of highly inter-connected, autonomous smart devicesthat will eventually permeate oureveryday lives, effectively placing usunder constant surveillance.

Furthermore, we cooperate with theUniversity of St. Gallen in the M-Labproject to identify and create businessapplications for smart things in the area ofbusiness-to-business. The goal of the M-Lab is to build up a critical mass of highlyqualified researchers and practitioners inthe field of applied ubiquitous computing.

Ubiquitous computing is also part of ourteaching activity. Besides teaching fun-damentals of distributed systems, dis-tributed algorithms, and Internet tech-

nology, our graduate level curriculumfocuses on state-of-the-art research inubiquitous computing. In our ubiquitouscomputing laboratory, students candevise and build their own smart envi-ronments using handheld devices (eg,mobile phones or pen-based computers),ID systems (eg, smart cards or RFIDtags), and wireless communicationequipment such as Bluetooth or WLAN.

Links:http://www.inf.ethz.ch/vs/http://www.inf.ethz.ch/vs/res/ubicomp.htmlhttp://www.inf.ethz.ch/vs/res/proj/smartits.htmlhttp://www.m-lab.ch/

Please contact:Friedemann Mattern — SARIT/ETH ZurichTel: +41 1 632 05 36E-mail: [email protected]



real time requirements, energy consump-tion and short distance wireless connection.

Our research theme is devoted to thedesign of such operating systems, throughthree main activities. One is related to thedesign and the implementation of a JavaDevelopment Environment that allows thedevelopment of Java multimedia applica-tion targeted appliances, the other oneconcerns spontaneous informationsystems based on ‘Short DistanceWireless’ (SDW) technologies, and thelast one is about location awareness.

Java Development Environment for PDAThe main goal of this activity is to envi-sion the use of wireless appliances thatallow the user to perform actions such asInternet accesses, application down-loading, and phone calls. However, theactual provision of the aforementionedenvironment on the wireless appliance isfar from straightforward; hardware andsoftware solutions still need to be devised.Our research concerns software solutionsdesign. A wireless software environmentmust meet the following requirements:• maximize efficiency so as to support

efficient execution of applications,possibly run concurrently

• minimize power consumption so as toprovide users with highly autonomous,small-sized appliances

• maximize availability so that users canbenefit from their appliances despitethe occurrence of failures

• support multimedia applications, whichhave ‘soft’ real-time constraints

• support dynamic downloading ofapplications (eg through Internet).

A Java environment achieves the latterrequirement. Furthermore, the de factostandard status of Java leads to a signifi-cant amount of available software, whichwill continue to grow. However, pro-viding a Java environment for appliancesthat both supports multimedia applicationsand exploits the underlying hardware,which could be a multiprocessor, is anopen issue and constitutes the main core ofour current research. A Java runtime envi-ronment has been designed that allows thedevelopment of Java multimedia applica-tions, targeted for appliances. Such anenvironment supports:

• concurrent execution of multimediaapplications, by handling associatedtimeliness requirements

• efficient execution of applications byexploiting the underlying architecture

• parallelism together with minimalpower consumption, through theprovision of complementary schedulingand load balancing policies.

Currently, the research is devoted totaking the benefit of the underlying hard-ware in order to improve the efficiency ofour environment.

Spontaneous Information SystemsA spontaneous information system is asystem composed of computing devices thatare carried by people and that communi-cate by a Short Distances Wireless(SDW) interface. Devices can only com-municate with each other when they arewithin a relatively short distance. Sincedevices are mobile, a device may leave anetwork at any time – simply by beingmoved out of communication range of otherdevices. Spontaneous networks are analo-gous to human groupings: people can onlydirectly communicate when they are close,strangers that meet may exchange informa-tion, and this exchange may lead to furtherexchanges (notably of more sensitive infor-mation) as the relationship develops. Aperson can find himself in a network at anytime, and the network around him or hervaries over time. It is very easy to imaginerelevant applications of this paradigm:medical domain, e-commerce, PDA-to-PDA information exchange, games, etc.

Among all the challenges we have identi-fied, we have selected some whichdemonstrate the real need for newresearch. One is coming from the fact thatthe network composition is dynamic andlocalized. Devices can only communicatewith devices in their neighbourhood, andonce a device leaves the range of commu-nication, it no longer forms part of thatnetwork. A device can only use the infor-mation available on its current network.The second one is autonomy. For instance,many spontaneous systems are formedmainly by PDAs, which are pro-grammable devices. On the one hand, thismakes the range of possible applicationslarge. On the other hand, it introducesextra complexity since devices mayinteract in unforeseen ways. For example,

a device may use a protocol that otherdevices are unaware of.

The last important feature of spontaneoussystems is that the value of the informationcan be highly localized in time and place.For instance, information about a trafficjam can be useful to a person in a car whois one kilometer from the jam, but theinformation is useless one day later or toanyone who is not in a car. Information ina spontaneous network is highly contex-tual and situational.

All of these problems combined are partic-ular to spontaneous information systemsand suggest a new form of building andoperating systems as the user’s physicalmobility has to be taken into account at theresource management level (energy, CPU,etc). Currently we are involved in twomain directions, one related to the pro-posal of predictive scheme in order to esti-mate the communication time dependingon users’ mobility. The other is concernedwith data location and accesses. In order tointegrate all these results a demonstrator isunder construction.

Location AwarenessLocation awareness is the last importantaspect to consider: ubiquitous computingenables the system to fade into the environ-ment, by its potential to guess the contextof the users without requiring explicitinteractions from them. Up to now, mostlocalization services were used to buildapplications related to navigation (helpingthe user to locate himself or other entities inthe environment, like GPS car navigationsystem). We believe that a lot of researchstill has to be done for a true context-awaresystem, which uses localization as a part ofthe context representation (along withtime), to go further than a navigation appli-cation. Currently we have designed anoriginal solution based on SDW technolo-gies to dynamically sense the relevantcontext aspects of a mobile user.

AcknowledgementsJava Development Environment for PDAactivity is supported by Texas-Instrumentsand the Spontaneous Information Systemsactivity is supported by Alcatel R&D

Please contact:Michel Banâtre — INRIATel: +33 2 99 84 72 85E-mail: [email protected]



Knowledge workers typically work inboth the physical world with physicalartefacts and in the electronic world withelectronic artefacts. At the same time theywork at various places and cooperate withcolleagues at the same or at a distantplaces. So, as the knowledge workers’activities and plans span across media,locations and work groups, they need tospend a considerable amount of time andenergy for orientation and coordination.

We are developing a hybrid setting thatbridges the gap between physical andelectronic worlds and provides a frame-work for orientation in both worlds. Inthis hybrid setting, ambient interfacesare the basis for a shared environment ofmutual information fostering intuitiveand adequate behaviour of distributedparticipants in a cooperative setting. Theambient interfaces act as sensors in thephysical world and capture various typesof information such as the position andorientation of a user, noise in a coffeeroom or movement in the hallway. Onthe other hand they act as indicators pre-senting digital information in the phys-ical environment of the user by means ofprojections of digital information on thewall and physical artefacts such as fishtanks, propellers, or robots (see Figure).

In a first generation of ambient interfacesthe sensors were mainly developedbased on existing technology. TheMOVY system was used to capture theposition and orientation of a user (see

ERCIM News No. 36, January 1999); aWebCam with microphones and standardsoftware was used to capture noise; andthe ‘Vision Command’ from LEGO wasused to capture movement. Some simplebinary ambient interfaces, which can beswitched on and off, were developed topresent the information. Relaisboards witheight relais allowed to control eight binaryambient interfaces. For instance, the fishtank starts releasing bubbles, or a propellerstarts blowing air into the user’s face whencertain events occur. Depending on thefrequency of the releases of the bubbles,the bubbling starts to aggregate, ie, wheninteresting events occur in high frequency,there are still bubbles from the formerreleases and the bubbling on a whole getsrougher indicating that a lot is going on.Additionally, the release of the bubblesproduces subtle sounds of the motor andthe bubbles. Thus the user can capture theinformation visually and acoustically. Thepropeller addresses the haptic sense of theuser by blowing air in a subtle way into theuser’s face.

This first generation was deployed inusers’ offices and in a public coffee room.From the users’ feedback three mayorrequirements for the second generation ofambient interfaces were derived: cus-tomisability (ie, the developers and usersshould be able to easily and rapidly buildand adapt ambient interfaces); aesthetics(ie, the users should like and appreciate thepresence of ambient interfaces); and lowcost (ie, it should be possible to deploy big

numbers of ambient interfaces to indi-vidual offices and public places in officebuildings).

In the second generation we developedAwareBots, ambient interfaces presentingawareness information with the gestalt ofrobots. Based on LEGO’s MindstormsRobotics Invention System, a broad varietyof AwareBots were designed and createdby project members. For instance, theAwareBot in the right picture of the figurecan indicate the login of a colleague bylifting its hat and the manipulation of a doc-ument by rotating its body. In order to loginto the system and to set the status to‘Available’, the user simply presses the armof the AwareBot. Other AwareBots can rolltheir eyes when the user’s Web pages areaccessed and capture movement outside ofthe owner’s office. Additionally, a WAPinterface was developed that allows mobileusers to query for information and enterinformation manually while on the move.

For the design of the third generation theexisting designs will be evaluated and pos-sibly consolidated. As an overall output weplan to come up with basic guidelines forthe design of ambient interfaces capturingand presenting information in the user’sphysical environment in order to provideshared guidance. The ambient interfacesdescribed are based on the TOWERsystem and its event and notificationinfrastructure (cf. ERCIM News No. 42,July 2000). They are being developed inthe IST-10846 project TOWER, partlyfunded by the EU, with the partnersAixonix, blaxxun interactive AG, FhG,UCL, and WS Attkins.

Links: http://tower.gmd.de/http://orgwis.gmd.de/~gross/

Please contact:Tom Gross — Fraunhofer Institute for Applied Information TechnologyTel: +49 2241 14 2091E-mail: [email protected]

The Computer-Supported Cooperative Work group of the Fraunhofer Institute forApplied IT develops ambient interfaces bridging the gap between the physical and theelectronic world and supporting distributed work groups with a shared environment ofmutual information and orientation providing a basis for smooth coordination.

Ambient Interfaces for Distributed Work Groupsby Tom Gross

User with TOWER world; fish tank; propeller; AwareBot.



GeoNotes started in the summer of 2000and is an ongoing project at theHUMLE-lab at Swedish Institute ofComputer Science. Although the finalapplication will rely on systems thathandle automatic positioning of mobiledevices, the primary workload so far hasbeen on the application side. TodayGeoNotes is a Java application thatallows a user to add content, a GeoNotesplace label, sender information, andaccess restrictions. The user thenchooses a lat-long to define the place ofthe GeoNote, either through a map inter-face or ‘here’.

The base server handles transactions ofnotes for all clients (see Figure 1). It per-forms requests from clients to insert notes,retrieve notes from a location, and toupdate a single note. It stores the actualnote data (its contents), be it text, audio, orvideo, as an entry in the Wherehoodatabase (http://wherehoo.media.mit.edu/).Wherehoo is a service for storing loca-tion dependent data in a client indepen-dent manner. It stores data coupled to afield that describes the physical locationof the data as a latitude/longitude posi-tion. To retrieve data, one specifies thelocation and a radius and the servicereturns all entries in that area. The server

stores meta information about the usageof the note, such as the number of timesit has been read, when it was last read,and so on, as data in a MySQL database.

GeoNotes is an open and mass-scaleannotation system, allowing not onlycommercial or organizational annotatorsto add to the information space, but allusers. In this way it is more similar tograffiti, toilet scribbles, sticky notes andposters than to commercial marketingchannels.

As a mass-scale annotation system wecan expect public and private spaces tobe cluttered with GeoNotes. These tracesof other people create a nice socialoverlay over geographical space, andthus support social awareness and pro-cesses in those spaces. At the same time,however, information overload willbecome a problem, making navigationsupport and filtering the core foci ofGeoNotes.

Users can browse a map for GeoNotes oruse the at-a-glance overview displayedin Figure 2, which provides GeoNotes ofthe immediate surroundings. The size ofthe circle represents the number ofGeoNotes at this particular place. Thepie-chart represents different sorts ofGeoNotes, for instance notes specificallydirected to the user or relating to somecontent. At the right hand side of theinterface, there is a scrollable list of allGeoNotes at this specific place. Onceclicked, the user accesses the individualGeoNote. The next step in the imple-mentation is to provide a search enginefor GeoNotes (perhaps in combinationwith a map interface).

GeoNotes pushed to the user will be fil-tered with social recommendation andcollaborative filtering techniques.Information about users putting up notes,reading notes, rejecting notes or even

rejecting GeoNotes senders (MacDonald’sadvertisements, for instance), providevaluable information to profile the userand to cluster those profiles with otherusers’. In this way only those GeoNotesthat are shared within such a cluster ofsimilarity will be pushed to the user. Suchfiltering techniques can also be comple-mented with traditional word-basedsearch engines: ‘only push GeoNotes thatcontain the words ONE-ROOMAPPARTMENT or CAR FOR SALE’.Future activities of the project involvedeploying the GeoNotes service on somepositioning enabled PDA and conductinguser evaluation.

Links: http://www.sics.se/humle/projects/GeoNotes/

Please contact: Per Persson — SICSTel: +46 70 600 1805E-mail: [email protected]

Positioning technology is often associated with locating people in geographical space.The GeoNotes system, however, positions pieces of information. GeoNotes allows allusers to annotate physical locations with virtual ‘notes’, which are then pushed to oraccessed by other users when in the vicinity.

GeoNotes: Social Enhancement of Physical Spaceby Per Persson and Fredrik Espinoza

Figure 1: The GeoNote SystemArchitecture. Clients connect to theserver using RMI. The server uses asocket connection to store content dataabout notes in the Wherehoo databaseand SQL commands to store meta dataabout the usage of notes in the MySQLdatabase.

Figure 2: The pie-chart circle provides an‘at-a-glance’ overview of the immediateGeoNote information space. On the righthand side of the screen is the list of thetitles of individual GeoNotes.



The baseline for the acceptance of infor-mation and communication systems is asimplified information presentationadapted to situation, task and user. On theone hand, simplified means a reductionof information by adapting the informa-tion presentation to the context (a task,situation and user adapted choice ofinformation) but on the other hand, itmeans driving at an intuitive interactionwith the SAiMotion-system itself. Thisarticle will concentrate on the first aspectof information adaptation. In SAiMotion,simplified information presentation isachieved through automatic proactiveselection of information and context-sen-sitive presentation of the contents withexplicit consideration of important envi-ronment variables, like daytime, noise,and the situation of communication.

Context in Nomadic Computing For context adapted information presen-tation and interaction the development ofa suitable context and process model isvital determining the individual informa-tion needs on the basis of location, envi-ronment, situation, user, task and activa-tion features. In most context definitionsfour dimensions of context are consid-ered (Klemke, 2000; Schilit, 1994): thelocation of the user in either electronic(eg an URL) or physical space, the iden-tity of the user implying a user modelwith information about the user’s inter-ests, preferences and knowledge, thetime (day/night time working hours,weekend, etc) and the environment (thetask or activity in a current situation;other users). Approaches that takecontext into their account usually focusone or only few aspects like the userscurrent position in physical space toadapt the information selection and pre-sentation. Beyond to these approachesSAiMotion aims to provide an exhaus-tive situation model identifying and

using all relevant situative parametersfor proactive information supply anduser interaction.

Contextualisation The process of information contextuali-sation requires filtering, annotating andaggregation of information contents. Theidentification of the relevant situativeparameters in agreement with the abovedescribed context dimensions help tosolve this task. The information need ofthe user is determined by the primary taskas well as his present location. The objec-tive is to assist the user by a proactivesupply of actually relevant information(and functionality) so that the search andselection of information does not distracthim from the primary task. Informationand communication systems thus arebeing used to enrich the professional andpersonal weekday with useful additionalinformation.

Information Selection and AnticipationContext adapted services are dependenton the reception of the situation of the

user implying the physical environment(location, objects in the vicinity, lightintensity, volume) actual tasks andtargets of the user, the current state ofaffairs in the processing of the task, thecharacteristics of the device – concerningin- and output of information, as well asthe user’s profile of interests and prefer-ences. Depending on the present situationthe probable current information need isnarrowed down – and is possibly pre-dictable, so that the information neededcould be anticipated by the system.

Additionally the emerging contextscreate a framework for user coordinationat different locations allowing animplicit coordination of location inde-pendent communication and cooperationof users. As already mentioned theproblem of privacy is to be taken intoaccount since users are not alwayswilling to disclose their situation or partsof their situation (eg their location) to theSAiMotion-system or other users.

In contrast to location and situationindependent information access deter-

Contextualised information presentation and interaction gives NomadicComputing its backing. At the Fraunhofer Institute for Applied InformationTechnology (FIT), prototypes and services are currently being developed in theframework of the project ‘Situation Awareness in Motion’ (SAiMotion), acooperation between FIT and other Fraunhofer Institutes that concentrates oncontext modeling and Human-Computer Interaction.

Contextualisation in Nomadic Computingby Markus Eisenhauer and Roland Klemke

Context Typology

Context

Timeday/night-time

working hours /leisure time

Locationphysical space

electronical

environmenttask / activity

other users

Identity

interests

preferences

knowledge



mining the paradigm of ‘anything,anytime, anywhere’ it is vital for theabove described activities to provideinformation dependent on the presentlocation and situation and to supportsuitable methods of user interaction andinformation presentation: TheSAiMotion-system provides support forthe completion of a task, by providingnavigation help, information aboutnearby local resources that are useful forthe completion of the task, selectinginformation adapted to the particular sit-uation, task and user preferences, moni-toring the current goals and sub goalsincluding a warning of possible conflictsand adapting the interaction with theSAiMotion-system to the modalityappropriate for the context. Moreoverthe SAiMotion-system is capable pur-suing multiple tasks: at the counter itreminds the user of an overdue cashtransfer and presents the current balance,at the supermarket it provides the shop-ping list and the current special offers, inan exhibition it provides a tour or infor-mation about the exhibits, on a journey it

provides useful information about thecity, including map, hotel information,transportation, current events, sight-seeing, and culture etc.

The most important advance will beanticipated information supply: Relevantinformation will be proactively selectedfor the present task sequence or interestof the user. With the help of user, situa-tion and task monitoring it will be pos-sible to reduce the information explosionto information that match the interest andneeds of the user to anticipate the nextinteraction with the SAiMotion-systemand the possible content of the nextrequest. The SAiMotion-system proac-tively scans for useful information andhas it ready for the next request.

ConclusionSAiMotion will lead to a great improve-ment in the efficacy of the user interac-tion with the system. Time saving andinformation focusing contribute to agreatly improved interaction and togreater contentment of the user.

Moreover multimodal interaction andpro-active presentation mechanisms thatadapt to environmental variables likedaytime, noise, situation of communica-tion will lead to context sensitive inter-action. The SAiMotion-system on itsown will meet a situation- and user-adapted interaction. Several partners areinvolved in the activities likeFraunhofer-Institut für GraphischeDatenverarbeitung IGD), Fraunhofer-Institut für Arbeitswirtschaft undOrganisation (IAO), Fraunhofer Institutfür Zuverlässigkeit und Mikro-integra-tion (IZM), Fraunhofer Institut fürIntegrierte Schaltungen (IIS);Fraunhofer Institut für Publikations-und Informationssysteme (IPSI).

Link:http://fit.gmd.de/projects/saimotion/

Please contact:Markus Eisenhauer — Fraunhofer Institute for Applied Information TechnologyTel: +49 2241 14 28 59 E-mail: [email protected]

By Scene Interpretation, we meandetecting and tracking several individ-uals in a scene (eg an office) and recog-nizing their behavior (eg ‘two personsare meeting near the blackboard’). Theconcept of video communication is dedi-cated to allow permanent and informalCommunication and group awarenessbetween spatially separated individuals(including video-conference). Currentvideo communication (VC) systems risetwo problems. First, the static nature ofthe cameras interferes with the real wordactivity; ie the VC system should auto-matically follow and zoom at the mainindividual in the office. Second, the levelof availability (protection of privacy) hasto be set up automatically, if not, thestudies show that VC systems will beless accepted.

For these reasons, we propose to use aninterpretation system to be a part of theglobal VC system to automate these tasks.

Given image sequences of everydayoffice life, our proposed interpretationsystem is able to analyze the behavior ofindividuals. This system is composed ofthree modules. First, from images thedetection and classification moduledetects moving regions by subtraction toa background image and it classifiesmoving regions into mobile objectsadding a label giving its type, such asindividual or noise. Second, the trackingmodule associates the mobile objectswith already tracked targets that corre-spond to real individuals. Third, the sce-narios recognition module recognizesthe scenarios relative to the behavior oftracked individuals. Finally, thanks to

the scenario analysis, the VC system candecide what to send to the network, azoomed sub-image of the camera or a fil-tered (blur) image. For example, if thesystem recognizes a work meeting situa-tion, the VC system can automaticallybroadcast a blurred image to protectvisitor’s privacy. Moreover, if the sce-nario ‘the user is writing on the black-board’ is recognized, the system canbroadcast an image zoomed on theblackboard area.

In this article, we focus (1) on thetracking module and (2) on the behaviorrecognition module. The trackingproblem is a central issue in scene inter-pretation as the lost of a tracked objectprevents from analyzing its behavior.We have developed a tracking method,based on a 3D model of the scene

Based on Scene Interpretation, a Video Communication System can improvepermanent and informal communication between distant individuals. Such a systemhas been realized by the Orion team at Inria Sophia Antipolis from 1999 to 2000.

Scene Interpretation for Video Communication by Monique Thonnat



Everyday, the average person with acomputer faces a growing flow of multi-media information, particularly via theinternet. But this ocean of informationwould be useless without the ability tomanipulate, classify, archive and accessthem quickly and selectively. Howevermultimedia is dominated by images,with respect to bandwidth and com-plexity.

The IMEDIA team achieves research,collaborations, and technology transfer

on the complex issue of intelligentaccess to multimedia data streams. Theprototype software IKONA illustratesthis research lead at IMEDIA.

Specific and Generic Image SignaturesVisual appearance is automatically mea-sured by numerical signature of imagefeatures such as color, texture, shape, ormost often a combination of them. Morespecific image signatures have to bedeveloped for special content and situa-tions. For designing an effective image

retrieval system, we find it convenient todivide image databases in two categories: • The first category concerns specific

image databases for which a groundtruth is available. When indexing thedatabase, the designer will considerthese ground truths and tune themodels or range of parametersaccordingly, maximising the systemefficiency. We have developpedspecific signature for face recognitionand detection and fingerprintidentification.

The IMEDIA research group at INRIA develops content-based image indexingtechniques and interactive search and retrieval methods for browsing large multimediadatabases by content.

Mining Multimedia Documents by Visual Content by Nozha Boujemaa and Marin Ferecatu

(including 3D geometric and semanticinformation), on explicit models of indi-viduals (including the 3D size of themobile objects and the presence of skinareas) and of individual’s trajectoriesand on the computation of several pos-sible paths for each individual (using atime delay to improve the robustness ofthe algorithm). Resulting algorithm isable to cope with severe occlusions,merging/splitting individuals, errors indetection and in classification. For thebehavior recognition module, we havedeveloped a description language, closeto natural language, which enable usersto specify the behaviors they want to berecognized. By this way, the users canconfigure their own VC System andsetup a desired level of their privacy.

This interpretation system has been testedon several office sequences. The longestsequence is 15 minutes time long. Someof these sequences are available on theweb, at http://www-sop.inria.fr/orion/personnel/Alberto.Avanzi. Thesesequences show people entering theoffice together, crossing and overlappingeach other. An example of scenariorecognition “two people are meeting inthe office” is illustrated by the figure.This scenario illustrates the case whenguests are visiting the office and whenthe availability level should decrease (wedon’t want to broadcast guest pictures).Guests can be recognized according tothe place where they sit. On a 3D anima-tion, a flag (cube), showing the avail-ability level, changes its color following

the number of people in the office: green(nobody in the office, maximum of avail-ability), orange (low probability of guestsin the office, medium level of avail-ability) and red (high probability ofguests in the office, meeting situation, noavailability). Currently, the two mainlimitations of the system are the mix ofindividual identifiers in difficult cases (eglong crossings) and to handle situationswhere individuals are mixed with mobileobjects corresponding to noise (eg a chairmoved by a person).

The first results of this VideoCommunication System based on SceneInterpretation are encouraging. We areplanning to model more sophisticatedscenarios and to complete the scenemodel to include 3D scene objects thatare able to move (like chairs and doors).This new model will prevent paths to bemixed with mobile objects correspondingto the motion of these scene objects. Thiswork has been done in collaboration withthe IIHM and PRIMA teams from theUniversity Joseph Fourrier in Grenoble.

Link:http://www-sop.inria.fr/orion/

Please contact:Monique Thonnat — INRIATel: +33 4 92 38 78 67E-mail: [email protected]

The left image illustrates the tracking results, whereas the right image comes from a 3Danimation illustrating the behavior recognition. Colored cylinders represent theindividual identifiers detected and tracked in the office. Each person has it’s own color.



• The second category includesdatabases with heterogeneous imageswhere no ground truth is available orobvious. Examples include stockphotography and the World WideWeb. The user should be assumed tobe an average user (not an expert). Inthis context, generic image signaturesare computed in order to describegeneral visual appearance such ascolor and texture.

IKONA System ArchitectureOur CBIR software IKONA, is based ona client-server architecture and aims tobe flexible, easily extensible, easy to use,intuitive, and does not enforce specialknowledge or training.

The server needs to be fast and is writtenin C++. It includes image feature extrac-tion algorithms (signatures), user inter-action policies (retrieve by visual simi-larity mode, relevant feedback mode,partial query mode, points of interest

mode, etc...) and a network module tocommunicate with the clients.

The client needs to be portable and iswritten in Java; it normally should run onevery computer architecture that sup-ports Java Runtime Environment (JRE).It presents the user with an easy to useGraphical User Interface (GUI), sets thequery mode for the server and displaythe search results.

Figure 1: Generic image database. Left: Selecting a query image in IKONA; the red border indicates the selected image. Right: Query results for the selected image.

Figure 2: Specific Image database. Left: Selecting a query image with relevance feedback mecanism in IKONA. Right: Query resultsfor the selected image. For more results on face retrieval, see http://www-rocq.inria.fr/~sahbi/Web/ercim.html



The communication protocol is modularand extensible, ie it is easy to add newfunctionality (new feature extractionalgorithms, searching methods, securityaccess restrictions, etc) without dis-turbing the overall architecture (imple-mentation). By default, IKONA does a‘retrieve by visual similarity’ in responseto a query, which means that it search allimages in all databases and returns a listof the most visually similar images to thequery image.

Interactive SearchIn order to deal with generic databases,Ikona includes a relevance feedbacktechnique which enables the user torefine their query by specifying overtime a set of relevant and a set of non-rel-evant images. Relevance feedback inter-action methods try to use the informationthe user supplies to the system in anattempt to ‘guess’ what are his inten-tions, thus making it easier to find whathe wants. IKONA has a RF mode for cat-egory search in image databases, whichmeans it can help the user to find morerapidly certain categories of images inlarge databases with browsing and userprofiling methods. This interaction allowto compensate the ‘semantic gap’ in suchimage search system.

Region based queries are being inte-grated into IKONA. In this mode, theuser can select a part of an image and thesystem will search images (or parts ofimages) that are visually similar to theselected part. This ineteraction allow tothe user to precise to the system whatpart or particular object is interesting inthe image. In this case, since the query isfocused, the system response is enhacedwith regards to the user target since thebackground image signature is not con-sidered. We have developed segmenta-tion based methods as well as point ofinterest methods to achive partialqueries.

Hybrid Image and Text IndexingWhile text indexing is ubiquitous, it isoften limited, tedious and subjective fordescribing image content. Visual contentimage signatures are objective but has nosemantic range. Combining both text andimage features for indexing and retievalis very promising area of interest ofIMEDIA team. We first work on a way todo keyword propagation based on visualsimilarity. For example, if an imagedatabase has been partially annotatedwith keywords, IKONA can use thesekeywords for very fast retrieval. Basedon the indexed visual features and thekeywords index, IKONA can suggest a

number of keywords for a non annotatedimage and their weight. Further researchon keyword propagation, semanticconcept search and hybrid text-imageretrieval mode are being carried on.

Short Term Involvement and EventSeveral industrial partenership are goingon: Judicial Policy for pedophilia prob-lems, TF1 French Chanel with more effi-cency in TV journal preparation, FranceTelecom R&D, Thalès and ThomsonMultimedia. IMEDIA has organized afirst workshop of a series on multimediacontent access management topic spon-sored by NSF Digital Library Initiativewith IBM (Watson research center) andBerkeley University. This workshop wasattended by approximately 35 distin-guished researchers from both academiaand industry and from both Europe andUnited States. It provide a forum foractive researchers in this area toexchange ideas on the most recentresearch progress and challenges in thearea of multimedia content browsing,indexing and retrieval.

Links: http://www-rocq.inria.fr/imedia/

Please contact: Nozha Boujemaa — INRIA Tel : +33 1 39 63 51 54E-mail: [email protected]

A young trend in consumer electronics isthe ubiquitous, transparent and intelli-gent integration of audio, video, commu-nication and information services thathave become so widely used in the 20thcentury. One of the key features of thisintegration is the ability of digital equip-ment to communicate reliably in a flex-ible fashion, which typically includes theplug-and-play ease of use. The look andfeel of ambient intelligence, however,can be instantly destroyed by a singlecrude error message.

HAVi and the IEEE 1394 high-speedserial bus, better known as FireWire,target a dynamic interconnection of dif-ferent pieces of equipment in the homeenvironment. HAVi is a common,openly-licensable specification for net-working digital home entertainmentproducts that has been developed byeight multinationals. Devices connectedin a HAVi network can inter-operate,using each other’s resources and func-tionality. HAVi is built on FireWire,which is designed to carry all forms of

multiple digital audio and video streamsquickly, reliably and inexpensively. Itsarchitecture is scalable and a user canadd or remove systems and peripheralseasily at any time. FireWire has rapidlybecome the de-facto standard in theinterconnection of digital devices.

Although intelligent household goodsare mostly not safety-critical, validationof their correctness is an important issue.The financial implications of errors orbad performance can be enormous,

In the context of the ERCIM Working Group on Formal Methods for Industrial CriticalSystems, protocols within HAVi (Home Audio Video Interoperability) and FireWire wereanalysed using a combination of state-of-the-art methods and tools for verification andtesting.

How can I be sure that my DVD player understands my TV? by Wan Fokkink, Izak van Langevelde, Bas Luttik and Yaroslav Usenko



because of their mass replication anddamage to consumer interests. The highcomplexity of the protocols underlyingsuch products (specifications typicallyrequire several hundreds of pages) turntheir verification into a major challenge.A number of protocols within HAVi andFireWire have been analysed success-fully using formal methods.

One notable study involved the analysisof the so-called HAVi leader electionprotocol, which was carried out at CWI.In HAVi, controllers, acting as hosts fordigital devices, elect among themselvesa leader with the best capabilities tosupervise their controlling tasks. Eachtime when a change in the network of theunderlying FireWire interface occurs,the controllers need to re-elect a leader.The HAVi leader election protocol wasspecified in the specification languagesLOTOS, Promela and µCRL. Theresulting models were analysed usingmodel checking techniques, which facil-itate the automated verification ofbehavioural properties. This analysisrevealed that the leader election protocoldoes not meet some critical safety prop-erties and that there are situations inwhich the protocol does not converge tothe election of a leader. These errorswere due to a lack of detail in the HAVispecification. They occur when commu-

nication between devices is faster thancommunication between componentswithin one device. These aspects weretaken into account in a subsequent speci-fication of the HAVi leader election pro-tocol.

A second study targeted the Link layer ofFireWire using the language µCRL andthe associated tool set, both developed atCWI. FireWire describes the serial busin three layers: the Transaction layeroffers an interface of high-level instruc-tions, such as reading and writing ofdata, the Link layer transforms theseinstructions into packets of bits, and thePhysical layer is concerned with theactual sending and receiving of the bits.The analysis brought to light a seriouscase of underspecification: when theLink layer receives a packet from thePhysical layer, the handling of a requestfrom the Transaction layer is left unspec-ified. Different solutions were proposedindependently at CWI and at theUniversity of Kiel. At INRIA Rhône-Alpes a model checking analysisrevealed a deadlock, due to a flaw in thecommunication scheme between Linkand Transaction layer. The shortest exe-cution trace leading to this deadlock con-sists of fifty consecutive transitions,which indicates that it would be very dif-

ficult to discover this deadlock by meansof testing.

Third, the FireWire tree identify protocolfor spanning a tree in the networktopology was tackled at the University ofNijmegen. This tree identification fails ifthe network topology contains a loop,which is signalled using a time-outmechanism. With the timed modelchecker UPPAAL it was detected thatthe timing constraints of FireWire aretoo lenient: they allow for scenarios inwhich there is a time-out but no loop inthe topology. Furthermore, the FireWireroot contention protocol was studied,which comes into play in the finale of thetree identify protocol where two nodescompete to become the root of the tree.

Correctness of this root contention pro-tocol was established by a number ofcase studies using a combination ofUPPAAL and probabilistic I/Oautomata. In collaboration with theUniversity of Aalborg, precise timingconstraints were derived. A young initia-tive is the analysis of the IEEE 1394.1standard for Bus Bridges, being devel-oped to connect FireWire networksthrough so-called `bridges’. The initialmodelling of the standard in Promela atCWI and the University of Nijmegen hasalready revealed a number of bugs in thedraft specification, in the algorithm thatupdates routing tables. Some of thesediscoveries have been taken into accountin the latest draft of the standard. Theaim is to conclude this verificationbefore the standard is finalised, thus con-tributing further to the standardisationprocess. Envisioned extensions of thisproject include the analysis of stochasticaspects at the University of Twente.

Concluding, these case studies bringhome the point that communicationbetween for example a DVD player anda TV can benefit from the analysis of theunderlying protocols by means of formalmethods.

Links: http://www.cwi.nl/sen2/http://www.inrialpes.fr/vasy/fmics/http://www.cs.stir.ac.uk/firewire-workshop/

Please contact: Wan Fokkink — CWITel: +31 20 592 4104E-mail: [email protected]

CWI conducts verificationof embedded systems.Illustration: I.A. vanLangevelde (2001).



Wireless sensor networks are receivingincreasing attention due to their ability tomonitor a wide variety of environments,ranging from remote geographical regionsto industrial plants, from office buildings totoxic urban locations. A wireless sensornetwork is usually composed of hundredsor thousands of sensors equipped withcomputation, sensing and communi-cation devices, which are coordinatedin a distributed mode in order tocollect information on their surround-ings. During the lifetime of thenetwork, the information collected bythe sensors is periodically transmittedto sink nodes, which can be eithermobile or base stations. Sink nodesare used by external operators toretrieve the information gathered bythe network.

Due to the limited energy supply ofsensors, protocols for sensor networksshould be designed with the goal ofminimizing energy consumption. Forthis reason, energy-efficient informationdissemination protocols for sensor net-works have been recently proposed in theliterature. These protocols exploit networkredundancy to achieve fault tolerance:when a sensor crashes (either because ofbattery depletion or due to a catastrophicevent), neighboring sensors can cover, atleast partially, its sensing task. However,none of these protocols provide explicitknowledge regarding the state (faulty orfault-free) of the sensors in the network. Inour opinion, the extraction of explicit diag-nostic information from the network couldbe important in those situations in whichsensors repair/ reconfiguration is feasible.

In order to motivate this hypothesis, con-sider the following example. A sensornetwork is used to help rangers to monitora vast natural park. The sensors, whichhave limited energy supplies and aremostly static, provide information aboutthe presence of animals, tourists, fire,flooding and so on. Rangers, who areequipped with mobile stations, move

around the park for control and mainte-nance. Mobile stations are connected to thesensor network through the nearest sensor.This way, the rangers can be alerted ofabnormal events, and they can quicklyintervene where needed. However, if thereis no provision for sensor maintenance,eventually the batteries will be depleted

and crash. As a consequence, the numberof non-operating sensors in the system willincrease until the system gets discon-nected, and is no longer functional. In sucha scenario, sensors should provide diag-nostic information along with sensor data,thus enabling rangers to maintain networkfunctionality by replacing faulty sensors orby recharging depleted batteries. For thispurpose, sensors in the network shouldexecute a distributed diagnosis protocol,either periodically or on-demand.

However, existing distributed diagnosisprotocols have been designed either formultiprocessor computers or for wiredcomputer networks. As a consequence, allthe protocols proposed so far assume thatunits communicate according to the one-to-one paradigm typical of wired networks.This means that, if applied to sensor net-works, these models are unable to takeadvantage of the shared nature of commu-nication, and are thus not feasible or at bestextremely energy consuming.

For this reason, we have developed a dis-tributed silent fault diagnosis protocolexplicitly designed for wireless sensor net-works. The protocol takes advantage of theshared nature of communications and aimsat minimizing the total number of bitsexchanged for the purpose of diagnosis,thus reducing the energy consumption

entailed by the protocol execution.The protocol first constructs a span-ning tree of the graph representing thenetwork topology, then exchangesdiagnostic information only along theedges of the tree. This allows a signifi-cant reduction in the number of mes-sages to be sent for the purpose ofdiagnosis. We have shown that theprotocol exchanges the minimumnumber of bits required by any diag-nosis algorithm for wireless sensornetworks, thus proving its optimality.

We have also studied the problem ofidentifying soft-faulted mobiles in adhoc networks. Contrary to the case of

silent faults, soft-faulted nodes can con-tinue to communicate with other units,although with altered behavior. We haveproposed a new comparison-based diag-nostic model based on the one-to-manycommunication paradigm. We show thatusing this strategy the ability to diagnosesoft faults in the presence of mobility is sig-nificantly reduced with respect to the sta-tionary case, meaning that a somewhatweaker notion of diagnosis should be con-sidered under this scenario. We have alsodesigned a distributed soft faults diagnosisprotocol for stationary ad hoc networksbased on the new model. The analysis ofthe time and communication complexity ofthe protocol shows that efficient soft faultsdiagnosis in this scenario is possible.

Please contact: Stefano Chessa — IEI-CNR/Pisa UniversityTel: +39 050 2212 736E-mail: [email protected] Paolo Santi — IMC-CNRTel: +39 050 3152411E-mail: [email protected]

Scientists at CNR address the problem of identifying faulty mobiles in ad hoc wirelessnetworks and propose an energy efficient silent fault diagnosis protocol.

Fault Diagnosis in Wireless Sensor Networksby Stefano Chessa and Paolo Santi

A park monitoring scenario.


TECHNOLOGY TRANSFER

The challenge of software engineering isto create flexible, reliable, scalable andmaintainable systems. To achieve this,the software engineer must apply a widevariety of tools and techniques. One ofthese is to take advantage of the under-lying application domain, by means of adomain-specific language (DSL), whichprovides a notation that can be used tocompose applications from a set of con-cepts tailored towards a specific applica-tion domain. Details of the underlyingimplementation platform are captured byknowledge built into the DSL compiler.This compiler can translate the DSL into,for example, executable code, calls tolibrary routines, database transactions, orHTML forms.

The Risla language is a case in point. Itwas developed by CAP Gemini andMeesPierson, in cooperation with CWI.Given a description of the essentialingredients of an interest rate product(loan, swap, financial future, ...), theRisla compiler can generate data struc-tures in VSAM format, data entryscreens in CICS format, and COBOLroutines for registering modifications inthe product or for yielding managementinformation. A DSL compiler relievesthe user from unnecessary details andmachine dependencies.

As a result, changes in the ICT infras-tructure have to be accommodated onlyonce in the DSL compiler. After re-com-pilation of all DSL programs, the desiredinfrastructure-related changes have beeneffectuated without making a singlechange to the DSL programs themselves.Contrast this with the traditional situa-tion that domain-related knowledge andinfrastructure-related knowledge areintermixed in the program code. The useof DSLs thus leads to greater flexibilityand less maintenance.

The DSL project is carried out under theauspices of the Telematics Institute (TI)and has as goals to develop methods for

selecting suitable DSL domains, and forcapturing domain knowledge into a DSLand its compiler; to develop meta-toolsfor the rapid prototyping of domain-spe-cific languages; and to gain more experi-ence, via case studies, with the use ofdomain-specific languages in a commer-cial setting. Legacy applications areoften a valuable (but hard to exploit)source of domain knowledge. Therefore,the DSL project takes legacy systemsand software understanding tools as oneof its point of departure. So far, we havegained experience with a variety of anal-ysis and renovation problems, includingdocumentation generation for (mostlyCOBOL-based) legacy systems (incooperation with the SoftwareImprovement Group, ABN AMRO,Roccade and others), and analysis of Cand and SDL systems, both incooperation with Lucent, partner in thenational Telematics Institute (TI). Inaddition, we gained experience with avariety of DSLs: RISLA for the financialdomain (in cooperation with TI-partnerCap Gemini), Sophus for writing partialdifferential equation solvers (incooperation with University of Bergen,Norway), used for instance in the area ofseismic modeling for oil exploration, anda router description language (in

cooperation with TI-partner Lucent).Last but not least, major progress hasbeen made with the development of toolsand meta-tools as incorporated in theASF+SDF Meta-Environment. Resultsof the project can be found atwww.cwi.nl/projects/dsl/

Links:DSL project: http://www.cwi.nl/projects/dsl/Interactive Software Development and Renovation at CWI:http://www.cwi.nl/sen1/

Please contactArie van Deursen and Jan Heering — CWITel: +31 592 4075 / 4130E-mail: [email protected], [email protected]

CWI is developing tools and methods to support the effective use of domain specificlanguages in real life software engineering projects.

Domain Specific Languagesby Arie van Deursen and Jan Heering

Seismic modelling and other computational modelling software benefits from theSophus approach to software construction which is being developed at CWI incooperation with the University of Bergen, Norway.


TECHNOLOGY TRANSFER

Created by several former researchers ofINRIA in March 2001, QuantifiCareS.A. is a French company, located inSophia-Antipolis and dedicated to theapplications of Medical ImageProcessing. The first phase of its activityis to design specific protocols for thefollow-up of disease evolution such asischemia, arrhythmia, osteoporosis,lesions or tumors volumes. Amongst thestudied pathologies are Coronary ArteryDisease, Multiple Sclerosis, Stroke andCancer. QuantifiCare’s techniques takeas input volumetric images such asMagnetic Resonance Images (MRI), cat-scans, Nuclear Medicine images(SPECT, PET∑), 3D ultrasound andprovide quantitative measurements ofphysical parameters.

QuantifiCare is relying upon a uniqueexpertise in 3D image matching in orderto track very precisely subtle changesalong time. Inter-patients image matchingis also used to relate the information ofdifferent individuals and allow for popu-lation model analysis or data miningapplications. An important technologytransfer, relying on a large number ofgeneral-purpose libraries, as well as ondedicated matching methods, includingthree patents, has been signed betweenQuantifiCare and INRIA in June 2001.This technology has been developed overthe years by an INRIA research teamcalled EPIDAURE, created in 1989 byNicholas Ayache. Jean-Philippe Thirion,the Chief Executive Officer ofQuantifiCare as well as several other co-founders of the company, are formersenior researcher of EPIDAURE.

Two product lines have already beendeveloped by QuantifiCare and are cur-rently improved: MPS-Quant, based onthe analysis of Myocardial PerfusionScintigraphy (MPS) studies forCoronary Artery Disease (CAD) andMS-Quant for the follow-up of MultipleSclerosis (MS).

According to the American HeartAssociation, more than 7 millionAmericans have a history of CoronaryArtery Disease and 600,000 people aredying each year in the US, due to thisdisease. Two current treatments are pro-posed, which are bypass grafting(600,000 operations per year) and angio-plasty (500,000 operations per year). Asthese operations are costly and invasive,many alternative treatments are currentlybeing developed, such as angiogenesis,laser revascularization, thrombolitics,radiation therapy. MPS-Quant is cur-rently used in two retrospective drug trialstudies, one for Angiogenesis, involving60 patients and one for laser revascular-ization, with 260 patients.

Multiple Sclerosis corresponds to fewer,but younger patients: about 350,000patients are identified in the US and450,000 in Europe. These patients aregenerally aged between 20 and 40 years.Some drugs have already been approvedfor a subset of MS patients, who are pre-senting the ‘relapsing-remitting’ form ofthe disease. It has been shown, fromMRI measurements, that the number ofrelapses was decreasing when beta-inter-feron is used. However, no long termbeneficial effects have been demon-

strated and many alternative treatmentsare currently under evaluation, for whichQuantifiCare’s technology is providinghighly accurate evaluation of changes,not only with respect to the number ofplaques, but also with respect to hiddeneffects of the pathology such as ‘masseffects’, that is, local volume variationswithin the brain or brain atrophies.QuantifiCare owns the right on apatented technology, which has allowed,for the first time, to quantify sucheffects.

The first customers are pharmaceuticalcompanies. QuantifiCare is designingdedicated measurement to quantify drugefficacy and takes care of all aspectsrelated to imagery in Phase II or PhaseIII drug trials. The proposed services areincluding the centralization of imagedata coming from multiple clinicalcenters, the archiving and remote pro-cessing of the image data and also orig-inal image-based data mining capacities.The results of QuantifiCare’s computa-tion can be used directly to support evi-dence of efficacy for the registration of anew drug or a new treatment.

The current effort of the company is toestablish a set of research contracts and

QuantifiCare is a newly created spin-off of INRIA, located in Sophia-Antipolis anddedicated to Medical Image Processing applications. Its first target is the quantificationof disease evolution for the objective measurement of drug efficacy.

QuantifiCare: A new INRIA spin-off in Medical Image Processingby Jean-Philippe Thirion

Evolution of multiple sclerosisplaques measured usingQuantifiCare’s technology. Topleft is a baseline image slice,extracted from a complete MRscan of the head. Top right,bottom left and bottom right aresuccessive exams of the samesubject, acquired with a 3 weeksinterval each, and matched andsubtracted with the baselineimage. Growing lesions appearas white annuli and shrinkinglesion as black annuli. Othertechniques are used to quantifymore precisely differentattributes of MS plaqueevolutions such as the ‘masseffect’.


TECHNOLOGY TRANSFER

drug trial participation with pharmaceu-tical companies. It is currently recruitingMarketing and Sales persons from thepharmaceutical field. A mid-term objec-tive is to establish formal collaborationswith other academic research teams inimage processing, within or outsideINRIA, to keep the technology of thecompany updated. Each percent gainedin sensitivity for the measurement of

disease evolution can lead to gains ofseveral orders of magnitude in the statis-tical results of a trial involving hundredsor thousands of patients and can decideupon the registration or rejection of thenew drug. The stakes are high andQuantifiCare‚s ambition is to maintainits technology to the up most level, toincrease the chances to measure theslightest effect of a drug, when medical

imaging is used to monitor the evolutionof patients.

Link: http://www.quantificare.com

Please contact: Jean-Philippe Thirion — QuantifiCare S.A., FranceTel: +33 4 93 00 60 86E-mail: [email protected]

Finding textual information on the evergrowing Web is already quite a chal-lenge, let alone searching the forth-coming wealth of multimedia docu-ments. This research provides the basisfor a new generation of search engines ona multimedia Web, and supports digitalarchiving in large companies. It is carriedout by CWI and seven partners (universi-ties as well as companies) with fundsallocated to the Amsterdam Science andTechnology Centre (WTCW).

Digital multimedia documents are beststored centrally and distributed throughbroadband Internet, allowing remoteaccess from anywhere anytime; shortly,even from your mobile phone. And,digital data are easily exchanged, alsoenabling the reuse of existing material.Furthermore, people at home create evermore digitized data: making your owncompact discs with music has become

extremely popular, and digital photo- andfilm-cameras become increasinglycommon. Technology developed in MIAfocuses therefore on retrieving ‘that songwith that nice little melody’, or ‘thatpicture of our sailing, with the Bostonskyline; when the weather was nice, wasit 1997, or 1998?’

MIA’s database team at CWI focuses ontwo themes: scalability and query formu-lation. The target in the first theme is pro-cessing queries in a collection exceedingone million pictures. Commonapproaches based on the colour distribu-tion of an example image break down forcollections of this size, because ofteneither all objects or none are found.Moreover, some ‘back of the envelope’calculations show that the required timeper iteration of the search process yieldsunacceptable response times as soon asthe collection grows beyond several

thousands of pictures. Yet, the toughestproblem in searching picture archiveswith current systems, is that query resultsare hard to understand even for their soft-ware developers. The end-user cannot beexpected to comprehend why an examplepicture of a sunset at sea results in(however beautiful) pictures of Africansavannas. Such inaccurate answers are anunwanted side-effect of the uncertaintyinherent to posing the question, exposingthe wide gap between our high-level per-ception and the (necessarily) muchsimpler methods for image analysis.

We seek to reduce this gap with better for-mulated search requests, based on twodifferent but complementary approaches.On the one hand, we assist the user withautomatic query formulation; in ouropinion, many improvements can beachieved by collecting more informationabout the user’s interest in a carefullycrafted, interactive query process. On theother hand, we attempt to give users moreinsight into the query process inside thesystem itself, enabling them to intervenein that process and eventually adjust it(query articulation). An interesting chal-lenge in our research is to find the correctbalance between these two approaches.

Links:http://www.cwi.nl/WTCW/MIA/http://www.cwi.nl/ins1

Please contact:Arjen de Vries — CWITel: +31 592 4306E-mail: [email protected]

The Multimedia Information & Analysis (MIA) project investigates the problem of accessto multimedial collections.

MIA — Querying Multimedia Databasesby Arjen de Vries

Initial query results (upper row) from a photo collection can be improved by feedbackfrom the user (lower row) (Illustration: CWI).


EVENTS

The 6th ERCIM FMICS workshop washeld in Paris on July 16-17, 2001. Theprogram committee was chaired byStefania Gnesi (IEI-CNR) and Ulrich Ultes-Nitsche (University of Southampton). Theworkshop was held in the LIAFA labora-tory (University Paris-7) and the local orga-nization chair was Mihaela Sighireanu.

The meeting was attended by 38 persons,both from academia and industry In par-ticular, it is worth mentioning that com-panies such as Ericsson, Giesecke &Devrient GmbH, IBM, InstrumentointiOy, Motorola, NASA, Simtech, andTransEDA were present at the 6thERCIM FMICS workshop.

The opening talk of the workshop wasgiven by Jean-Eric Pin, ScientificDirector of ERCIM. During the two days,fifteen talks were presented, carefullyselected by the Program Committee witha rejection rate of 50%. Also, two invitedtalks were given, the first one by CindyEisner (‘Automatic Detection of Vacuityin Temporal Logic’) and the second oneby Joseph Sifakis (‘Modeling Real-timeSystems’).

A special issue of the scientific journal‘Formal Methods in System Design’ willgather the best papers presented duringthe workshop.

The 6th FMICS workshop benefitedfrom the financial support of INRIARhone-Alpes and ERCIM. The work-shop proceedings (240 pages) have beenpublished by INRIA.

Links: FMICS Working Group web page : http://www.inrialpes.fr/vasy/fmics/6th FMICS Workshop web page: http://www.dsse.ecs.soton.ac.uk/FMICS2001

Please contact: Hubert Garavel — INRIATel: +33 4 76 61 52 24E-mail: [email protected]

FMICS is the ERCIM Working Group on Formal Methods for Industrially CriticalSystems. Launched in 1996 by Diego Latella and Stefania Gnesi (CNR Pisa), the FMICSworking group is currently chaired by Hubert Garavel (INRIA Rhone-Alpes).

Sixth ERCIM FMICS International Workshopby Hubert Garavel

The success of engineering depends onthe recognition and use of several syn-ergic factors. While the analyticapproach is the key in solving somewell-defined subproblems, creativityhelps us to get free of prejudices and infinding useful analogies. Moreover, crit-ical thinking opens up the way of argu-mentation with inexact premises andrules. While engineering is based onsuch complex mental features, it isperhaps more transparent than someother human activities: its world is ofman-made artifacts that should be usefulin a broad sense of the word.

This resulted in the opinion that by con-sidering the advancements in engi-neering of intelligent systems, one mayhave a proper view of the state ofresearch in the whole area of Artificial

Intelligence (AI). Knowledge gleanedhere is valid not only in this particularapplication area but provides valuablehints to the perspectives in the theory ofAI and in its potential for solving otherproblems as well.

Having recognized the above features ofengineering problem solving, theInternational Society of AppliedIntelligence (ISAI), more than a decadeago, initiated a series of conferencesnamed Industrial & EngineeringApplications of Artificial Intelligence &Expert Systems (IEA/AIE). TheHungarian AI community was honoredby the invitation of the ISAI to host theIEA/AIE-2001 conference (June 4-7,2001, Budapest, Hungary), which wasthe fourteenth in the series. General chairof IEA/AIE-2001 was Moonis Ali,

program chair and co-chair were LászlóMonostori and Soundar Kumara, respec-tively. The conference was organized bySZTAKI in co-operation with a numberof international scientific societies asAAAI, ACM/SIGART, ECCAI, ERCIMto name but a few ones. The main spon-soring institutions of the event wereISAI, ERCIM, the Hungarian Academyof Sciences, and the Hungarian Ministryof Education.

The Call for Papers announcementattracted many researchers and engineersfrom all over the world. A total of 104papers — representing 21 countries andall continents — were selected for pre-sentation. The accepted papers coveredall the main research areas of AI andmany different application domains.

IEA/AIE-2001 — The Fourteenth International Conference on Industrial & EngineeringApplications of Artificial Intelligence & Expert Systems was held during 4-7 June 2001,in Budapest, Hungary. The conference that covered all main research areas and manyapplication domains of AI was supported in part by ERCIM.

IEA/AIE-2001 Conference Reportby László Monostori and József Váncza

SPONSORED BY ERCIM

SPONSORED BY ERCIM


EVENTS

The mission of the workshop was to con-tinue the tradition of previous workshopsin Interfaces to Database Systems (IDS1992, IDS 1994, IDS 1996) and the 1stUIDIS 1999. The emphasis, however,has been put to a human-centered way ofhuman-computer interaction, especially,when data intensive systems are con-cerned. The workshop tried to find solu-tions to problems such as interactionwith high-dimensional spaces, accessinglarge and semantically difficult to inter-pret database schemas, understandabilityof data and user/system interactions,through a variety of 12 accepted paperswhich have been selected out of 17 sub-mitted papers by the program com-mittee.

In particular, the workshop broughttogether researchers practitioners ininformation management systems, com-puter-human interaction, knowledgebased methodologies and visualization.The papers were organized in three ses-sions which mainly addressed: • context aware data querying and

information retrieval techniques inorder to restrict the navigationalspace and increase understandability

of domain knowledge and data insemantically difficult to interpretdatabases. The presentations wererelated to context aware querylanguages as well as to informationretrieval techniques

• designing effective user interfaces forinteractive systems in terms ofembedding user/system interactionissues into conceptual design (UML),constraints based specification ofuser interface patterns, considering ofpossible outcomes in a user/systeminteraction environment, as well asthe semi-automatical derivation ofuser interfaces by using XML

• providing comprehensive visualizationtechniques which enable therecognition of interesting informationpatterns as well as the presentation ofinformation.

The two invited talks addressed theaccessing and organizing large informa-tion spaces. The first talk put theemphasis on the semantics based (onto-logical) approach, on the example ofcontents management for business-to-businsess electronic commerce, whilethe second talk discussed the use of the

parallel coordinates approach as a visu-alization technique for high-dimensionalinformation spaces. The workshop hasbeen considered a success by the partici-pants and organizers alike. A number ofsponsors have contributed to its suc-cessful organization: ERCIM; theAssociation of Computing Machinery(ACM - Special Interest GroupComputer-Human Interaction(SIGCHI); the Swiss Federal Institute ofTechnology (ETH), Zurich; theGlaxoSmithKline pharmaceuticalcompany, and the University of Zurich.

Please contact:Epaminondas Kapetanios — ETH ZurichTel: +41 1 632 7261E-mail: [email protected]

The second international workshop on User Interfaces to Data Intensive Systems(UIDIS) took place at the Swiss Federal Institute of Technology, in Zurich, Switzerland,31 May - 1 June 2001. The two-days workshop has been attended by 24 participants.

Workshop on User Interfaces to Data Intensive Systemsby Epaminondas Kapetanios

The plenary talk of the conference wasdelivered by Tibor Vámos (SZTAKI) on‘Triple R: Representation, Retrieval,Reasoning’. The technical presentationswere held in 2-3 parallel tracks in ses-sions on search, knowledge representa-tion, model-based reasoning, machinelearning, data mining, soft computing,evolutionary algorithms, distributedproblem solving, expert systems, patternand speech recognition, vision, languageprocessing, planning and scheduling,robotics, autonomous agents, design,control, manufacturing systems, finance

and business, software engineering, andtutoring.

Proceedings of IEA/AIE-2001 were pub-lished by Springer Verlag under the title‘Engineering of Intelligent Systems’inthe series of Lecture Notes in ArtificialIntelligence (LNAI 2070). Electronicaccess to the papers of the volume is pos-sible (see links). The extended versionsof some selected papers are consideredfor publication in the InternationalJournal of Applied Intelligence.

Links:Conference:http://www.sztaki.hu/conferences/ieaaie2001/ Proceedings:http://link.springer.de/link/service/series/0558/tocs/t2070.htm

Please contact:László Monostori, József Váncza — SZTAKI Tel: +36 1 466 5644E-mail: {laszlo.monostori;vancza}@sztaki.hu

SPONSORED BY ERCIM


EVENTS

The conference was chaired by GertiKappel from Johannes KeplerUniversity, Linz and László Varga fromEötvös Loránd University, Budapest.The Programme Committee was chairedby Jørgen Lindskov Knudsen fromUniversity of Aarhus, Denmark.

The first two days (Monday 18 andTuesday 19) were dedicated to work-shops and tutorials. During the last threedays of the week (20-22), the main con-ference took place together with an exhi-bition, demos and poster sessions.

22 workshops and 21 tutorials wereselected out of respectively 30 and 48high quality proposals. The openingkeynote address was given by CharlesSimonyi from Microsoft Research. Hispaper entitled ‘Languages, Objects, andIntentionality’, was a very impressivediscussion on ‘intentionality’ as animportant aspect of natural language,and the uses of intentionality in com-puter languages and in object-orientedprogramming. Erik Meijer, Microsoftgave the second invited talk entitled‘Scripting .NET Using Mondrian’. He

spoke about Mondrian, a functionalscripting language designed especiallyfor the new Microsoft .NET platform.Mondrian is useful for functional pro-grammers who would like to be able tointer-work more closely with other lan-guages that target the .NET CommonLanguage Runtime and for object-oriented programmers who would like toexplore being able to write and accessobjects written in functional languages.

Alistair Cocburn, Humans andTechnology gave the third invited talk

ECOOP 2001 took place at Eötvös Loránd University, Budapest, Hungary from 18 to 22June 2001. It was organized under the auspices of AITO (Association internationale pour lestechnologies objects), with sponsorship from IQSOFT, the Hungarian Ministry of Education,NOKIA, GRAPHISOFT, Microsoft Research, ERCIM, ERICSSON, VT-SOFT, SZÁMALK andSZÁMITÁSTECHNIKA. About four hundred people participated in the events.

ECOOP 2001 - 15th European Conference on Object-Oriented Programming by László Kozma

SPONSORED BY ERCIM

The establishment of this new interna-tional conference is largely due to thededicated work of the ERCIM WorkingGroup ‘User Interfaces for All’, which,since its institution in 1995, has carriedout a variety of activities bringingtogether researchers and teams whoshare common interests and aspirations,and contribute to the endeavours towardsmaking the emerging InformationSociety equally accessible to all citizens.

The UAHCI 2001 conference hasattracted strong international interest,with 228 papers presented, covering awide range of topics in the area ofUniversal Access in Human-Computer

Interaction, such as Universal Design,Adaptive and Intelligent Interfaces,Architectures and Tools, Multimodal,Continuous and Ubiquitous Interaction,User Diversity and User Participation,Human Factors, Ergonomics, Guidelinesand Standards, Access to Information,Applications, Assistive Technologies,and Cultural, Social, Ethical and LegalIssues. The Proceedings of UAHCI 2001have been published by LawrenceErlbaum Associates (ISBN 0-8058-3609-8).

The Second International Conference onUniversal Access in Human-ComputerInteraction will be held in Crete, Greece,

22-27 June 2003, jointly with the TenthInternational Conference on Human-Computer Interaction (HCI International2003, http://hcii2003.ics.forth.gr), theSymposium on Human Interface (Japan)2003, and the 5th InternationalConference on Engineering Psychologyand Cognitive Ergonomics.

Links: Conference website: http://hcii2001.engr.wisc.edu/Working Group ‘User Interfaces for All’: http://ui4all.ics.forth.gr/

Please contact: Constantine Stephanidis — FORTHTel: +30 81 391741E-mail: [email protected]

The 1st International Conference on Universal Access in Human-Computer Interaction(UAHCI 2001) took place in New Orleans, Louisiana, USA, 5-10 August 2001, jointly withthe Ninth International Conference on Human-Computer Interaction (HCI International2001), the Symposium on Human Interface (Japan) 2001, and the 4th InternationalConference on Engineering Psychology and Cognitive Ergonomics.

1st International Conference on Universal Access in Human-Computer Interactionby Constantine Stephanidis


EVENTS

entitled ‘People and the Limits ofMethodology’. He spoke about thatpeople had a nasty habit of runningneatly drown up methodologies. It is notthat they intend to do, it just that peopleare packed as ‘individuals’ while metho-dologies are packed in ‘roles’.

Eighteen technical papers out of the 108submitted papers were selected for pre-sentation and covered a broad range of

topics related to the object-orientedparadigm including sharing and encap-sulation, type inference and static anal-ysis, language design, implementationtechniques, reflection and concurrency,testing and design.

From an organizer’s point of view, theconference was very successful. Thetechnical presentations, tutorials andworkshops were of high quality. We

were successful in attracting manyyoung researchers whose enthusiasmand fresh ideas greatly contributed to thesuccess of the event. The conference pro-ceedings were published by Springer -Verlag, Berlin - Heidelberg in Germany.

Please contact:László Kozma — Eötvös Loránd University,Hungary Tel: +36 1 209 0555E-mail: [email protected]

Being the first Summer School organ-ised by the DELOS Network ofExcellence, it was decided to have abroad programme covering most of thefundamental issues underlying the multi-disciplinary area of Digital Libraries.Future editions of the DELOS SummerSchool may focus on specific issues andtechnologies of interest in the area of theDigital Libraries, providing more indepth coverage of the topics selected.

The ISDL programme was organised innine half-day lectures, having as invitedlecturers leading researchers in thedigital library field, both from the USand Europe. Each lecture was devoted toa fundamental topic of Digital Libraries,addressing both the methodologicalaspects and the presentation of theresults of some of the most innovativeand significant Digital Library projects.The School was divided into two parts:the first part, from Monday toWednesday morning, devoted to moregeneral topics, and the second part, fromThursday to Friday, addressing morespecific topics. On Wednesday after-noon there was an invited talk, a shortdemo and the social programme (aguided tour of Lucca and a social dinnerin an old villa outside of Pisa). Most ofthe participants and lecturers resided atthe conference centre for the duration ofthe School, allowing ample opportuni-ties for getting to know each other and

for an informal exchange of ideas. Theprogramme of the school included:• ‘Introduction to Multimedia Digital

Collections’ by Carl Lagoze, CornellUniversity, USA

• ‘New Models for ScholarlyDissemination’ by Robert Wilensky,University of California, Berkeley,USA

• ‘Architectures and Open Access toDigital Libraries’ by William Y.Arms, Cornell University, USA

• ‘Information Retrieval Models andMethods, Metadata, and Evaluation’by Norbert Fuhr, University ofDortmund, Germany

• ‘Online Information Access fromHandheld Devices’ by AndreasPaepcke, Stanford University, USA

• an invited talk on ‘Public Access toDigital Materials’ by Brewster Kahle,Internet Archive, USA

• a short demo by Rui Amaral, INESCID, Portugal

• ‘Cross-language Retrieval’ by CarolPeters, CNR-IEI, Italy

• ‘Text Categorization and InformationFiltering’ by Fabrizio Sebastiani,CNR-IEI, Italy

• ‘Video Digital Libraries’ by HowardWactlar, Carnegie Mellon University,USA

• ‘Libraries in the Digital World’ byElizabeth Lyon, UKOLN, UK.

The School was attended by 52 studentsfrom 16 different countries (Australia,Denmark, France, Germany, Greece,India, Ireland, Israel, Italy, Korea, Poland,Portugal, Spain, Switzerland, TheNetherlands, UK). 36 students were fromuniversities and research institutions, andthe remaining 16 were from libraries andarchives. DELOS was able to partiallyfund the participation of 24 students bycovering their enrolment fees.

An evaluation form was distributed at theend of the course asking students to ratedifferent aspects of the school on a scalefrom 1 to 5 (bad to good). About 75% ofthe students filled in the form with 86%of the students giving a rating of between4 and 5 for the school overall, 85% givinga rating of between 4 and 5 for the socialevents overall, and 87% giving a rating ofbetween 4 and 5 for support overall.Lectures were rated individually, with70% of the ratings falling between ascore of 4 and 5. A section of the evalua-tion form provided space for commentsand suggestions which will be consideredwhen planning the next Summer Schoolin June 2002. All in all, the school can bejudged to have been a success, with manystudents expressing interest in attendingISDL2002.

Please contact: Vittore Casarosa — IEI-CNRTel: +39 050 315 3115E-mail: [email protected]

The First DELOS International Summer School on Digital Library Technologies (ISDL 2001) was held in Pisa, Italy, from 9-13 July 2001, under the direction of Maristella Agosti, University of Padua, Italy.

First DELOS Summer School on Digital Library Technologies by Maristella Agosti and Vittore Casarosa


EVENTS

The concept of personalisation is aboutmaking systems different for individualpeople or groups of people and one typeof personalisation that is growing in use isrecommender systems. These take inputdirectly or indirectly from users and basedon user needs, preferences and usage pat-terns, they make personalised recommen-dations of products or services. Thesevary from recommending books to buy orTV programs to watch, to suggesting webpages to visit. The ultimate goal of suchrecommender systems is to reduce theamount of explicit user input and tooperate, effectively, based on usage pat-terns alone, thus giving users what theywant without them having to ask.

Within the digital library environment,personalisation and recommendersystems may have different characteris-tics because individual user behavior andtraffic patterns may differ significantlyfrom those of Web and E-commerceenvironments. For example, few digitallibraries will see millions of transactionswithin a short period of time, digitallibrary resources may be structured andmore stable as compared with commer-cial sites, and digital library characteris-tics and usage patterns may provideopportunities for types of long-termlearning that would be difficult orimpractical in other environments.

Fifteen papers and three invited talks werepresented at the workshop, while moder-ated discussion sessions at the end of eachday provided an opportunity for partici-pants to reflect on the day’s talks and toaddress recurring themes and issues.Although the workshop included talks onsoftware architectures, several deployedsystems, and a user study, most of the pre-sentations focused on algorithms formaking recommendations. In spite of this

emphasis the papers were generally con-sistent with past research in this area,which has been adaptation of existingtechnologies, mostly from InformationRetrieval or Machine Learning, oftencombined with creative methods ofacquiring training data. This suggests thatthe area is not yet mature enough to havespawned its own specific and tailoredtechniques, or the theoretical models thatwould support them, which is a sign thatthere is much more work yet to be done.

Workshop participants naturally felt thatpersonalisation is an important researcharea, and that it will receive greater atten-tion in the coming years but there was lessagreement on how it will be applied in thecontext of digital libraries. Most of the‘success’ stories are systems thatencourage additional consumption, forexample of movies and books but it isharder to find systems that provide othertypes of improvements that might be moreapplicable in a digital library environment.

As is often the case, research has beeninfluenced strongly by the data availableand movie recommender systems are apopular research vehicle because moviedatabases are freely available on theWeb. Some researchers have access tomore interesting datasets, but these areoften proprietary, which makes researchresults difficult to evaluate and repro-duce, and which raises a barrier to entryby new researchers. Workshop partici-pants agreed that there is a strong needfor a more diverse set of generally-avail-able data resources for personalisationresearch, and encourage the fundingcommunity to consider this in theirfunding decisions.

Evaluation is a related problem. Most ofthe systems discussed in the workshop

were evaluated in some manner, but fewof the evaluations could be called rig-orous. It might be clear how to evaluatea movie recommender system, but it isless clear how to evaluate more complexsystems. Researchers in this area need tobegin considering a broad approach toevaluation that embraces not just quanti-tative evaluation, but also methods andtools from disciplines such as sociology.While recognising the importance ofevaluation, we worry that an overem-phasis on evaluation will stifle new ideasin this fledgling field. Finding the rightbalance between creative developmentof new ideas and scientific evaluationremains an issue.

Finally, prior research on personalisationand recommender systems has focusedon relatively short periods of time.Systems are beginning to be deployedthat are intended for daily use over longperiods of time (two were discussed inthis workshop), but little is known abouthow such systems and their users mightevolve over time. It is important to beginstudying long-term personalisationissues, for example following a partic-ular group of users over a several yearperiod. Workshop participants felt thatfunding bodies need to give greaterattention to longer-term projects.

The complete workshop proceedings areavailable at http://www.ercim.org/ publi-cation/ws-proceedings/DelNoe02/.

Please contact: Alan Smeaton — Dublin City University, IrelandE-mail: [email protected]

Jamie Callan — Carnegie Mellon University,USAE-mail: [email protected]

The Joint DELOS-NSF Workshop on Personalisation and Recommender Systems inDigital Libraries held at Dublin City University June 18-20, 2001, brought together 57researchers and practitioners from 14 countries to discuss the development ofpersonalisation and recommender systems and techniques, particularly as they applyto digital libraries.

Joint DELOS-NSF Workshop on Personalisation and Recommender Systems in Digital Librariesby Alan Smeaton and Jamie Callan


EVENTS

Cross-Language Evaluation Forum pro-vides an infrastructure for the testing andtuning of mono- and cross-languageinformation retrieval systems workingwith European languages. The mainfocus of the activity is to stimulate thedevelopment of multilingual informationretrieval systems, ie systems capable ofmatching a query in one language againstdocument collections in many languages.

As a result of the success of theCLEF2000 campaign, it was decided toextend the test suite provided for thisyear’s participants. In particular, the mul-tilingual corpus was increased to nearlyone million documents in six languagesinstead of four languages: Dutch andSpanish were added to the English,French, German and Italian collections.The main task of CLEF2001 thusrequired participants to search a docu-ment collection in five languages,extracting their queries from a topic setoffered in twelve different languages(nine European and three Asian lan-guages). The same topic set could be usedin the bilingual retrieval task to searcheither a Dutch or an English target collec-tion. Tasks testing monolingual, domain-specific, and interactive text retrievalsystems were also offered. Thirty fourgroups participated in one or more of thetracks – fourteen more than in 2000.

The results of the CLEF2001 experi-ments were presented in a two-dayWorkshop held in Darmstadt at thebeginning of September. The Workshopwas attended by nearly fifty researchersand system developers from bothacademia and industry.

In the opening session, one of the orga-nizers, Martin Braschler of EurospiderInformation Technology, Zürich, gave anoverview of the campaign and sum-marised the main trends that emerged thisyear. All the traditional approaches to thecross-language task had been employed

but many innovational strategies werealso tested. It was notable that manygroups had built on results reported inCLEF 2000. The rest of the first day wasdedicated to descriptions of the experi-ments and discussions on a number of thekey issues affecting cross-languagesystems. These included the problem ofmerging the results of a search over anumber of document bases in differentlanguages in a meaningful way, and acomparison between methods used forindexing texts in multiple languages.

On the second day, ideas for new activi-ties were presented. Julio Gonzalo(UNED, Madrid) and Douglas Oard(University of Maryland) first reportedthe results of the experimental interac-tive track that was focussed on the enduser and the document selection task.The aim was to investigate how theresults of a cross-language system canbest be presented to users who have noknowledge of the target language(s) inorder to facilitate their selection of thosedocuments that do contain relevantinformation and merit further examina-tion. Marcello Federico of itc-IRST,Trento, then illustrated a proposal for thedesign of a track for a preliminary testingof systems for cross-language spokendocument retrieval, thus moving CLEFtowards an investigation of multilingualmultimedia systems.

In the following session, Donna Harmanfrom the US National Institute forStandards and Technology (NIST) andNoriko Kando (National Institute ofInformatics, Japan) reported on theactivities of the TREC and NTCIR eval-uation campaigns for cross-languagesystems (for Arabic and Asian lan-guages, respectively) . In their next cam-paign, NTCIR intends to adopt the CLEFmodel for evaluating multingualretrieval systems operating on Chinese,Japanese, Korean and English. Thesession concluded with an invited talk by

Ellen Voorhees (NIST) on the philos-ophy of IR evaluation in which she out-lined the methodology that has beenstudied for TREC and adopted by CLEF.In the afternoon, the criteria adopted byCLEF to create the topic sets andproduce the relevance assessments weredescribed by Christa Womser-Hacker(Univ. Hildesheim, Germany) andDjoerd Hiemstra (Univ. Twente,Netherlands).

In the final session of the Workshop, thepreliminary programme for CLEF 2002was outlined. While the multilingual taskwill remain unaltered, the bilingual taskhas been considerably strengthened.Groups will be able to test their systemschoosing from target document collec-tions in at least five European languagesand perhaps more. Investigations arecurrently underway as to the feasibilityof adding both Finnish and Swedish.Monolingual system testing will againbe offered and the interactive track willbe extended to become an official CLEFtask in 2002.

Copies of the Workshop presentationsare available on-line on the CLEF Website. The proceedings will be publishedfor the second year running by Springerin their Lecture Notes for ComputerScience series. CLEF is a collaborationbetween the US National Institute forStandards and Technology and theEuropean Commission. CLEF2000 and2001 were supported by the DELOSNetwork of Excellence for DigitalLibraries. CLEF 2002 will be fundedindependently under the 5th FrameworkIST programme of the Commission.

Link: http://www.clef-campaign.org/

Please contact:Carol Peters – IEI-CNRCLEF CoordinatorTel: +39 050 315 2897E-mail: [email protected]

The results of the second evaluation campaign of the Cross-Language EvaluationForum were presented at a Workshop held on 3-4 September, immediately beforeECDL2001 in Darmstadt, Germany.

CLEF2001: Workshop of the Cross-Language Evaluation Forumby Carol Peters


EVENTS

CALL FOR PARTICIPATION

MMM 2001 — 8th InternationalConference on MultimediaModeling

Amsterdam, 5-7 November 2001

MMM 2001 will continue theMultimedia Modeling series’ explo-ration of models and abstractions formultimedia creation, representation,storage, delivery, processing, presenta-tion and final interface. This year’s fociwill be on authoring and creation, stan-dards, intelligent delivery and presenta-tion, and the recent (re-)introduction inthe field of semantic models for multi-media and for the Web.

Topics concentrate on mulitimediaauthoring; aspects of information pre-sentation; multimedia standards andmodels; semantic models for multimediaand for the Web; multimedia modelingof real world scenarios; model-basedgraphics, video, vision and virtualreality; formal methods and multimediainformation; Multimedia databases.

More information:http://www.cwi.nl/conferences/MMM01/

SPONSORED BY ERCIM

CONCUR 2002 — 13thInternational Conference on Concurrency Theory

Brno, Czech Republic, 20-23 August 2002

The purpose of the CONCUR confer-ences is to bring together researchers,developers and students in order toadvance the theory of concurrency, andpromote its applications. Interest in thistopic is continuously growing, as a con-sequence of the importance and ubiquityof concurrent systems and their applica-tions, and of the scientific relevance oftheir foundations. Submissions aresolicited in all areas of semantics, logicsand verification techniques for concur-rent systems.

Principal topics include (but are notlimited to) concurrency related aspectsof: models of computation and semantic

domains, process algebras, Petri nets,event structures, real-time systems,hybrid systems, decidability, model-checking, verification techniques, refine-ment techniques, term and graphrewriting, distributed programming,logic constraint programming, object-oriented program-ming, typing systemsand algorithms, case studies, tools andenvironments for programming and veri-fication.

More information:http://www.fi.muni.cz/concur2002

SPONSORED BY ERCIM

ISSTA — InternationalSymposium on SoftwareTesting and Analysis

Rome, 22-24 July 2002

ISSTA is a leading research conferencein software testing and analysis, bringingtogether academics, industrialresearchers, and practitioners toexchange new ideas, problems, andexperiences. The ISSTA programmewill include research papers, panels, andinvited presentations.

Call for PapersAuthors are invited to submit papersdescribing original research in testing oranalysis of computer software. Papersdescribing theoretical or empiricalresearch, new techniques and tools, andin-depth case studies of software testingand analysis methods and tools arewelcome.

As ISSTA 2002 will be colocated withthe Workshop on Software Performance,papers on software testing, analysis, andverification techniques targeting soft-ware peformance issues are particularlysought.

More information:http://www.iei.pi.cnr.it/ISSTA2002/

ERCIM News is the magazine of ERCIM.Published quarterly, the newsletter reports onjoint actions of the ERCIM partners, and aimsto reflect the contribution made by ERCIM tothe European Community in InformationTechnology. Through short articles and newsitems, it provides a forum for the exchange ofinformation between the institutes and alsowith the wider scientific community. ERCIMNews has a circulation of 7000 copies.

ERCIM News online edition is available athttp://www.ercim.org/publication/ERCIM_News/

ERCIM News is published by ERCIM EEIG, BP 93, F-06902 Sophia-Antipolis CedexTel: +33 4 9238 5010, E-mail: [email protected] 0926-4981

Director: Bernard LarrouturouCentral Editor:

Peter KunzE-mail: [email protected]: +33 1 3963 5040

Local Editors:CLRC: Martin Prime

E-mail: [email protected]: +44 1235 44 6555

CRCIM: Michal HaindlE-mail: [email protected]: +420 2 6605 2350

CWI: Henk Nieland E-mail: [email protected]: +31 20 592 4092

CNR: Carol Peters E-mail: [email protected]: +39 050 315 2897

FORTH: Constantine StephanidisE-mail: [email protected]: +30 81 39 17 41

FhG: Dietrich StobikE-mail: [email protected]: +49 2241 14 2509

INRIA: Bernard HidoineE-mail: [email protected]: +33 1 3963 5484

SARIT: Harry RudinTel: +41 1 720 3707E-mail: [email protected]

SICS: Kersti HedmanE-mail: [email protected]: +46 8633 1508

SRCIM: Gabriela Andrejkova E-mail: [email protected]: +421 95 622 1128

SZTAKI: Erzsébet Csuhaj-VarjúE-mail: [email protected]: +36 1 209 6990

TCD: Carol O’SullivanE-mail: [email protected]: +353 1 60 812 20

VTT: Pia-Maria Linden-Linna E-mail: [email protected]: +358 9 456 4501

Free subscriptionYou can subscribe to ERCIM News free ofcharge by: • sending e-mail to your local editor• contacting the ERCIM office (see address

above) • filling out the form at the ERCIM website at

http://www.ercim.org/


IN BRIEF

CALL FOR PAPERS

Pervasive 2002 — International Conference on Pervasive Computing

Zurich, Switzerland, 26-28, 2002August

The objective of the InternationalConference on Pervasive Computingwill be to present, discuss, and explorelatest technical developments in theemerging field of pervasive computingas well as potential future directions andissues. Leaving aside low-level hard-ware and transmission technology topicsas well as peripheral socio-economicaspects, the conference will focus ontechnical infrastructure and applicationissues.

Relentless progress in basic InformationTechnologies (IT) - microprocessors,memory chips, integrated sensors,storage devices, and wireless communi-cation systems - continues to bring eversmaller, lighter, and faster systems-on-a-chip. As a result, beyond large hosts,PCs, and laptop computers, IT systemsare now invading every aspect of life tothe point that they are disappearinginside all sorts of appliances or can beworn unobtrusively as part of clothingand jewelry. This explosive spread of ITcan be compared to the spread of electricmotors over the past century butpromises to revolutionize life muchmore profoundly than elevators or elec-tric car windows ever did. At the sametime, pervasive computing poses pro-found challenges including technical,safety, social, legal, political, and eco-nomic issues.

The objective of this conference will beto present, discuss, and explore latesttechnical developments in the emergingfield of pervasive computing as well aspotential future directions and issues.Leaving aside low-level hardware andtransmission technology topics as wellas peripheral socio-economic aspects,the conference will focus on technicalinfrastructure and application issues. Itwill include presentations, panel discus-sions, poster sessions, and demos on sub-jects like:

• System architectures and platformsfor pervasive computing

• Middleware and pervasive computinginfrastructures

• Mobile, wireless, and wearabletechnologies

• Innovative small computing andintelligent devices

• Emerging applications and mobilebusiness issues

• Scenarios for information appliances • Service discovery protocols • Content distribution and delivery • User interfaces for invisible and

embedded computing • Context awareness • Security and privacy issues.

More information:http://www.pervasive2002.org/

SPONSORED BY ERCIM

STACS — 19th InternationalSymposium on TheoreticalAspects of Computer Science

Antibes Juan-les-Pins, France,March 14-16 2002

Topics presented at STACS include (butare not limited to): • Algorithms and data structures,

including: parallel and distributedalgorithms, computational geometry,cryptography, algorithmic learningtheory

• Automata and formal languages • Computational and structural

complexity • Logic in computer science, including:

semantics, specification, andverification of programs, rewritingand deduction

• Current challenges, for example:theory, models, and algorithms forbiological computing, quantumcomputing, mobile and netcomputing.

More information:http://www-sop.inria.fr/stacs2002/

INRIA — Claude Castelluccia,reseach fellow at INRIA is the winner ofthe 2001 Communication & SystemsAward. Intended to reward youngresearchers who bring significant inno-vations to important applications, theCS 2001 scientific price was awarded toClaude Castelluccia for his work onmobility in the future Internet protocolIPV6. Detailed information athttp://www.inrialpes.fr/planete/

INRIA — Olivier Faugeras is recipientof the ‘2001 World technology Award’in the category ‘InformationTechnology - Software’. The Awardrecognises the programmes and capabil-ities of which information technologyhardware makes use. It includes tradi-tional software applications using inno-vative or enhanced techniques, as wellas more comprehensive subjects such asvirtual reality or artificial intelligence.http://www-sop.inria.fr/robotvis/robotvis-eng.html

CLRC — Professor John Wood suc-ceeded Gordon Walker as CCLRC’s

Chief Executive.He took up hispost on 1 April2001 and hisappointment isfor four years.John Wood wasappointed Pro-fessor ofMaterials Engin-

eering and Head of Department at theUniversity of Nottingham in 1989 andsubsequently became Dean ofEngineering in 1998. He is Chairman ofthe Office of Science and Technology’sForesight Panel on Materials, havingbeen appointed in 1997. ProfessorWood has held a number of director-ships and consultancies with industryand has acted as an adviser on materialsissues to governments. He is a Fellow ofthe Royal Academy of Engineering andhas been awarded both the Grunfeld andIvor Jenkins Prizes of the Institute ofMaterials. Professor Wood received theCitizen of Honour of Cluj-Napaca inRomania for his ‘help in restructuringmaterials engineering education inRomania’.

Fraunhofer GesellschaftSchloß Birlinghoven, D-53754 Sankt Augustin, GermanyTel: +49 2241 14 0, Fax: +49 2241 14 2889http://www.fhg.de/

Consiglio Nazionale delle Ricerche, IEI-CNRArea della Ricerca CNR di Pisa, Via G. Moruzzi 1, 56124 Pisa, ItalyTel: +39 050 315 2878, Fax: +39 050 315 2810http://www.iei.pi.cnr.it/

Centrum voor Wiskunde en InformaticaKruislaan 413, NL-1098 SJ Amsterdam, The NetherlandsTel: +31 20 592 9333, Fax: +31 20 592 4199http://www.cwi.nl/

Institut National de Recherche en Informatique et en AutomatiqueB.P. 105, F-78153 Le Chesnay, FranceTel: +33 1 3963 5511, Fax: +33 1 3963 5330http://www.inria.fr/

Central Laboratory of the Research CouncilsRutherford Appleton Laboratory, Chilton, Didcot, Oxfordshire OX11 0QX, United KingdomTel: +44 1235 82 1900, Fax: +44 1235 44 5385http://www.cclrc.ac.uk/

Swedish Institute of Computer ScienceBox 1263, SE-164 29 Kista, SwedenTel: +46 8 633 1500, Fax: +46 8 751 72 30http://www.sics.se/

Technical Research Centre of FinlandVTT Information Technology, P.O. Box 1200, FIN-02044 VTT, FinlandTel:+358 9 456 6041, Fax :+358 9 456 6027http://www.vtt.fi/

Foundation for Research and Technology – HellasInstitute of Computer Science, P.O. Box 1385, GR-71110 Heraklion, Crete, GreeceTel: +30 81 39 16 00, Fax: +30 81 39 16 01http://www.ics.forth.gr/FORTH

Czech Research Consortium for Informatics and MathematicsFI MU, Botanicka 68a, CZ-602 00 Brno, Czech RepublicTel: +420 2 688 4669, Fax: +420 2 688 4903http://www.utia.cas.cz/CRCIM/home.html

Swiss Association for Research in Information TechnologyDept. Informatik, ETH-Zentrum, CH-8092 Zürich, SwitzerlandTel: +41 1 632 72 41, Fax: +41 1 632 11 72http://www.sarit.ch/

Slovak Research Consortium for Informatics and MathematicsDept.of Computer Science, Comenius University, Mlynska Dolina M SK-84248 Bratislava, SlovakiaTel: +421 7 726635, Fax: +421 7 727041http://www.srcim.sk

Magyar Tudományos Akadémia – Számítástechnikai és Automatizálási Kutató IntézeteP.O. Box 63, H-1518 Budapest, HungaryTel: +36 1 4665644, Fax: + 36 1 466 7503http://www.sztaki.hu/

ERCIM – The European Research Consortium for Informatics and Mathematics is an organisationdedicated to the advancement of European research and development, in information technologyand applied mathematics. Its national member institutions aim to foster collaborative work withinthe European research community and to increase co-operation with European industry.

Trinity College, Department of Computer Science,Dublin 2, IrelandTel: +353 1 608 1765, Fax: 353 1 677 2204http://www.cs.tcd.ie/ERCIM