˚3(&,$/ 6+(0(: ˘2%,/( C20376,1* - ERCIM NewsA/52 ,1 6+,5 ,557(: Keynote Strategies of the European ICST Public Research Organisations towards Horizon2020 by Domenico Laforenza Joint

Also in this issue:

KeynoteStrategies of the European ICST

Public Research Organisations

towards Horizon2020

by Domenico Laforenza

Joint ERCIM ActionsUniversity of Southampton joins

ERCIM

Research and InnovationVMC: A Tool for the Analysis of

Variability in Software Product Lines

ERCIM NEWSwww.ercim.eu

Number 93, April 2013

Special theme:

Mobile Computing

ERCIM NEWS 93 April 2013

ERCIM News is the magazine of ERCIM. Published quarterly, it

reports on joint actions of the ERCIM partners, and aims to

reflect the contribution made by ERCIM to the European

Community in Information Technology and Applied

Mathematics. Through short articles and news items, it provides

a forum for the exchange of information between the institutes

and also with the wider scientific community. This issue has a

circulation of about 8,500 copies..

ERCIM News is published by ERCIM EEIG

BP 93, F-06902 Sophia Antipolis Cedex, France

Tel: +33 4 9238 5010, E-mail: [email protected]

Director: Jérôme Chailloux

ISSN 0926-4981

Editorial Board:

Central editor:

Peter Kunz, ERCIM office ([email protected])

Local Editors:

Austria: Erwin Schoitsch, ([email protected])

Belgium:Benoît Michel ([email protected])

Cyprus: George Papadopoulos ([email protected])

Czech Republic:Michal Haindl ([email protected])

France: Thierry Priol ([email protected])

Germany: Michael Krapp ([email protected])

Greece: Eleni Orphanoudakis ([email protected]),

Artemios Voyiatzis ([email protected])

Hungary: Erzsébet Csuhaj-Varjú ([email protected])

Italy: Carol Peters ([email protected])

Luxembourg: Thomas Tamisier ([email protected])

Norway: Truls Gjestland ([email protected])

Poland: Hung Son Nguyen ([email protected])

Portugal: Joaquim Jorge ([email protected])

Spain: Silvia Abrahão ([email protected])

Sweden: Kersti Hedman ([email protected])

Switzerland: Harry Rudin ([email protected])

The Netherlands: Annette Kik ([email protected])

United Kingdom: Martin Prime ([email protected])

W3C: Marie-Claire Forgue ([email protected])

Contributions

Contributions should be submitted to the local editor of your

country

Copyright Notice

All authors, as identified in each article, retain copyright of their

work

Advertising

For current advertising rates and conditions, see

http://ercim-news.ercim.eu/ or contact [email protected]

ERCIM News online edition

The online edition is published at http://ercim-news.ercim.eu/

Subscription

Subscribe to ERCIM News by sending email to

[email protected] or by filling out the form at the

ERCIM News website: http://ercim-news.ercim.eu/

Next issue

July 2013, Special theme: Intelligent Cars

Editorial Information

ERCIM NEWS 93 April 2013 3

Strategies of the European

ICST Public Research

Organisations towards

Horizon 2020

The scientific and professional societies in ICST in Europehave a key responsibility in helping to understand and toshape the digital future. They should collectively identify thevision, needs and priorities, and offer their expertise to theentire discipline and to society as a whole.

A round table on the “Role and Strategies of the EuropeanICST Public Research Organisations towards Horizon 2020”,was held in October 2012 in the framework of the ERCIMFall meetings (http://www.ercim.eu/fall-meetings-2012) atINRIA Sophia Antipolis. The panel addressed how Europeanpublic research organisations working in Information andCommunication Sciences and Technologies (ICST) can con-tribute to the success of Horizon 2020 (H2020), and in partic-ular to the strategic direction of the “Excellence in Science”agenda. Members of the panel were: Domenico Laforenza(CNR, ERCIM Vicepresident – panel moderator), KeithJeffery (STFC-RAL, ERCIM President), Michel Cosnard(CEO and Chairman, INRIA), Jan van Leeuwen (UtrechtUniversity, Chair of the European Forum for ICST), JosBaeten (CEO, CWI), Matthias Jarke (Chairman, ICT Group,Fraunhofer Gesellschaft).

The initial presentation addressed the new challenges intro-duced by H2020. After an overview of the objectives andstructure of H2020, focusing on the three pillars of the H2020strategy (Excellent Science, Industrial Leadership andSocietal Challenges), DG “Connect” (the new DirectorateGeneral for Communications Networks, Content andTechnology created by the EC in July 2012 in order tomanage the Digital Agenda for Europe) was presented.

The main question raised in the panel was: How canEuropean public ICST research organisations contribute toshaping the H2020 work programmes? In the past, eachorganisation interacted independently with the EC; however,the lack of a single voice has seriously impacted on theirability to be heard by EC decision makers. The developmentof common viewpoints and strategies for ICST in Europe and,whenever appropriate or needed, a common representation ofthese viewpoints and strategies at the international level arethe foundational principles of the European Forum forInformation and Communication Sciences and Technologies.EFICST (http://www.eficst.eu/) was established in November2011 by the joint action of seven leading organisations andsocieties in ICST in Europe: ACM Europe (ACM EuropeCouncil), European Association for Programming Languagesand Systems (EAPLS), European Association of SoftwareScience and Technology (EASST), European Association forTheoretical Computer Science (EATCS), EuropeanCoordinating Committee for Artificial Intelligence (ECCAI),European Research Consortium for Informatics andMathematics (ERCIM), and INFORIE (Informatics Europe).

The Forum is intended to be an open platform for cooperationamong the scientific ICT societies in Europe.

As reported in an EFICST white paper (Shaping the DigitalFuture of Europe) under preparation, the advances in infor-mation and communication science and technology continueto dramatically impact on our economies and our society.Computational thinking and virtualization are revolutionizingboth science and technology. Driven by the need for automa-tion and the exciting opportunities that are emerging, thedevelopment of ICST must be accompanied by a sound andfar-reaching vision of the way we will live, work, and do busi-ness in the years to come. This requires a permanent dialoguebetween ICST professionals and the many other stakeholdersinvolved, aimed at understanding where future innovationswill take us. In its strategy for 2013-2015, the EuropeanForum for ICST is defining a set of actions intended to meetthese challenges in view of the crucial role of ICST as a keydriver of innovation and change. The strategy focuses onthree areas: Software as the key enabling technology ofEurope’s digital future; ICT as the catalyst of industrial andsocietal innovation, and Informatics as a scientific disciplinein vocational and professional education.

It was observed in the panel that although software is world-wide considered a strategic key enabling technology, unfortu-nately “software” is not included in the list of the KeyEnabling Technologies (micro- and nano-electronics; pho-tonics; nanotechnologies; advanced materials; biotechnology;advanced manufacturing and processing) reported in officialEC H2020 documents. In order to raise awareness that thelack of recognition of the strategic importance of softwaretechnology will lead to a significant reduction in globalEuropean competitiveness, an interesting ISTAG report, titled“The Missing KET: Toward a Strategic Agenda for SoftwareTechnologies in Europe”, has been released. The objective ofthis report is to create a sense of urgency in the European soft-ware industry and awareness of software as the prime indus-trial differentiator and basis for innovation.

An additional topic discussed during the panel was how toreinforce relationships with the Knowledge and InnovationCommunities (KICs) of the European Institute for Innovationand Technology (EIT), and in particular with the EIT KICICT Labs, in order to stimulate innovation through a more rig-orous and dynamic link with higher education, research andbusiness.

The final message of the panel was that European digitalfuture is in the hands of many stakeholders, driven by dif-ferent goals and needs. It is important to achieve coordinatedthinking and actions so that society can benefit in the bestpossible way and not run the risk of missed opportunities orunwanted effects.

Domenico Laforenza

Domenico Laforenza

Director of the Institute for

Informatics and Telematics (IIT),

Italian National Research

Council (CNR), and ERCIM

Vice-President

Keynote


Contents

SPECIAL THEME

The special theme section “Mobile Computing” has beencoordinated by Edgar Weippl, SBA Research, AARIT, Austria andPietro Manzoni, Universitat Politècnica de Valéncia, SpaRCIM

8 Introduction to the Special Themeby Edgar Weippl and Pietro Manzoni

Invited articles10 Reality Mining at the Convergence of Cloud Computing

and Mobile Computingby Matthias Steinbauer, Ismail Khalil and Gabriele Kotsis

11 Positioning Terminals in Mobile Computing Networksby Francisco Barcelo-Arroyo, Israel Martin-Escalona andMarc Ciurana-Adell

12 DarkDroid - Exposing the Dark Side of Malicious MobileApplicationsby Engin Kirda

Computing services 14 Automatic Offloading of Mobile Applications Using

Evolutionary Algorithmsby Gianluigi Folino and Francesco Sergio Pisani

15 Scheduling Data Mining Applications in MobileComputing Environmentsby Carmela Comito, Deborah Falcone, Domenico Talia andPaolo Trunfio

17 On the Benefits of a Poly-Cultural SensorSetup:Controlling Embedded Sensors with a Smart Phoneby Rolf Adelsberger and Gerhard Tröster

18 SmartLab: Empowering Mobile Computing Researchthrough an Open Smartphone Cloudby Georgios Larkou, Panayiotis Andreou, AndreasKonstantinidis and Demetrios Zeinalipour-Yazti

19 Phone Accessories as an Interfaceby Mattias Jacobsson, Stina Nylander and Ylva Fernaeus

21 Boosting Performance of Wireless Networks withConcurrent Access by Smart Traffic Splitting by Gerard Hoekstra and Rob van der Mei

Platforms 22 CoMobility: A Mobile Platform for Transport Sharing

by Carlos E. Cuesta, Paloma Cáceres, Belén Vela and JoséMaría Cavero

23 Quality of Service Information System: Get to Know thePerformance of Your Mobile Network OperatorAnywhere-Anytimeby Katarzyna Wac

25 Mobile Service Platforms Based on OpportunisticComputing: The SCAMPI Projectby Marco Conti, Franca Delmastro and Andrea Passarella

KEYNOTE

3 Strategies of the European ICST PublicResearch Organisations towards Horizon 2020by Domenico Laforenza

JOINT ERCIM ACTIONS

6 Cor Baayen Award 2013 Call for Nominations

6 ERCIM Postdoc Fellowship Programe: LastRound of the Co-funded “ABCDE” Project

7 University of Southampton joins ERCIMby Inés Teresa-Palacio


27 TravelDashboard - a Framework for the Delivery of

Personalized Mobility Services to Urban Travellers

by Licia Capra, Pierre Chatel, Animesh Pathak and RobertoSpeicys Cardoso

28 APISENSE: Crowd-Sensing Made Easy

by Nicolas Haderer, Romain Rouvoy, Christophe Ribeiro andLionel Seinturier

Security and privacy

30 Mobilitics: Analyzing Privacy Leaks in Smartphones

Jagdish Prasad Achara, Franck Baudot, Claude Castelluccia,Geoffrey Delcroix and Vincent Roca

31 Privacy-Preserving Interest-Cast for Android Smartphones

by Gianpiero Costantino, Fabio Martinelli and Paolo Santi

32 Market-Based Security for Mobile Devices

by Gabriele Costa, Alessio Merlo and Luca Verderame

34 Revealing Social Links Between Owners of Wi-Fi Enabled

Smartphones

by Mathieu Cunche, Mohamed Ali Kaafar and Roksana Boreli

35 Secure Collaboration for Smartphones

by Abdessamad Imine and Michaël Rusinowitch

Localization

36 Indoor User Localization Using Mobile Devices

by Jonáš Ševčík

37 Airplace: Indoor Geolocation on Smartphones Through

WiFi Fingerprinting

by Christos Laoudias, Georgios Larkou, DemetriosZeinalipour-Yazti and Christos G. Panayiotou

39 Knowledge Representation and Management in Indoor

Mobile Environments

by Imad Afyouni, Cyril Ray and Christophe Claramunt

Applications

41 Mobile Real Time Applications for Enhancing Public

Transport User Experience - The MOVE-ME Project

by João Falcão e Cunha, Teresa Galvão and Jeremy Pitt

42 U-AirPoll: Mobile Distributed and Collaborative Air

Pollution Measurement

by Marino Linaje and Luis Miguel Dominguez-Peinado

43 ARGO Sentinel: The Mobile App for Reporting Oil

Spillages at Sea

by Massimo Martinelli, Davide Moroni and Ovidio Salvetti

45 Mobile Devices to Improve Breast Cancer Information

Management

by Damià Segrelles, Maite Giménez and Ignacio Blanquer

reSearch aNd INNoVaTIoN

This section features news about research

activities and innovative developments from

European research institutes

46 Secure and Privacy-Aware Mobile Identity

Management

by Fabio Martinelli

47 Social Electricity: When Awareness About

Electricity Becomes Social

by Andreas Kamilaris, George Taliadoros andAndreas Pitsillides

49 New Interaction Paradigms in Energy

Management

by Paulo Carreira and Alfredo Ferreira

50 VMC: A Tool for the Analysis of Variability

in Software Product Lines

by Maurice ter Beek, Stefania Gnesi andFranco Mazzanti

52 Wood Variety Recognition on Mobile Devices

by Pavel Vácha and Michal Haindl

53 The New SHIELD Architectural Framework

by Mariana Esposito, Andrea Fiaschetti,Francesco Flammini

54 A Biofeedback System for Self

Empowerment and Improved Quality of Life

by Johanna Mercurio

55 Software and Hardware-Intensive Activities

for Supporting Creative Learning

by Michail Giannakos and Letizia Jaccheri

eVeNTS

56 W3C at the Mobile World Congress 2013

by Marie-Claire Forgue

56 PROMISE Winter School 2013 on Bridging

between Information Retrieval and

Databases

by Nicola Ferro

57 Announcements

IN BrIef

59 CWI researcher Floor Sietsma youngest PhD

in the Netherlands

59 Warsaw Center of Mathematics and

Computer Science Established

Joint ERCIM Actions

Cor Baayen Award 2013

Call for Nominations

The Cor Baayen Award is awarded each year to a promising

young researcher in computer science and applied mathe-

matics. The award was created in 1995 to honour the first

ERCIM President. The award consists of a cheque for 5000

Euro together with an award certificate. The selected fellow

will be invited to the ERCIM meetings in autumn. A short

article about the winner, together with the list of all candi-

dates nominated, will be published in ERCIM News.

Conditions

Nominees must have carried out their work in one of the

“ERCIM countries”: Austria, Belgium, Cyprus, Czech

Republic, Finland, France, Germany, Greece, Hungary, Italy,

Luxembourg, Norway, Poland, Portugal, Spain, Sweden,

Switzerland, The Netherlands and the United Kingdom.

Nominees must have been awarded their PhD (or equivalent)

after 30 April 2010. A person can only be nominated once for

the Cor Baayen Award.

Submitting a nomination

Nominations should be made by a staff member of an

ERCIM member institute. Self nominations are not accepted.

They should concern nominees having performed their

research in any research institution from the country of the

nominating institution. Nominations must be submitted

online

Selection

The selection of the Cor Baayen award winner is the respon-

sibility of the ERCIM Human Capital Task Group, who will

consult expert opinion when reaching their decision.

Deadline

Deadline for the 2013 award nominations is 30 April 2013

Further Information:

http://www.ercim.eu/activity/cor-baayen-award

ERCIM Postdoc

Fellowship Programe:

Final Round of the

Co-funded “ABCDE” Project

The final round of the successful “ABCDE” programme co-

funded by the EC Marie Curie Actions is open. For this

round, the fellowships are of 12-month duration to be spent

in one ERCIM member institute. Topics cover most disci-

plines in Computer Science, Information Technology, and

Applied Mathematics and applicants can also propose their

own research agenda. The deadline for applications is 30

April 2013.

Conditions

Applicants must:

• have obtained a PhD degree during the last 8 years (prior

to the application deadline) or be in the last year of thesis

work with an outstanding academic record.

• be fluent in English.

• be discharged or get deferment from military service.

• have completed the PhD before starting the grant (proof

will be requested).

• start the fellowship before May 2014.

An opportunity for ERCIM members

Since the start of the programme in 2010, some 130 fellow-

ships have been co-funded by the European Commission’s

Marie-Curie Actions. Participating institutes appreciate the

high quality of the candidates and that part of the costs of

hosting a fellow is refunded. For this last round, co-funding

is still available for about 50 fellowships and all ERCIM

member institutes are entitled to benefit from this pro-

gramme.

More information and application form:

http://fellowship.ercim.eu/


The Cor Baayen Award is named after

the first president of ERCIM and the

ERCIM 'president d'honneur'. Cor

Baayen played an important role in

ERCIM’s foundation.

Cor Baayen was scientific director of the

Centrum Wiskunde & Informatica

(CWI) in the Netherlands, from 1980 to 1994. He joined the

institute in 1959 as a researcher of the department of pure

mathematics, where he became group leader in 1965. As

scientific director, Cor Baayen convinced the government to

include CWI in a program that stimulated Dutch computer

science research. He initiated or stimulated several new research

areas including cryptography, computer algebra and performance

analysis.

About Cor Baayen

In 2013, Inria is offering 55 government-funded post-doctoral

positions among others, each lasting about 12 or 24 months, for

holders of a PhD or other doctorate. Post-doctoral positions offer

the opportunity to pursue rewarding research work in the field of

computer science and mathematics in an internationally

recognized working environment with links to industry.

Applicants must have defended their thesis in 2012 or 2013. The

offered positions focus on the institute’s priority research areas.

However, high-quality applications to work on non-priority

subjects will also be considered. Positions are open to candidates

from all over the world. This campaign begun in January and

ends on 14 June.

For subjects and available positions, see:

http://www.inria.fr/en/institute/recruitment/offers/post-doctoral-

research-fellowships/campaign-2013

55 Post-Doctoral Positions at Inria in 2013


University of Southampton

joins ERCIM

by Inés Teresa-Palacio

The University of Southampton is one of the top 15 researchuniversities in the UK and is ranked among the top 1% ofuniversities in the world (2012 QS World UniversityRankings). It is a founder member of the prestigious RussellGroup of leading research universities in the UK and is partof the Worldwide Universities Network, engaging in collabo-ration and exchange agreements with some of the world’sleading universities. The University of Southampton hasaround 22,000 students of which over 5,000 are internationalfrom more than 130 different countries. It has eight facultiescovering a wide range of subject areas: from humanities, artand social sciences, to biology, chemistry, electronics andcomputer science, engineering, health sciences, medicine,mathematics, oceanography, environmental sciences, opto-electronics, physics and astronomy.

Electronics and Computer Science (ECS), which is part ofthe Faculty of Physical Sciences and Engineering, is one ofthe world’s largest and most successful electronics and com-puter science departments, with more than 60 years of devel-opment in leading edge technology. It has a global reputationfor computer science, agent technologies, web science, bio-metrics, open access, digital libraries, nanofabrication, wire-less communications, energy harvesting and pervasive com-puting. ECS contains six research groups, an IT InnovationCentre, 100 academic staff, 120 research staff and 270 PhDstudents. ECS also plays a major role in two of theUniversity’s Doctoral Training Centres (DTCs) – theInstitute for Complex Systems Simulation and Web Science.Both DTCs offer four-year postgraduate programmes inthese specialist areas enabling students to work effectivelyacross disciplines and helping them to develop the skills theyneed to address some of society’s biggest problems.

Southampton is ranked first in the UK for Electronics andElectrical Engineering, and we are consistently placed in thetop 10 places in the UK for Computer Science and IT. ECSresearchers achieved exceptional success in the 2008Research Assessment Exercise (RAE). Computer Sciencewas ranked joint second in the UK for the quality of itsresearch, with 85 per cent of its research work receivingeither the top 4* rating (defined as ‘world leading’) or the 3*rating (‘internationally excellent’). In Electronics andElectrical Engineering (in which ECS was assessed jointlywith the University's Optoelectronics Research Centre), ECS(and the ORC) came second in the ‘medals’ tables, with 42researchers rated as achieving research of either world-leading or internationally excellent quality. We are also oneof only a handful of universities to receive the prestigioustitle Regius Professor, an honour bestowed by The Queen.The rare professorship marks our excellence in the field ofComputer Science and reflects our exceptionally highquality of teaching and research.

ECS benefits from state-of-the-art facilities, including theMountbatten Complex (one of the world’s leading clean-

room laboratory complexes for materials and device researchin diverse fields ranging from electronics to photonics andbio-nanotechnology) and the High Voltage Laboratory (oneof only a handful of similar facilities in Europe that providethe high-level testing and research facilities required by elec-tricity supply companies).

“I am very excited that the University of Southampton isjoining the ERCIM network,” says Professor Dame WendyHall who is representing the university in the ERCIMGeneral Assembly. “Over the years we have hugely bene-fitted from the international networks that we have been partof, including the World Wide University Network and theWeb Science Trust international network of research labora-tories. More recently we have formed a strategic partnershipwith the Fraunhofer-Gesellschaf through the SoFWIReD(Southampton Fraunhofer Web Science, Internet Research& Development) project. It is a natural step for us to join theERCIM network, which represents some of the best com-puting research laboratories in Europe and enables thedevelopment of networks of excellence to tackle the grandchallenges in our subject today, which require big teamsworking collaboratively across disciplinary and geograph-ical boundaries”.

Links: http://www.southampton.ac.uk/http://www.ecs.soton.ac.uk/

Please contact:Professor Dame Wendy Hall, ERCIM General AssemblyrepresentativeLeslie Carr, ERCIM General Assembly substituteUniversity of Southampton, UKE-mail: [email protected], [email protected]

Mountbatten Complex

A few years ago, the IT industry forecast thatconsumer demand for data would surpass thatof the voice market. At the time, few couldconceptualize such an outlook. These days thedata market is at least 70% larger than themobile market, and still growing. Mobile tech-nology has advanced in leaps and bounds overrecent years, such that experts predict that,within the next few years, mobile computingwill be strictly bound to cloud computing.Mobile cloud computing is set to impact andtransform the mobile communication land-scape and the whole computing infrastructure.Computing offloading, for example, is one ofthe main features of mobile computing toimprove the battery life of mobile devices andto improve the performance of applications.However, there are many associated issues tosolve, including efficient and dynamicoffloading in a variable environment.

Mobile users may face certain problems, suchas congestion owing to wireless bandwidthlimitations, network disconnection, and signalattenuation caused by mobile users’ mobility.Although many researchers have proposedsolutions to optimally and efficiently allocatebandwidth, bandwidth limitation is still amajor concern because the number of mobileand cloud users is increasing dramatically. 4Gnetwork and Femtocell are emerging as prom-ising technologies that are revolutionizingbandwidth optimization, helping to overcomethe traditional limitations. Efficient networkaccess management not only improves linkperformance for mobile users but also opti-mizes bandwidth usage. Cognitive radio islikely to offer a solution to achieve wirelessaccess.

Today’s mobile phones harness the power of acomputer, thus making sensitive informationmore widely available (eg information aboutuser behaviour, GPS position, and personaldata). Mobile phones are often fully integratedinto social applications such as social net-works, e-mail clients, messengers or data-har-vesting analytics scripts on websites, enablingnot only a holistic analysis of an individual,but also paving the way for new attack vectorsoperating against individuals and the compa-nies they work for. Malware targeting mobileclients, especially those on the more openAndroid-system, have become very common.


Special Theme: Mobile Computing

Introduction to the

Special Theme

by Edgar Weippl and Pietro Manzoni

The articles presented in this issue reflect the many faces of our special theme, “Mobile

computing”, and address topics including: Mobile service platforms and new ways of

establishing networks, privacy concerns and security challenges, location services and

collaborative data capture.

The invited article by Engin Kirda summarizes a research project that focuses on

Android malware detection. Matthias Steinbauer et al discuss effects of the conver-

gence of cloud and mobile computing. Francisco Barcelo-Arroyo et al present results

related to the important topic of providing indoor localization through the combined use

of communication networks.

Different kinds of computing services form the focus of the first six articles:

Folino and Pisani describe a framework for generating decision tree-based models that use

evolutionary algorithms to take automatic decisions regarding the offloading of mobile appli-

cations into a cloud environment. Larkou et al present SmartLab, an open IaaS cloud of

smartphones that improves the efficiency of systems-oriented mobile computing research.

Mobile device networks also offer enormous potential for data mining. Comito et al have

defined a distributed architecture with an energy-aware scheduling strategy that assigns

data mining tasks in a peer-to-peer network of mobile devices.

Adelsberger and Tröster present a means of using smartphones to synchronize and con-

trol sensors wirelessly, more effectively, and in a more energy-efficient way than is usu-

ally the case for wireless synchronizing of multiple data streams.

With mobile Internet on the rise, the demand for wireless networks is growing at an

unprecedented pace. Hoekstra and van der Mei describe new methods, developed in the

Netherlands, that use smart algorithms to split traffic over the numerous overlapping net-

works in the country, thereby increasing wireless speed.

Moving towards the viewpoint of the user, Jacobsson et al describe a design concept for

changing setups and user interface styles of smartphones by physically attaching phone

shells or accessories such as jewellery or headphones to the device, enabling quick shifts

between, for example, business and leisure modes.

The user perspective is also present in different mobile service platforms. The near ubiq-

uity of smartphones makes them ideally suited for transport planning services. Capra et

al introduce the TravelDashboard project, which will allow customizable trip planning

according to personal preferences, also incorporating user-generated content with infor-

mation on issues such as crowdedness of buses. In the same vein, Cuesta et al’s

CoMobility platform also allows travel planning but integrates carpooling with public

transportation. Wac focuses on a Quality of Service Information System and reports on a

mobile application that uses measurement data provided by mobile users to predict a net-

work’s expected performance.

Conti et al discuss opportunistic computing, a self-organizing dynamic networking para-

digm that combines pervasive environmental network devices with mobile devices to

allow communication and services. They present the CAMEO middleware platform,

which focuses on the management and elaboration of context information in such oppor-

tunistic networks.

The pervasiveness of mobile devices has raised interest in collecting their sensor data via

“crowd sensing”. Haderer et al address the needs of various research communities with

their APISENSE platform, which gives researchers an online environment in which they

can set up an experiment without in-depth technical knowledge while ensuring the pri-

vacy and security of the participants collecting the data.

Privacy and security are, indeed, important issues as consumers and companies enter the

world of mobile computing with unprecedented enthusiasm, and a number of contribu-

tions in this issue are dedicated to them.

Articles in the special theme

From localization and transportation services

to sensor control and environmental moni-

toring, we hope you enjoy the contributions in

this issue, which show the wide applicability

of mobile computing and give us a taste of

things to come.

Regarding the future, we can see that the area

of mobile computing, within the wider area of

the Information and Communications

Technologies (ICT) is a very promising and

strategic discipline. The European

Commission, through its new program

"Horizon 2020", has presented an €80 billion

package for research and innovation funding,

as part of the process to create sustainable

growth and new jobs in Europe.

In particular, and strongly related to mobile

computing, ICT in Horizon 2020 will be a

crucial actor and will support the develop-

ment of solutions for industrial leadership, by

supporting the development of the next gener-

ation of computing thanks to advanced com-

puting systems and technologies supported by

enhanced network infrastructures, technolo-

gies and services for the future Internet,

including content technologies and informa-

tion management.

The final objective is to provide answers to

societal challenges such as health, demo-

graphic change and wellbeing (eg, e-health,

self management of health, improved diag-

nostics, improved surveillance, health data

collection, active ageing, assisted living);

secure, clean and efficient energy (eg, Smart

cities; Energy efficient buildings; smart elec-

tricity grids; smart metering); smart, green

and integrated transport (eg, smart transport

equipment, infrastructures and services; inno-

vative transport management systems; safety

aspects); climate action, resource efficiency

and raw materials (eg, ICT for increased

resource efficiency; earth observation and

monitoring); and finally inclusive, innovative

and secure societies (eg, digital inclusion;

social innovation platforms; e-government

services; e-skills and e-learning; e-culture;

cyber security; ensuring privacy and protec-

tion of human rights on-line).

Please contact: Edgar Weippl

SBA Research/AARIT, Austria

E-mail: [email protected]

Pietro Manzoni

Universitat Politecnica de Valencia, Spain

E-mail: [email protected]


One reason for the widespread use of mobile devices is, of course, the wide availability of

Wi-Fi networks. Transmitting data over unsecured networks has well-known risks, but

Cunche et al. go further and show how the automatic search for networks enabled on

most smartphones can be used to not only fingerprint individual devices, but also to iden-

tify social links between their owners.

Networked calendars and other collaborative applications are very popular but pose con-

siderable security challenges. Imine and Rusinowitch have developed a decentralized

and secure shared calendar that is independent of third-party servers, instead allowing

users to share their calendar events in dynamic groups in ad-hoc networks.

Constantino et al present an implementation of the cryptographic FairPlay framework for

Android smartphones. It protects users’ privacy in opportunistic networks by ensuring that

information is exchanged with other users’ devices via Bluetooth only if they have matching

interest profiles, which can be determined without sending sensitive information in plaintext.

Achara et al have examined the information that can be gained from smartphones to edu-

cate users about risks. The Mobilitics project investigates both the Android and iOS oper-

ating systems and apps for these platforms for potential privacy leaks and has found that

many apps access information not necessary for their operation.

One way of making apps more secure can be found in Costa et al’s proposal for a secu-

rity-enabled app marketplace, where applications are analyzed to ensure that they comply

with the security policy and can be installed without affecting the device’s security con-

figuration.

While GPS-based localization offers a multitude of location-based services (eg, naviga-

tion) in outdoor environments, indoor mobile environments require different approaches,

some of which are presented in the next three contributions.

Laoudias et al’s Airplace indoor positioning system uses a radiomap built from received

signal strength (RSS) fingerprints. This radiomap is transmitted to the user’s Android

smartphone upon entering the building and allows the device to determine its position

based on the signal strength it receives from surrounding wireless access points without

revealing its personal state.

RSS fingerprinting is also one of the three techniques that are combined in the user

localization method presented by Ševčík, the others being dead reckoning and sequential

Monte Carlo filtering. The prototype displays the position of the user on a floor plan.

Further developments are planned to allow navigation to a selected point on the map and

augmented reality.

Afyouni et al focus on the representation and management of spatial data and location-

dependent query processing required for the development of efficient and flexible con-

text-aware indoor navigation systems. They developed a hierarchical data model of an

indoor environment and algorithms that utilize it to process location-dependent queries

continuously and effectively.

Finally, we have four contributions dedicated to applications. From bus travel to oil spills,

they present four innovative uses of mobile computing combined with crowdsourcing.

While Falcao e Cunha et al present their MOVE-ME multimodal travel planning project

with user contribution, Segrelles et al’s TRENCADIS allows the secure sharing, organi-

zation and searching of medical images on mobile devices and has been prototyped for

breast cancer diagnosis and treatment.

The papers by Trigueros et al. and Martinelli et al both present applications for envi-

ronmental monitoring with a strong focus on crowdsourcing: Trigueros and Peinado’s U-

AirPoll is a novel approach to distributed and collaborative air quality measurement,

while Martinelli et al.’s ARGO Sentinel allows volunteers at sea to immediately report

sighted oil spillages.

Over the last decade, we have witnessedan increasing usage of mobile devicesfor capturing, analysing, and predictinghuman behaviour in everyday activities.

Most modern mobile devices areequipped with a plethora of sensors thatcapture every aspect of the user’s phys-ical context represented by attributessuch as time, location, light, sound,weather, temperature, or even physio-logical state. Combined with socialcomputing applications such as blogs,email, instant messaging, social net-working (Facebook, Twitter, Linked,

Google+), Wikis, and social book-marking, computers are able to capturethe social context of the users in terms ofinterpersonal relationships and roles.

These physical and social contextsdescribe the existence of a relationshipbetween two entities. Although thestructure and nature of such relation-ships can be interpreted as a semanticnetwork that can be used as the basis forunderstanding the meaning of an inter-action, it fails to reflect the dynamics ofa relationship over time. The dynamicpatterns of interaction are essential in

formalized procedures such as work-flows, recurring sequences of actions(such as routine tasks), types of motion(such as walking, running andstanding), tasks (such as having lunch,washing dishes, and driving) and goals(for example, socializing, hiring,selling, keeping fit or simply havingfun) [1].

Reality mining is the collection andanalysis of machine-sensed environ-mental data pertaining to human socialbehaviour, with the goal of identifyingpredictable patterns of behaviour,

including how computers can learn toextract social clues from social systems[2].

In reality mining, techniques inheritedfrom data mining and data analysis areapplied to data generated by humaninteractions, ie phone call logs, e-mailmessages, Bluetooth proximity logs andcell tower logs, etc.

Reality mining analyses traces left bymobile devices, social networks andcommunication systems in the envi-ronment to extract social clues about

the social network from which theyoriginate.

Such clues can be used to detect com-munication bottlenecks in organizations(such as when a single individual con-nects departments) and behaviouralstereotypes applying to each member ofa work group in order to determine theroles of individuals within groups (whois a leader in which group, who is anexpert in which group).

In our research we focus on cloud-basedreality mining. We are aiming to pro-

vide tools and techniquesthat allow work groupsto analyse their socialnetwork in near realtime. Our objective is toimprove group perform-ance by providing aholistic overview of thegroup, its activities, situ-ations and goals, in orderto improve the group’soverall performance.

The overall architectureof our approach is dis-played in Figure 1.Reality mining com-prises three phases:Sensing, modelling andinterpretation.

The sensing part is concerned with therecording, storing and transmission ofsensor data that are generated by mobilephone sensors. These data are sent to acloud based dynamic graph model forfurther processing.

The modelling phase, which shows thedata processing pipeline in the cloud,consists of a sensor monitoring compo-nent in which raw sensor data areretrieved from roaming mobile devices.A REST service allows mobile clients topost their data to the service. In the data



Reality Mining at the Convergence of Cloud

Computing and Mobile Computing

by Matthias Steinbauer, Ismail Khalil and Gabriele Kotsis

During any social interaction, nonverbal social signals convey just as much information as the

conversation itself. While transmitting and analysing conversation is quite a common task for

machines, the transmission and analysis of social signals is not. The convergence of cloud

computing and mobile computing leads to a situation where insight into social systems is possible,

thus paving the way for exciting new applications.

Call Logs

Proximity

Social NetworkActivity

ShortMessages

Visualizations

Recommendations

Alerts

Sensing Modeling Interpretation

Cloud-basedDynamic Graph

Model

SensorMonitoring

DataCleansing

DataOrganization

Graph Modeling

V

nsotizailuasVi

s

Reco

otiammendReco

nso

sgoLllCa

ciSo

tymixiorP

rkoetwNlaci

Sens

roSens

taDa

taDa

G

ph ra

ciSo

rkoetwNlacityviActi

trShoesgasesM

oM

ngritonio

nginseaCl

zaniargO

noti oM

ngield

tserAl

esgasesM

Sens

ngiSens

M

ngieldoM

etaInterpr

notieta

Figure 1: Overall architecture

Sensing Modeling Interpretation

cleansing component, the raw data arecleaned of duplicates and unnecessarycontent. This step also allows us toclean data from blocked phone numbersand e-mail addresses.

The data organization component per-forms a first alignment of the raw datato the dynamic graph model. Here welook up user profiles, and conversationtopics or create them on demand if theydid not exist.

Finally, in the graph modelling compo-nent, the data are inserted into the graphmodel as vertices and edges in a singletransaction.

The final phase is the data interpretationin which the graph model is used to gen-erate visualizations, recommendationsand alerts.

Current research on reality mining tendsto focus on one of two aspects: (1)implementing mechanisms and tools

that run on mobile devices or (2) imple-menting methods to analyze data sets inthe cloud with low latency.

Our research focuses on dynamic graphmodels. These models keep their historyas the graph changes. In this way we areable to analyse various phenomena, forinstance changes in the structure ofsocial networks. On top of thesedynamic models we use queries thatadapt their results whenever the graphchanges in order to reflect the newmodel.

We are working on integrating methodslearned from this research into thedesign of collaborative applications andin applications of computer supportedcooperative work.

Mobile devices have become an impor-tant platform for understanding socialdynamics and influence, because oftheir pervasiveness, sensing capabilitiesand computational powers. The conver-

gence of mobile and cloud computing isforming the breeding ground for realworld applications in a research fieldknown as reality mining. This isundoubtedly an interesting field, whichis gaining new momentum with the con-vergence of mobile and cloud com-puting.

References:[1] B. Begole: “Ubiquitous computingfor business”, Upper Saddle River, NJ:Financial Times Press; 2011[2] Nathan Eagle and Alex Pentland:“Reality mining: sensing complexsocial systems”, Personal andUbiquitous Computing, 2006.

Please contact:Matthias Steinbauer, Ismail Khalil,Gabriele KotsisInstitute of Telecooperation, JohannesKepler University Linz, AustriaE-mail: {matthias.steinbauer,ismail.khalil, gabriele.kotsis}@jku.at


The fingerprinting technique is usedextensively for WLAN positioning. Theterminal collects the received signalstrength from several access points and,during a precalibration phase, comparesthe achieved vector to the vectors previ-ously recorded along with their posi-tions. This technique does not involvemodifications to the hardware. Othertechniques use the time of flight (ie thetime needed by a signal to travelbetween two nodes) to estimate the dis-tances to several access points at knownpositions and then apply a trilaterationprocess. The time of flight is more con-sistent than the signal strength. But inorder to avoid modifications to the ter-minal’s hardware, the time of flight must

be obtained from communication mes-sages by using only software.

Recent research at the TechnicalUniversity of Catalonia has led to a pro-cedure to measure distances betweenterminals [2]. This procedure is aimedat obtaining the time of flight afteradding timestamps to messages sent andto the corresponding acknowledge-ments received. The round trip time(RTT) is computed as the differencebetween both timestamps, and the dis-tance between the nodes is inferred byconsidering that the trip occurs at thespeed of the radio signal. The softwaremust interact with the link layer of theprotocol stack of the device. A simple

approach is to use a network interfacecapable of providing time measure-ments made by the hardware, but thetimestamps performed do not have anacceptable accuracy for many locationapplications (eg the characteristics ofIEEE 802.11 lead to a resolution of onemicrosecond corresponding to 300metres in distance.)

A more sophisticated approach is pre-sented in Figure 1. The protocol stack ofthe terminal is updated by introducingtwo software layers that are registeredin each terminal. The registrationprocess is run once and replaces the net-work interrupt handler (responsible forhandling the events related with the net-

Positioning Terminals in Mobile Computing

Networks

by Francisco Barcelo-Arroyo, Israel Martin-Escalona and Marc Ciurana-Adell

The ability to pinpoint a terminal’s position is useful for many applications of mobile computing and for

network optimization (eg handovers, tariffs, resource management). A range of techniques are available to

obtain a terminal’s position [1]. GPS, for example, is used externally to the network and achieves good

accuracy outdoors, with the trade off of increased energy consumption. Communication devices, however,

are frequently used indoors, connecting to private networks, such as WLAN. Since GPS is inaccurate

indoors owing to signal blockage and multipath errors, further research on indoors localization through

communication networks is required. Mobile computing is linked to indoors positioning in applications such

as: aged care, remote health control and security of buildings such as hospitals.

work interface) with a new one. Thisnew handler analyses the traffic trans-mitted and received and adds time-stamps to certain messages so that theRTT can be finally computed. Figure 1also shows the network architectureincluding the interrupt handler in aLinux-based device. The sources of theLinux kernel have been patched toallow location metrics (ie RTT) to beobserved. These changes alter themac80211 subsystem, which imple-ments most of the common MAC fea-tures in Linux. The goals of thesechanges are 1) to allow the location-related capabilities to be registered andreleased and 2) to add timestamps to themessages exchanged between the ter-minal and the access point. The capabil-ities are implemented as plugins, so that

each works as standalone. This allowsisolation of the bugs and extension ofthe capabilities without impacting thosethat are already working. An RTTplugin has been developed in order tocalculate the RTT between a node andan IEEE 802.11 access point. Thisplugin is responsible for most of thetasks developed by the interrupt han-dler. Specifically, it filters the traffic notsuitable for location purposes andmatches the transmission and receptionmessages involved in an RTT, so thatthe RTT can finally be computed. Theinteraction between the user’s applica-tions and the RTT plugin is done bymeans of system calls to a new modulenamed pos80211 [3]. This module pro-vides the computed and buffered RTTsto the user’s applications.

Currently, the proposed implementationhas been prototyped and tested at dis-tances shorter than 30 metres with goodresults. Future work includes moretesting (eg different scenarios, longer dis-tances) and developing new plugins forhyperbolic, instead of circular, trilatera-tion. The current plugin provides the RTTwhich is useful for circular trilateration,while time differences are appropriate forhyperbolic. The use of this proposal in adhoc networks is also investigated.

References:[1] Y. Gu, A. Lo, I. Niemegeers: “Asurvey of indoor positioning systemsfor wireless personal networks”, IEEECommun. Surveys & Tutorials, vol. 11,no. 1, pp. 13-32, 1Q 2009[2] F. Barcelo-Arroyo, M. Ciurana, I.Martin-Escalona: “Process and systemfor calculating distances betweenwireless nodes”, U.S. Patent 8 289963, October 26, 2012[3] J. Corbet, A. Rubini, G. Kroah-Hartman: “Linux Device Drivers”,O'Reilly Media, Third edition, 2005

Link: http://grxca.upc.edu

Please contact:Francisco Barcelo-Arroyo, UniversitatPolitecnica de Catalunya, SpainTel: +34 934016010E-mail: [email protected]



Figure 1: Proposed Linux implementation of the measurement system.

The Android security model is a particu-larly compelling reason for security-con-scious organizations to adopt Android asa platform for developing and distrib-uting mobile applications. In this secu-rity model, applications must declare aset of permissions that describes the setof privileged actions that might be per-formed during execution. Examples ofsuch actions include accessing cameradata, determining the location of a

device, or placing phone calls. Prior toinstallation, the user must explicitlyapprove the set of permissionsrequested by the application. Duringexecution, the Android OS is respon-sible for ensuring that the applicationonly performs those privileged actionsthat have been approved by the user.

This containment-based approach topreventing malicious behavior has the

benefit of potentially high scalability,since Android applications are not nec-essarily subject to a manual reviewprocess (as is the case for iOS applica-tions distributed in Apple's App Store).Unfortunately, the discovery ofAndroid-based malware in the wild hasprompted concerns that the Androidsecurity model is inadequate, especiallyif Android devices are to be adopted inhigh security environments (eg, the US

darkdroid - Exposing the dark Side

of Malicious Mobile Applications

by Engin Kirda

The Android operating system is a burgeoning platform for deploying mobile applications to

users, with more than 550,000 activations per day and an approximate 75% share of the

global smartphone market that eclipses the once-dominant Apple iOS [1,3]. This trend is

expected to continue, considering that Android's liberal licensing structure, open development

environment, wide adoption across multiple hardware manufacturers and carriers, and

modern end-user experience make it an attractive platform for both civilian and military use.


Department of Defense, the EuropeanCommission, etc.) [2] A major draw-back of the platform's approach to secu-rity is its reliance on user securityawareness to detect potentially mali-cious applications. Specifically, mostusers are not possessed of the requisitemotivation or security knowledge todetermine whether an application legiti-mately requires a declared set of per-missions, or to determine whether anattack has occurred at runtime.Therefore, there is a clear and pressingneed to supplement the Android secu-rity model with the means to detectmalicious code and remove the applica-tions that contain such code fromAndroid marketplaces.

Given their centralized nature, Androidapplication marketplaces are well-suited as a venue for deploying staticanalyses to detect potentially maliciousbehavior in Android applications. Staticanalysis has the significant advantagethat the entire program code can be ana-lyzed, and thus, it is possible to guar-antee the absence of certain, maliciousbehaviors. This is crucial for govern-mental stakeholders who might place asignificant amount of trust into theirapplications. Furthermore, market-places also present an opportunity todeploy application instrumentationframeworks to remove maliciousbehavior or harden vulnerable applica-tions against known classes of attacksprior to application installation.

DarkDroid is a joint project betweenNortheastern University in Boston andUniversity of California, Santa Barbara,that is being sponsored by DARPAsince 2012. The goal of DarkDroid is toresearch static mechanisms for provingthe non-existence of malicious code inAndroid applications. We use scalableand precise static analyses on Dalvikbytecode to discover attacks rangingfrom privacy violations to unauthorizedmodifications to critical sensor data.Using the results of our analyses, ourtools can then excise the malicious codeor harden other apps against attack.

Static code analysis has come a longway; it is nowadays an accepted tech-nique that is routinely used for exam-ining large source code repositories forthe presence of bugs and security vul-nerabilities. In most scenarios, the goalof a static analyzer is to find as manyprogram flaws as possible rather than to

guarantee the complete absence ofbugs. While this unsound approachmakes the analysis scalable and resultsin few false positives, it fails in the pres-ence of malicious code. The reason isthat malware authors write their pro-grams to specifically exploit the diffi-cult corner cases that an unsoundanalysis misses. Hence, we require aprecise static analysis that accuratelyhandles such difficult cases.

In the case of Dalvik (Android) applica-tions, we identify two main challengesfor precise analysis: First, programsmake use of complex data structures(such as hash tables) and polymorphicclasses with virtual methods. Second,Dalvik applications interact with theuser in non-trivial ways, and multipleapplications can collude (exchangemessages) to achieve a single, mali-cious goal. We propose novel staticanalysis techniques that improve theprecision of the analysis (and data-flowtracking) when facing complex datastructures, user interaction (throughintents), and virtual method calls. Thisguarantees that we will identify andcapture malicious code with a lownumber of false positives, even whenmalware authors attempt to obfuscatetheir actions.

Another significant problem is that theabsence of a particular class of mali-cious code does not imply that the appli-cation is not malware. In particular, it iscrucial that the analysis guarantees theabsence of a broad and diverse range ofmalicious behaviors for the analysis tobe useful. In DarkDroid, we areworking on novel ways to characterize

malicious activity. First, we leveragedata-flow analysis to capture maliciouscode that aims to breach confidentiality.Moreover, we extend data-flow trackingwith a fine-grain value set analysis todistinguish cases where certain data ele-ments are legitimately transmitted fromcases where unintended informationflows occur. Finally, we introduce codepatterns that represent attacks againstintegrity and availability of data, suchas patterns that indicate that the pro-gram tries to draw an unnecessaryamount of energy from the battery.

If successful, the impact of this projectwill be significant. In particular,DarkDroid will guarantee the absenceof broad classes of malicious code fromAndroid applications. Given the antici-pated low false positive rates and thedetailed output to render humananalysis easy, we believe that it is pos-sible to detect malicious applicationsbefore they are deployed on an Androidmarketplace.

References:[1] GOOGLE, INC. 2011 Q2 EarningsConference Call Slides, July 2011http://kwz.me/01[2] MCAFEE, INC. McAfee Q2 2011Threats Report Shows SignificantGrowth for Malware on Mobile Plat-forms, August 2011, http://kwz.me/02[3] GOOGLE, INC. Advanced sign-insecurity for your Google account,February 2011, http://kwz.me/05

Please contact: Engin KirdaNortheastern University, USAE-mail: [email protected]

Figure 1: Summary of the DarkDroid Project

A key problem of modern smartphonesis the limited life of their batteries. Theintroduction of larger screens and thelarge usage and availability of cpu-con-suming and network-based mobileapplications has aggravated thisproblem. The offloading of computationon cloud computing platforms can con-siderably extend battery duration.However, it is important to be able toverify not only whether offloading guar-antees real advantages with respect tothe computing power needed for datatransfer but also if user requirements aresatisfied, with respect to the quality ofservice and the costs of using the clouds.All the issues involved in the offloadingdecision, such as network disconnec-tions and variability, data privacy andsecurity, variations in load of the server,etc. need to be evaluated carefully.

At ICAR-CNR, we have designed aframework for the automatic offloadingof mobile application using a geneticprogramming approach, which attemptsto address the issues listed above. Theframework comprises two parts: amodule that simulates the entireoffloading process, and an inferenceengine that builds an automatic decisionmodel to handle the offloading process.

The simulator and the inference engineboth apply a taxonomy that defines fourmain categories concerning theoffloading process: user, network, dataand application. The simulator evalu-ates the performance of the offloadingprocess of mobile applications on thebasis of user requirements, of the condi-tions of the network, of thehardware/software features of themobile device and of the characteristicsof the application. The inference engineis used to generate decision tree basedmodels that take decisions concerningthe offloading process on the basis of theparameters contained in the categoriesdefined by the taxonomy. This is basedon a genetic programming tool that gen-erates the models using the parameters

defined by the taxonomy and driven bya function of fitness, giving differentweights to the costs, time, energy andquality of service depending on the pri-orities assigned.

TaxonomyA taxonomy of parameters and proper-ties has been defined and is used to takedecisions in order to build the modelthat decides the offloading strategy.The taxonomy only considers aspectsthat influence the offloading processand is based on four different cate-gories: Application (parameters associ-ated with the application itself), User(parameters assigned according to theuser needs), Network (parameters con-

cerning the type and the state of the net-work), and Device (parametersreflecting the hardware/software fea-tures of the devices). The decisionmodel built by the GP tool will take thedecision whether to offload or not onthe basis of the parameters associatedwith these categories.

ArchitectureFigure 1 shows the software architec-ture of the framework.The first fourmodules contain sets of data for each of

the four categories considered (user,device, network and application), whichwill be used by the sampler module inorder to generate the training and vali-dation datasets, simply by randomlycombing the data. These two datasetswill be used to generate and validate thedecision models.

The two main modules depict the infer-ence engine, consisting of a GeneticProgramming module which develops apopulation of models capable ofdeciding the possible offloading of amobile application, and the simulationmodule comprising the well-knownGreenCloud simulator (simulating thecloud part of the offloading process)

and an ad hoc mobile simulator model-ling the mobile device's behaviour. Inpractice, each model generated by theGP module, will be input to the simu-lator module and a weighted fitnessfunction will be computed that evalu-ates the performance of the model onthe basis of the energy wasted, timeconsumed, costs and QoS.

At the end of the process, the best model(or the best models) will constitute therules adopted by the offloading engine,



Automatic Offloading of Mobile Applications

Using Evolutionary Algorithms

by Gianluigi Folino and Francesco Sergio Pisani

We present a framework for generating decision-tree based models that take automatic decisions

on the offloading of mobile applications onto a cloud computing platform using an algorithm based

on the genetic programming (GP) approach.

Figure 1: The

software architecture

of the system

which will decide whether an applica-tion must be offloaded, according to theconditions assigned (user requirements,bandwidth, characteristics of the mobiledevice and so on). All these modelsmust be validated using the simulationengine with the validation dataset. If theresult of this evaluation is above a pre-defined threshold, the model will beadded to a model repository for futureuse.

Conclusions and PerspectiveThis work presents an automaticapproach to generate decision-taking

models for the offloading of mobileapplications on the basis of userrequirements, conditions of the net-work, the hardware/software features ofthe mobile device and the characteris-tics of the application. The system con-stitutes a general framework for testingoffload algorithms and includes amobile simulator, which computes theenergy wasted in the process ofoffloading. Ongoing and future activi-ties involve testing the framework withreal datasets and verifying whether theobtained models improve battery per-formance in real environments.

Reference: K. Kumar and Y.-H. Lu: “Cloudcomputing for mobile users: Canoffloading computation save energy?”,IEEE Computer, 43(4), 2010.

Links:http://www.genetic-programming.org/http://www.icar.cnr.it/

Please contact:Gianluigi FolinoICAR-CNR, ItalyE-mail: [email protected]


Scheduling data Mining Applications

in Mobile Computing Environments

by Carmela Comito, Deborah Falcone, Domenico Talia and Paolo Trunfio

Data mining is emerging as a promising topic in mobile computing environments. We have defined a

distributed architecture in which mobile devices cooperate in a peer-to-peer style to perform a data

mining process, tackling the problem of energy capacity shortage by distributing the energy consumption

among the available devices. An energy-aware (EA) scheduling strategy assigns data mining tasks over a

network of mobile devices optimizing energy usage. The main design principle is to find a task allocation

that prolongs the network residual life by balancing the energy load among the devices.

The wide availability and growing com-puting power of mobile devices hasopened the way for data analysis andmining in mobile scenarios [1]. Mobileapplications exploiting data miningtechniques have appeared on the marketin recent years. Examples include smart-phone-based systems for body-healthmonitoring, vehicle control, and wire-less security systems. An importantaspect that must be addressed is that ofensuring energy efficiency, as mostmobile devices have battery powerwhich would last only a few hours. Datamining tasks in mobile environmentsshould be allocated and scheduled so asto minimize the energy consumption oflow-capacity mobile devices.

We have worked in this direction bydefining a distributed architecture inwhich mobile devices cooperate in apeer-to-peer style to perform datamining tasks, tackling the problem ofenergy capacity shortage by distributingthe energy consumption among theavailable devices. Efficient resourceallocation and energy management isachieved through clustering of mobiledevices, as shown in Figure 1. With thisapproach, mobile nodes can be assigned

different roles, such as cluster-head orcluster member. A cluster-head servesas the local coordinator for its cluster,performing intra-cluster transmissionarrangement and data forwarding, whilea cluster member is a non-cluster-headnode without any inter-cluster links.

To evenly exploit all availableresources, a proper distribution of datamining tasks among clusters and indi-vidual devices is crucial. To this end, wedefined an energy-aware (EA) distrib-uted task scheduling strategy whosegoal is to find a task allocation that pro-longs the network residual life. The EAscheduler implements a two-phaseheuristic-based algorithm. When anassignment decision has to be made fora task, the first phase, denoted localassignment phase, is responsible forlocal task arbitration: it considers theenergy consumption of task executionon the different devices within the localcluster. The algorithm tries to minimizethe total energy consumed in the clusterby assigning the task to that device thatpermits the cluster residual life to beextended. If the first phase is not fea-sible, the second phase, denoted globalassignment phase, is responsible for

task arbitration among clusters: the taskwill be assigned to the most suitabledevice, in the network of clusters, thatmaximizes the network residual life.

We performed an evaluation of the EAallocation strategy using a network sim-ulator, which allowed us to assess itseffectiveness on a set of data miningtasks [2]. The simulation mainly aims tostudy the behaviour of the schedulerwith respect to network residual life,number of alive devices, and number ofcompleted tasks at the end of the simu-lation. To assess the effectiveness of theEA strategy, we compared its perform-ance with that achieved by the well-known round-robin (RR) schedulingalgorithm. As a first step, the simulatorbuilds a network composed of 100mobile devices, and lets them groupinto clusters. Then, an initial energycapacity ranging from 3,000 to 11,000Joules is assigned to each device, fol-lowing a normal distribution. After theinitial setup, mobile devices start gener-ating a set of data mining tasks to beexecuted; these are allocated to theavailable nodes according to the EAstrategy. For the purpose of our simula-tion, we considered three reference data

mining algorithms from the Wekaproject [3], namely J48, K-means andApriori.

For each reference algorithm, we ran aset of tasks with a fixed dataset size(200 kB) and task arrival rate λ varyingfrom 80 to 1280 tasks per hour. Figure 2shows the network residual life meas-ured at the end of the experiments forthe three algorithms, using EA and RR.As expected, increasing the task arrivalrate, the network residual live tends tozero both for EA and RR. However, forthe lightest of the three data miningalgorithms (Apriori), the residual lifedoes not reach zero and the differencebetween EA and RR increases with λ infavor of EA. Figure 3 shows the numberof alive devices for the three algorithms,using EA and RR. Also in this case, theresults demonstrate that the number ofalive devices with EA is greater thanthat achieved by RR. Finally, Figure 4compares the performance of EA andRR in terms of completed tasks for thethree algorithms. With Apriori, both EA



Figure 2: Network residual life

Figure 3: Number of alive devices Figure 4: Number of completed tasks

Figure 1: Reference architecture

and RR are able to complete more tasksas λ increases, but EA ensures better per-formance. With J48 and K-means, with agiven task arrival rate, the number ofcompleted tasks cannot increase becausethe network residual life is zero, asshown in Figure 2. However, even inthese cases, there is an advantage for EAcompared to RR with λ lower than 320tasks/hour.

In summary, the experimental resultsshowed that a significant improvementcan be achieved using our EA schedulercompared to the time-based RR sched-uler. Our algorithm: i) is effective inextending the network residual life byreducing the energy consumption,without limiting the number of com-pleted tasks; ii) keeps alive most of themobile devices, in all the experimentsperformed, thanks to its energy load bal-ancing strategy.

Link:http://grid.dimes.unical.it

References:[1] D. Talia, P. Trunfio: “Service-oriented distributed knowledgediscovery”, Chapman and Hall/CRC,2012[2] C. Comito et al: “A distributedallocation strategy for data miningtasks in mobile environments”, in proc.of the 6th International Symposium onIntelligent Distributed Computing,Studies in Computational Intelligence446: 231-240, 2012[3] M. Hall et al: “The WEKA datamining software: an update”. SIGKDDExplorations 11(1): 10-18, 2009.

Please contact:Carmela Comito and Domenico TaliaICAR-CNR and DIMES, University ofCalabria, ItalyE-mail:{ccomito,talia}@dimes.unical.it

Deborah Falcone and Paolo TrunfioDIMES, University of Calabria, ItalyE-mail:{dfalcone|trunfio}@dimes.unical.it


Multi-device sensor systems often intro-duce the problem of synchronizing mul-tiple data streams, as sub-millisecondaccuracy is often needed. Synchronizingmultiple sensors at run-time is a com-plex task. After data acquisition, offlinesynchronization by correlation analysisbetween two data streams solves theproblem. Two IMUs, for instance, canbe synchronized if there are shared fea-tures in some of the data signals.However, for statistically independentdata sources this approach is not fea-sible. In this situation, a physical con-nection can help by providing a synchro-nization signal. Synchronizationbetween two identical devices via wire-less communication consumes a lot ofpower and requires an advanced net-work protocol. Our solution uses asingle smartphone to control an arbitrarynumber of IMUs (ETHOS Sensors [1]).

Controlling Sensors via ANT+TMWe use small, light-weight IMUs thatfeature on-device storage [1]. AnANT+TM-capable network chip takescare of the wireless communication.ANT+TM defines a data source, ie asensor, as a master device. An aggre-gator, such as a smartphone, collectsdata and is dubbed slave in this setting.The network chip uses a TDMA-likeadaptive isochronous network tech-nology for communication.

An embedded device's energy scarcity isexacerbated by the requirements ofwireless communication. Our sensornodes draw current from 300 mAh Li-Ion batteries; hence, power awarenesswas crucial during the development.Deployment of the devices in mastermode is suboptimal for two main rea-sons: Constantly sending data, even ifnot used by an aggregator, wastesenergy: not only does the cost of wire-less communication add to the powerbudget, but also the increased load onthe operating system uses more power

since a node cannot go to sleep. Further,frequent collisions on the physical layerat high transmission rates with multiplemaster devices would not only reducethe throughput of the network:Retransmissions also impact the real-time data processing of an aggregatorand thus aggravate the problem of

power scarcity. These considerationsmotivated our solution. We configuredthe sensors to slaves: they do not emitdata packages on their own, but listen tonew packages on the channel. On theother hand, our aggregator device - anX-Peria S by Sony - assumes the role ofa master: in idle mode it emits heartbeatpackages that are received by anysensor device tuned to a common radiofrequency.

In addition to this network-layer con-nection, we also implemented a simpleapplication layer: A user of the systemcan control all or specific nodes with aset of commands. Each node is config-ured to a node ID - a simplistic equiva-lent of an IP-address. The master emitseither broadcast or addressed packages.A sensor node only reacts to a networkpackage if that package either is abroadcast package or if the destinationID of the package matches the node'sID. There is one command to start sam-

Figure 1: System setup: A smartphone

communicating to an arbitrary number of

sensors; (size ratio of sensors to smart phone

is accurate).

Figure 2a: Communication with

ANT+TM-. A smartphone requests

data from sensor 1 (green slots) which

answers in the same slot (red). Later,

the smartphone sends a command to

sensor 2 (blue slot).

Figure 2b: Network interrupt priority.

A smartphone requests data from

sensor 1 (green slots) which answers

in the same slot (red). Later, the

smartphone sends a command to

sensor 2 (blue slot).

On the Benefits of a Poly-Cultural Sensor Setup:

Controlling Embedded Sensors with a Smart Phone

by Rolf Adelsberger and Gerhard Tröster

We present our wireless solution to device synchronization, control and real-time feedback for small and

light-weight Inertial Measurement Units, (IMUs). The controller is implemented on a smartphone. We base a

custom application layer on the protocol layer implemented on a low-power radio chip by ANT+TM. Our

system is capable of synchronized (< 15 μsec) control of an arbitrary number of sensors (ETHOS), streaming

data for real-time visualization and it reduces power consumption compared to other approaches.

pling. The same operating system(firmware) runs on each sensor node.Network communication is handled byan interrupt service routine with highestpriority.

In fact, every other software interrupt isoverwritten by the network-stack inter-rupts. This way, we can ensure that eachnode receives the commands at virtuallythe same time: the micro-controller ofour sensor boards features fixed-over-head interrupt handlers.

For 100 iterations we started the sam-pling on 10 rigidly connected sensornodes, exposed all of them to a commonpeak in acceleration and calculated themaximal inter-node offset between any

two sensor nodes. The mean offset was12μ sec. A user can select a sensor tostream its data to the aggregator: theGET DATA command requests datapackages from an addressed sensornode. The ANT+TM-protocol allowsslave devices to reply to a master devicein the same time slot. We exploit thisfeature to implement real-time datastreaming: in streaming mode themaster sends addressed packages to onedevice which replies with packagescontaining the requested sensor data.

ConclusionsWe have shown that multiple types ofwearable devices can be combined into asensor system to create something morepowerful than a mono-cultural system.

By carefully programming the softwaremodules wireless synchronization canbe as accurate as 12 μsec.

Link: http://www.ife.ee.ethz.ch/

Reference:[1] H. Harms et al: “ETHOS:Miniature orientation sensor forwearable human motion analysis” inproc. of IEEE Sensors conference,IEEE, 2010.

Please contact: Rolf Adelsberger Federal Institute of Technology Zurich,ETHZ, SwitzerlandE-mail: [email protected]



The year 2011 marked the beginning ofthe post personal computer (PC) era, asthe number of smartphone shipmentsexceeded, for the first time in history, thesales of all types of PCs combined (ie,notebooks, tablets, netbooks and desk-tops). According to IDC’s “WorldwideMobile Phone 2012-2016 Forecast andAnalysis”, Android is projected to domi-nate the future of the smartphone industrywith a share exceeding 53% of all devicesshipped in 2016. Currently, an Androidsmartphone provides access to more than650,000 applications and those appsintroduce unprecedented possibilities,knowledge and power to humankind.

Re-programming smartphonesand instrumenting them forapplication testing and datagathering is currently a tedious,time-consuming process thatposes significant logisticalchallenges. To this end, wehave implemented and demon-strated SmartLab [1], a first-of-a-kind open Infrastructure-as-a-Service (IaaS) cloud thatenables fine-grained controlover both real and virtualsmartphones via an intuitive

web-based interface. Our current infra-structure is ideal for scenarios thatrequire fine-grained and low-level con-trol over real smartphones, eg, OS,Networking, DB & storage, security,peer-to-peer protocols, but also for sce-narios that require the engagement ofphysical sensors and geo-location sce-narios. Our preliminary edition hasbeen utilized extensively in-house forour research and teaching activities[2,3] but it has also been open toselected research groups around theglobe. SmartLab provides a diverse,high-availability platform that can beutilized by the mobile computingresearch community to engage more

effectively in systems-oriented researchon smartphones.

SmartLab’s hardware comprises 40+Android smartphones and our data centrethat encompasses over 16TB of RAID-5/ SSD storage on an IBM X3550, as wellas 220GB of main memory on 5 IBM /HP multiprocessor rackables. SmartLabsupports a variety of connection modali-ties. In particular, most of our devices areconnected to the server with USB hubs.Additionally, a number of smartphonesare connected from within our DMSLresearch lab (in the same building)through the network on virtual serversthat reside on our data centre. The given

mode is particularly promisingfor scenarios we want to applyto scale our testbed outside thedepartment. Finally, a fewdevices within the departmentare also wirelessly connectedusing WiFi.

SmartLab supports four modesof user interaction with thesmartphones: i) RemoteControl Terminals (RCT), aweb-based remote screen ter-minal for Android, developed

SmartLab: Empowering Mobile Computing

Research through an Open Smartphone Cloud

by Georgios Larkou, Panayiotis Andreou, Andreas Konstantinidis and Demetrios Zeinalipour-Yazti

SmartLab is a first-of-a-kind open cloud of smartphones that enables a new line of systems-oriented

mobile computing research.

Figure 1: Part of the Smartlab smartphone fleet connected locally to

our datacentre


in-house using Ajax that mimics touchscreen clicks and gestures among otherfunctionalities; ii) Remote Shells (RS),a web-based shell developed in-housewith Ajax that enables a wide variety ofUNIX commands to be issued to theAndroid Linux kernels; iii) RemoteScripting Environment (RSE), whichallows users to author AndroidMonkeyRunner automation scripts(written in python) and upload them tothe devices to perform automated tasks;and iv) Remote Debug Tools (RDT),which provide web-based debuggingextensions to the Android Debug Bridge(ADB). Through these tools, SmartLabfacilitates research in smartphone net-work programming environments, com-munication protocols, system designand applications.

In the near future, we are planning anumber of exciting extensions forSmartLab. Firstly, we are integrating a

prototype Apache HBase installationwithin our data centre, to store sensorreadings in a tabular and scalable (ie,column-oriented) format. The givenstore can be utilized to store billions ofsensor readings that can be fed to ourGPS/Sensor Mockup subsystem.Additionally, we are currently workingwith local telecommunication compa-nies in order to obtain 3G data time forour mobile fleet, and local transportationcompanies, which will be moving ourdevices around within a city. This willallow testing of algorithms, protocolsand applications within a real mobileurban environment, thus providing afirst-of-kind open mobile programmingcloud. Furthermore, we are developing aWeb 2.0 JSON-based API of our testbedusing the YII framework. In particular,this effort will allow users to access thesubsystems of our testbed in a program-mable manner (ie, Web 2.0 JSON inter-actions). Finally, we are also looking

into mechanisms for supporting securityand privacy experiments.

Links:http://smartlab.cs.ucy.ac.cy/http://dmsl.cs.ucy.ac.cy/

References:[1] A. Konstantinidis et al: “Demo: aprogramming cloud of smartphones”,ACM Mobisys '12[2] D. Zeinalipour-Yazti et al:“Crowdsourced Trace Similarity withSmartphones”, IEEE TKDE, 2012[3] G. Chatzimiloudis, et al: “Crowd-sourcing with Smartphones”, IEEEInternet Computing, Volume 16, 2012.

Please contact:Demetrios Zeinalipour-Yazti Department of Computer ScienceUniversity of CyprusTel: +357 22 892755E-mail: [email protected]

Figure 2: Screenshot of the SmartLab web-based user interface

Phone Accessories as an Interface

by Mattias Jacobsson, Stina Nylander and Ylva Fernaeus

Mobile ActDresses is a design concept that utilizes existing practices of accessorizing, customization

and manipulation of a physical mobile device to predict and control the behaviour of its software.

Existing smartphone technology can be augmented by taking inspiration from existing practices in

relation to costume, jewellery, personalization and fashion.

There are various physical means bywhich people personalize their digitaldevices: by placing stickers on laptops;by using customized cases; and byattaching mascots and charms to smart-phones, for instance [1]. Another area ofcustomization is the use of personalizeddigital themes and the growing amateurpractice of making small and personal

mobile applications. The mobile phoneas such is thus not merely a technolog-ical device but also an object for per-sonal expression. Inspired by the trendof personalizing devices, we have beeninvestigating the potential of usingphysical clothing, accessories andlabelling as an alternative means of con-trolling mobile interactive systems by

developing a prototype system: MobileActDresses.

The design space of Mobile ActDressesranges conceptually, from single on/offmode switchers to more complex con-figurations with combinations of activelabels and accessories. Our prototypeenables people to attach physical acces-

sories to their smartphone and, at thesame time, change its digital function-ality or appearance. Below we outlineexamples of different strategies forimplementing and deploying this designconcept based on existing standards onmobile phones currently on the market.

Various implementation experimentsThere are many forms of technologyavailable for mobile phone handsetsthat could potentially be used to imple-ment the concept of ActDresses [2].However, wireless protocols such asBluetooth and WiFi require active trans-mitters, and also have a relatively longrange of communication, features thatare incompatible with having signs inthe immediate physical context of thedevice that they control. Using thephone’s camera to read barcodes on anobject, for instance, requires explicitreading and is not ideal given that itconflicts with the immediate physicalcontext requirement.

Another emerging wireless method isNear Field Communication (NFC),which enables the exchange of databetween devices at a distance of up toten centimetres from one another.Drawbacks in this case are that thereading must be constantly active on thedevice and that most devices currentlydo not support simultaneous reading ofmultiple tags. As with the wireless pro-tocols, there is a range of possible wiredor direct contact solutions for realizingthe concept. Examples include experi-mental solutions such as Pin & play,iButtons, conductive stickers, resistors,USB, or even the built in memory cards.

There are thus a number of partiallycomplete technical options currentlyavailable. Our conclusion is that it willbe crucial to have a dedicated channel,whether it is wireless or wired. The keypoint is that physical ‘sockets’ restrict-positioning of tags whereas a wirelesssolution can be designed to be both freeand ‘socketed’.

The design prototypeIn our first prototype (Figure 1), weequipped physical shells with small butstrong Neodym (NdFeB) magnets posi-tioned at different locations. Theabsolute distance between the magnetand the magnetometer in the phone issensed, and can be used to triggerevents in software. In this case a simpleservice application was developed that

changes the theme on a mobile Androidphone according to the style of the shell.Furthermore, we used theheadphones/handsfree jack as anexample of a wired ActDress togetherwith jewellery (Figure 2).

A general design concept for expressionTo conclude, our goal with this proto-type was to explore how the concept ofActDresses could be extended to digitalartifacts in general [3]. Our inspirationcomes from how clothes are worn bypeople to serve a range of communica-tive functions, indicating appropriatebehaviours, group belongings, andexpected interactions, for example.Similarly, physical accessories attachedto a device could be used as a resourceto indicate what mode the device is cur-rently in, and what behaviours andinteractions could be expected.

Link: www.sics.se/projects/actdresses

References: [1] J. E. Katz, S. Sugiyama: “Mobilephones as fashion statements: evidencefrom student surveys in the US andJapan”, New Media Society, 8(2),321–337, 2006 .doi:10.1177/1461444806061950[2] M. Jacobsson, Y. Fernaeus, R.Tieben, R.: “The look, the feel and theaction”, in proc. of the 8th ACMConference on Designing InteractiveSystems - DIS’10,p. 132, New York,USA: ACM Press, 2010doi:10.1145/1858171.1858196[3] Y. Fernaeus, Y., M. Jacobsson:“Comics, robots, fashion andprogramming: outlining the concept ofactDresses” (pp. 3–8). Cambridge, UK:ACM, 2009doi:10.1145/1517664.1517669

Please contact:Mattias Jacobsson, SICS Swedish ICTTel: +46 703 128557E-mail: [email protected]



Figure 1: Mobile phone prototype using

magnets on shells to control themes and other

content.

Figure 2: Headphone and jewellery hack to

control applications and other content.

Jill rarely leaves the house without her

mobile phone, which she customizes

physically as well as digitally to match her

own clothes of the day. Thereby the phone

itself works as a fashion item that she

seeks to match with her outfit. Upon

entering her office, Jill attaches the

company shell to her mobile phone

handset, which enables the phone to let

her into the building, as well as acting as

a company identity marker and label.

Plus, it goes well with her work outfit. The

phone is now set into a mode that

automatically loads her work contacts as

her primary address book in the phone.

While the shell is on, all charges on the

phone get placed on the company, rather

than on her personal, phone bill. When

leaving the office she removes the shell on

her phone, which then replaces her office

applications with her favorite spare time

applications on the front screen. Later in

the evening Jill goes out for a drink and

attaches her ‘after work’-charm that

shows her VIP status at one of the local

clubs. Not only does it enable rebates at

the club when paying with the phones

digital wallet, it often becomes a

conversation piece, and therefore a

marketing tool for that specific club.

A day in Jill’s Mobile Life


Most locations in developed regions arecovered by a multitude of overlappingwireless networks from several net-work providers that have 3G/4G net-works deployed, as well as hotspots andlocal Wi-Fi-networks. Usually, userscan only access one of these networksat a time. But what if several of theavailable networks could be usedsimultaneously? Would splitting dataover two wireless networks also makemobile communication twice as fast?And would it increase the overall per-formance of the communication? Theresults of research by CWI and Thalesexceeded expectations: simultaneoususe of two networks did not just doublethe speed of wireless communication,but could increase it up to a factor often.

High speed gainThe key factor in this speed gain is thestrong fluctuation in available networkcapacity. In this context, one may usetemporal under-utilization in one net-work to compensate for congestion inthe other network. Even if both net-works are fully utilized, there are stillvery high fluctuations in supply and

demand of network capacity on a microscale. The main idea is to split trafficstreams over the different access net-works in such a way that the user-per-ceived performance is optimal, forexample by assigning packets to thenetwork with the lowest utilization.This smart assignment of packets toparallel wireless access networks notonly leads to an increase in the networkthroughput, but also leads to greatenhancements of response-time per-formance, availability and robustnessof the applications.

Removing peaks in capacity demand isvery effective in reducing congestion.As in several logistic problems, speedis not linearly, but rather exponentially,related to available capacity. If thenumber of cars on a road could bedecreased by 10% during rush hour,traffic jams would not decrease inlength by 10%, but would be halved oreven eliminated. The same happens inwireless networks: the download timesof individual users can be up to tentimes shorter when capacity peaks areflattened by using two networks at thesame time.

Sharing networksThe developed techniques can be real-ized in the short term, simply bydevising a smartphone that can combineany mobile networks, hotspots or Wi-Fi-networks, and a piece of softwarebased on our models. But for a moreeffective implementation, networkproviders need to adapt their softwareand hardware to this new mode of oper-ation and more importantly, reach anagreement with competing providers toshare networks.

Application areasMany application areas, such as emer-gency and security services and the mil-itary, could greatly profit from fast andreliable networking. Such organizationscould use their own networks to profitfrom these new techniques.

Developing the modelThe mathematical challenge of model-ling the use of multiple networks at thesame time lies in devising accuratemodels that determine how the fileshould be split. The complication lies inthe fact that the networks involved arecoupled: the download starts at the

Boosting Performance of Wireless Networks

with Concurrent Access by Smart Traffic Splitting

by Gerard Hoekstra and Rob van der Mei

The spectacular growth of mobile internet on smartphones and tablet computers has boosted the

demand for fast wireless networks. By 2015, mobile data exchange is expected to be 26 times larger

than it was in 2010 [1]. In the first six months of 2012, the mobile internet traffic volume in the

Netherlands was 21% higher than that recorded in the last six months of 2011 [2]. A highly promising

means to meet the increasing demand is to take advantage of the fact that many geographical areas

are covered by a multitude of overlapping networks. This phenomenon is referred to as concurrent

access (CA). CWI and technology company Thales Nederland B.V. have developed new methods to

make efficient use of the possibilities of CA by developing and implementing smart algorithms to split

traffic over the multitude of wireless access networks.

Figure 1: Users covered by multiple, overlapping wireless networks from different technologies and/or providers experience improved performance by

using concurrent access.

same time in all networks, with trans-ported data through each networkdepending on the original file size andthe statistical (stochastic) variation inavailable capacity in each network.

We first developed an accurate, experi-mentally verified model of file down-loads in a single network. This model isbased on queuing; abstractions used incomputer science to model waitinglines. The behaviour of multiple net-works was modelled by a network con-sisting of several of these queuingabstractions. The performance of these

networks and the efficiency of the split-ting algorithms was analysed and opti-mized using Processor Sharing modelsand optimization techniques likeMarkov Decision Processes andBayesian analysis. By assuming perfectsplitting of files, the researchers found atheoretical solution for this model.Experimental results subsequentlyshowed that this theoretical solutionindeed provides nearly optimal per-formance in practice, with a differencebetween modelled and practical out-comes of no more than a few percent.More details can be found in [3].

References:[1] TNO Monitor draadloze technolo-gieën 2011, TNO, August 2011[2] OPTA, Marktcijfers tweedekwartaal 2012, 06-11-2012[3] G.J. Hoekstra: “Concurrent Accessand Traffic Control Methods in WirelessCommunication Networks”, Ph.D. thesis,VU University Amsterdam, 2012http://kwz.me/0S

Please contact:Rob van der Mei, CWI, The NetherlandsTel: +31 20 592 4129 E-mail: [email protected]



Mobile systems are not only useful intheir ability to provide pervasive accessto computing systems (ie enabling com-puter access anywhere), but they alsoprovide the inputs of a mobile anddynamic environment into a computa-tional system; it is now possible to per-form computations that, until recently,were simply impossible. A classicexample is geolocation; it is now easy toprovide the physical location of a user,and to use this data for a variety of pur-poses.

In particular, such data can be used inthe context of transportation. Transportis an important issue in modern soci-eties, especially in big cities, and repre-sents a significant proportion of globalenergy consumption. In many cases,energy consumption can be significantlyreduced by an efficient use of transportmedia. Not only is choosing the rightpath within the transport network impor-tant, but also being able to share avehicle (including private vehicles) withothers. There are several initiativesassisting people in sharing transport, themost popular being carpooling: morethan one person sharing a car.

However, no existing solution combinesprivate transport sharing with the use of

public transport, which would make itmore flexible. We have designed an ITplatform, called CoMobility, to assist inintermodal transport sharing, inte-grating the use of carpooling withpublic transport, as well as other privatetransport media.

To take advantage of this approach,individuals must be convinced of thebenefits of reducing the number of pri-vate cars, and of the need of a newmodel of transport. For this purpose,

our CoMobility platform has "cus-tomized" analytics on savings andenergy consumption, to make individ-uals aware of the benefits of this newway of travelling. These data areobtained by combining private datafrom carpooling, and open data frompublic transport networks, and fromenergy-aware institutions.

Our system requires the integration oftwo types of data set: private data (pro-vided by private providers and con-

CoMobility: A Mobile Platform for Transport Sharing

by Carlos E. Cuesta, Paloma Cáceres, Belén Vela and José María Cavero

Mobile systems are becoming ubiquitous, which, in combination with wide-range service-oriented

architectures, offers enormous potential for a range of uses. In this context, we have developed the

CoMobility platform, a system designed to integrate carpooling and the use of public transport networks,

with the goal of reducing energy consumption and CO2 emissions. CoMobility defines a service-oriented

platform to help mobile users plan their use of transport, including sharing, with the purpose of saving both

energy and money.

Figure 1:Conceptual architecture of the CoMobility platform


sumers of the transport system) andopen data (ie data from public transportnetworks and public institutions). Theformat of public data within the opendata initiatives prevents non-expertsfrom using them directly, and thus itrequires additional semantics, as pro-vided by “Linked Open Data” [1][2] .

In summary, CoMobility provides asystematic approach to (i) accessingopen, integrated and semantically anno-tated transportation data and streetmaps, (ii) combining them with privatedata, and (iii) supplying mechanisms toallow the actors to share and searchthese data. Therefore, the CoMobilityconceptual architecture provides themeans to perform the following tasks(also depicted in Figure 1).

First, the platform can identify, select,extract and integrate data from differentand heterogeneous sources, stemmingfrom the transportation, geographicaland energy domains. Second, data frompublic institutions is obtained automati-cally in the form of open data. Third,these data are annotated as linked data,and a set of heuristics generate linksbetween data items from different

sources without human intervention.Fourth, these data are integrated withprivate data provided by users them-selves. And finally, CoMobility pro-vides intuitive and customized data ana-lytics and visualization, allowing indi-viduals to become aware of the environ-mental impact of their transportchoices.

The CoMobility platform is provided onthe Internet “as a service”, where bothpublic transport information and dataprovided by users themselves are storedand accessed “in the cloud”. The cloudapproach is necessary as scalability isone of the most important requirementsof this kind of wide-range service archi-tecture. The platform needs to access agreat amount of data, which is alsostored in the cloud – both the privatedata of carpoolers, and the public dataaccessed in a linked open dataapproach. Users are able to access theirinformation in several formats, particu-larly in mobile devices (currently,Android devices) and web applications.Through these devices, they are able toplan their paths in the transport net-work, moving from a shared car to theunderground, and from there to a bus

line; and at the same time receiving anestimation of the saving of both moneyand energy.

With this system, we hope to encouragesocial and individual change towards anew - more efficient and environmen-tally friendly - model of transport.

Link:VorTIC3 Research Group (Rey JuanCarlos University):http://www.vortic3.com

Reference:[1] Linked Open Data.http://linkeddata.org/[2]T. Heath, C. Bizer: “Linked Data:Evolving the Web into a Global DataSpace”, Morgan & Claypool, 2011.

Please contact:Carlos E. Cuesta, Paloma Cáceres,Belén Vela and José María CaveroVorTIC3 Research Group, Rey JuanCarlos University, Madrid, SpainE-mail: [email protected],[email protected],[email protected],[email protected]

Quality of Service Information System:

Get to Know the Performance of Your Mobile

Network Operator Anywhere-Anytime

by Katarzyna Wac

Quality of Service Information System (QoSIS) focuses on measurement-based performance

evaluation of wireless access networks provided by diverse mobile network operators in diverse

locations and times. We have developed an Android OS mobile application that uses measurement

data provided by real mobile users living in the Geneva area to predict the networks' expected

performance. Measurement data, and therefore predictions, are available for Swiss operators:

Swisscom, Sunrise, Orange CH, as well as French operators: SFR, Bouygtel and Virgin.

The effect iveness of any mobileservice depends on the quali ty ofservice (QoS) provided by the wirelessaccess network it uses. However, theQoS is often unknown, as public andprivate wireless access networkproviders, mobile network operators(MNOs) for instance, tend to not dis-close detailed, real-world QoS-infor-mation. For marketing purposes, theseproviders usually advertise only thebest data rate values for their net-works.

According to the 4G vision, in the nearfuture wireless access networks of var-ious providers - employing differentwireless access technologies - will beubiquitously available for mobileservice users. Also, a seamless han-dover between these networks will sup-port users’ mobility. Ideally, usersshould have a priori knowledge aboutthe QoS provided by different networks.Based on that knowledge, a mobiledevice, on behalf of its user, could makean informed choice about which wire-

less access network provider and tech-nology to use for the preferred mobileservices of the user.

To date, there are no unbiased, externalproviders of such information in themobile business landscape. Users, whogain personal experience of variousmobile services, may eventually usetheir acquired knowledge to manuallyreconfigure their devices and share theirknowledge with family and friends.However, there is no service platform

enabling users to collaboratively sharetheir collected knowledge.We aredeveloping a Quality of ServiceInformation System (QoSIS) to fill thisgap. QoSIS distributes predictions tomobile devices about the QoS providedby the different wireless access net-works available at a given geographicallocation and time. These predictions arethen used by a mobile device, on behalfof the user, to choose a wireless accessnetwork provider and technology to beused. The knowledge furnished byQoSIS also allows mobile serviceproviders to adapt their service deliveryto the predicted QoS, thus increasingthe service quality and improving userexperience.

Mobile devices, on behalf of their users,can contribute to the QoSIS database ina collaborative information-sharingmanner by submitting collected dataabout the QoS provided during theirmobile service use, given the selectedwireless access network provider andtechnology. The principal dimensionsof the QoSIS database are: geographicallocation, time, wireless access networkprovider and wireless access tech-nology. QoSIS is based on a QoS pre-diction engine. Based on machinelearning algorithms, the engine builds aheuristic from which to derive predic-tions.

We have assessed the feasibility ofderiving predictions in a case studybased on delay measurement data col-lected from the mobile device of ahealth telemonitoring service user. Wehave shown that it is feasible to predictthe value of delay for a device, based onits own measurement data or on datacollected by another device being usedat the same location and time and usingthe same wireless access networkprovider and technology. It is also fea-sible to predict the value of delay for adevice, based on delay measurementdata collected by both devices [1].

In contrast to existing approaches, theproposed QoSIS is entirely user-driven:mobile service users collaborativelycreate the QoS-information that wouldprovide the basis for QoS prediction forother users. QoSIS therefore imple-ments the Mobile Web 2.0 paradigm. Inline with this approach, we have alsoassessed the business scenarios for anenterprise based on QoSIS services -provisionally named “QoSIS.net”. This

enterprise could provide a QoS predic-tion service to its customers: mobileservice providers and MNOs, as well asmobile service users. As part of the casestudies, we have investigated business-to-business (B2B) and business-to-con-sumer (B2C) scenarios for QoSIS.net.We have shown that these scenarios arebeneficial for all parties in terms ofincreased revenue, increases in mobileservice quality and improvement ofmobile user’s experience [2].

The Quality of Life technologies groupat University of Geneva, Institute ofServices Science, is developing theQoSIS.net project. Future activitiesinclude deployment of QoSIS.net in themHealth area – benefiting mobilepatients and their caregivers by pro-viding QoS-assurance for health tele-monitoring and treatment services [3].

Links:http://www.qosis.com/http://www.qol.unige.ch/

References:[1] K. Wac, M. Hilario, B.J. vanBeijnum, et al: “QoS-PredictionsService: QoS Support for ProactiveMobile Services”, Wireless NetworkTraffic and Quality of Service Support:Trends and Standards, IGI GlobalPublisher, 2010[2] K. Wac: “Collaborative Sharing ofQuality of Service-Information forMobile Service Users”, PhD thesis,Information Systems, June 2009,University of Geneva, Switzerland.[3] K. Wac, M. Bargh, B.J. vanBeijnum, et al: “Power- and Delay-Awareness of Health TelemonitoringServices: the MobiHealth System CaseStudy”, IEEE JSAC, Special Issue onWireless and PervasiveCommunications in Healthcare, 27(4):525-536, IEEE Press, May 2009.

Please contact:Katarzyna WacUniversity of Geneva, SwitzerlandTel: +41 22 379 0242E-mail: [email protected]



Figure 1:

QoSIS service

coverage in

Geneva area

(background

colors indicate

performance

level, dots

indicate

separate

network cells)

(status: end of

2012)

Figure 2: Most

frequent users'

mobility paths

for which

QoSIS service is

available (dots

indicate

separate

network cells)

(status: end of

2012)


Modern mobile devices (smartphones,tablets, etc.) bundle a number of sensing,computing and networking resources,such as cameras, microphones, wirelessinterfaces and memory. The environ-ment is becoming increasingly saturatedwith pervasive devices (fixed camerasand sensor networks, for instance) thatalso feature computation, sensing andnetworking capabilities. As a conse-quence, the environment features a mul-titude of heterogeneous resources withdynamic availability due to factors suchas the users’ mobility. If orchestratedand managed through novel, appropriatecomputing paradigms, the availability ofsuch resources has the potential to sup-port innovative applications. The newmobile computing paradigm, “oppor-tunistic computing” [1] aims to addressthis vision.

Opportunistic computing assumes theexistence of a heterogeneous set of hard-ware and software resources contributedby users’ devices and by the devicesavailable in the environment. It com-poses and makes them available to theusers’ applications in a dynamic way,based on the current, situated needs ofthe applications and the mobility pat-terns of the users. Opportunistic com-puting is an evolution of opportunisticnetworking, a self-organizing net-working paradigm that enables commu-nication in dynamic pervasive networks.Each contact between mobile nodes isseen as an opportunity to forward con-tent towards final destinations, such thatend-to-end communication is supportedeven when simultaneous multi-hoppaths between sources and destinationsare not available. Opportunistic com-puting generalizes this approach, andviews contacts between nodes as oppor-tunities to exploit each other ’sresources, represented as service com-ponents. Ultimately, thanks to thedynamic composition of these compo-nents, applications can enjoy far richer

functionalities with respect to what isavailable in each individual node.

Developing the opportunistic com-puting concept and designing serviceplatforms for future self-organizing per-vasive networks is the goal of the EUSCAMPI project (Service Platform forSocial Aware Mobile and PervasiveComputing [2]), which started in 2010under the FP7 FIRE (Future InternetResearch and Experimentation) initia-tive. The conceptual view on oppor-tunistic computing of SCAMPI isdescribed by the logical architecture ofFigure 1.

The social layer at the bottom capturesproperties of users’ movements andtheir social networks. The structure ofsocial relationships between users canbe used to derive very good predictorsof mobility patterns and of communica-tion opportunities [1]. Contact opportu-nities between users naturally translateinto opportunities for utilization of theirresources. In the intermediate layer aresource can be a CPU, a shared portionof memory, a sensor, a network connec-

tion, a piece of content, a functionimplemented as a piece of code, etc. Alink between two resources in this layerrepresents the fact that these resourceshave a probability of “encountering”each other through opportunistic con-tacts, typically as a side effect of users’mobility. The opportunistic servicelayer (OSL) is responsible forabstracting resources into service com-ponents. It provides a functionaldescription of resources, their limits andcapabilities, and is responsible fororchestrating the composition of mul-tiple components into a single service,as required by the applications. Notethat this may require passinginput/output parameters between dif-ferent nodes providing different compo-nents, which is achieved through oppor-tunistic networking techniques.

One issue that we are investigating inSCAMPI is how information collecteddynamically on users’ devices can beused to describe their context, and ulti-mately to optimize service provisioningin opportunistic networks. To this end,we designed and implemented

Mobile Service Platforms Based on

Opportunistic Computing: The SCAMPI Project

by Marco Conti, Franca Delmastro and Andrea Passarella

Pervasive networking devices, including mobile devices, generate an environment saturated by

heterogeneous hardware and software resources. Novel mobile computing paradigms allow this

environment to be organized and orchestrated. “Opportunistic computing” is a new approach that

allows applications to take advantage of self-organizing services built, in a dynamic way, out of the

mobile resources that are available in pervasive environments.

Figure 1: Logical architecture of opportunistic computing

CAMEO, a middleware platformfocused on the management and elabo-ration of context information for oppor-tunistic computing environments [3].The architecture of CAMEO isdescribed in Figure 2, and is composedof two main building blocks. The LocalResource Management Framework(LRM-Fw), aims at implementing fea-tures related strictly to the interactionwith the local resources of the device,both hardware (eg, embedded sensors,capacity, battery, wireless interfaces)and software (eg, communication prim-itives and programming libraries). TheContext-Aware Framework (CA-Fw),aims at storing, elaborating and dissem-inating all the context information, andgathering a view on the resources avail-able on other devices, which can then becomposed according to the applica-tions’ needs.

In order to test the functionality ofCAMEO with real applications, wehave developed a Tourist-MSN (MobileSocial Networking) application [3],designed to improve the tourist experi-ence, for example, during a visit to acity, by collecting and sharing usefulinformation and content possiblydynamically elaborated and enriched byother users themselves. Figure 3 pro-vides some snapshots of the applicationinterface. Let us consider, for instance,a couple visiting Rome; before leaving,they plan to visit several attractions infew days trying to optimize their time.While they are moving around the citythey encounter other people that havejust visited some of those attractionsthey are interested in. Such users mayprovide useful information and/or mul-timedia content related to the attractionsthey are visiting. Tourist-MSN canexploit the opportunistic computingfunctionalities to dynamically re-schedule the tourists’ visit based on thecurrent conditions (eg, expected waitingtimes at queues). Moreover, it can allowits users to share each other’s resourcesin order to cooperatively generate mul-timedia content starting from elements(pictures, clips, video, etc.) they gen-erate during their visit. More generally,through CAMEO, applications likeTourist-MSN can provide the users withfunctionalities such as (i) identifyingusers in the social context interested in aspecific content, post or discussion; (ii)disseminating selected contents to inter-ested users; (iii) generating ratings of

available contents depending on thelocal user’s interests; (iv) establishingdiscussion forums with other users; (v)cooperatively annotating content andenriching it thanks to multimediaediting functionalities contributed bydevices available in the environment.

Link: http://www.ict-scampi.eu

References:[1] M. Conti et al: “From Opportunis-tic Networks to Opportunistic Comput-ing”, IEEE Communications Maga-zine, Vol. 48, Issue 9, 2010[2] M. Pitkänen et al: “SCAMPI: serv-ice platform for social aware mobileand pervasive computing”, ComputerCommunication Review 42(4): 503-508 (2012)[3] V. Arnaboldi, M. Conti, F. Delmas-tro: “Implementation of CAMEO: aContext-Aware Middleware for Oppor-tunistic Mobile Social Networks”,IEEE WoWMoM 2011, 20-23 June2011.

Please contact:Marco ContiIIT-CNR, ItalyTel: +39 050 315 3062E-mail: [email protected]

Franca DelmastroIIT-CNR, ItalyTel: +39 050 315 2405E-mail: [email protected]

Andrea PassarellaIIT-CNR, ItalyTel: +39 050 315 3269E-mail: [email protected]



Figure 3: Screenshot of Tourist-MSN application exploiting the CAMEO middleware

Figure 2: The CAMEO software architecture


With over 70% of the world’s populationexpected to be living in cities by 2050,the support of citizens’ mobility withinthe urban environment is a priority formunicipalities worldwide [1]. Althoughpublic multi-modal transit systems arenecessary to better manage mobility,they alone are not sufficient. Citizensmust be offered personalized travelinformation to make their journeys moreefficient and enjoyable. Notably, suchinformation should not only be objective(eg, bus timetable, live bus tracking), butcrucially personalized – since every pas-senger’s preferences and interests differ[2] (eg, crowdedness of trains, heat oftube platforms, sociability of thecoaches).

Goal and objectivesOwing to the recent proliferation ofsmartphones, the relatively new field ofmobile participatory sensing [3] couldbe leveraged towards providing a morefine-grained and up-to-date view of acity’s transportation system. A pan-European team including partners fromAlcatel/Lucent, Ambientic, Inria,Systematic, Thales, Transport forLondon (TfL) and University CollegeLondon (UCL), have pooled their

resources for TravelDashboard; aproject to produce an open source mid-dleware platform, enriched with person-alized mobility services for urban trav-elers, evaluated via real-life demonstra-tors’ assessments, and accompanied bynovel business models. The partnerswill tackle:

1. Development of a self-adaptive datacollection middleware, that gathersboth passive sensory information (eg,bus location from GPS) and activeuser-generated content (eg, road haz-ard reports, crowdedness of journey,etc.), and disseminates transport net-work status updates to travellers.

2. Development of personalized mobili-ty services, including: user prefer-ences about various aspects of mobil-ity (eg, punctuality, crowdedness, andsociability of transport systems); per-sonalized information aggregators, tocombine the data streams that are ofinterest to a user in their current con-text.

3. Development of a demonstrator forreal-life assessment, and deploymentwith actual end-users.

4. Elicitation of viable Business Mod-els, based on the services whose rapid

development and deployment aresupported by the middleware.

The middleware core is being workedon by Inria, Ambientic, and Thales; theplanned added-value services are beingdeveloped by UCL and Alcatel; thedemonstrator will be built and deployedunder the experience of Ambientic,thanks also to key input from TfL;finally, Systematic brings the know-how to elicit business models for inno-vative technological services, forAmbientic to deliver new urban trav-eller services.

From research to innovationThis 12 month project (January toDecember, 2013) is funded by theEuropean Institute of Innovation andTechnology (EIT)’s “ICT Labs” knowl-edge and innovation community (KIC)under the “Intelligent Mobility andTransportation Systems” action line.The core goal of EIT ICT labs is to spurinnovation in European ICT by lever-aging results from ongoing researchprojects – as presented below forTravelDashboard. This activity expandson them by conducting domain-focusedresearch, concretized through demon-strators, and by developing businessmodels specifically tailored to the urbanmobile participatory sensing domain.More precisely:• Inria leverages its work on large scale

mobile participatory sensing as partof the EC FP7 CHOReOS project onchoreographies for the future inter-net, applying it to the urban transportdomain, while Thales expands on thetransportation-related use cases that itis developing as part of CHOReOS aswell as the French ANR SocEDAproject on large distributed socialEvent Driven Architecture (EDA)platform, focusing on user mobilityand mobile device embeddingaspects.

• UCL builds upon the recommendersystem it has been developing as partof the FP7 i-Tour project on intelli-gent urban transport systems, addingqualitative metrics such as crowded-

Traveldashboard - a framework for the delivery of

Personalized Mobility Services to Urban Travellers

by Licia Capra, Pierre Chatel, Animesh Pathak and Roberto Speicys Cardoso

European researchers and companies join hands to address the woes of the urban traveller.

Figure 1: The TravelDashboard vision of urban transport

ness to its information filtering anddata mining services.

• Alcatel-Lucent is involved in anongoing project on Human MobilityPattern Mining, which combines dif-ferent sources of user mobility tracesto understand how communities areformed. In TravelDashboard, theirresearchers perform an empiricalstudy to understand mobility patternsand characteristics (time, mode, routeand price) of different user groups,and identify contextual factors thatdictate user mobility in order toimprove their user profiling system.

• Founded on its work on multi-plat-form multimedia streaming service inthe FP7 CONNECT project, the SMEAmbientic leads the work onenabling access to data streams gen-erated by mobile urban users, as wellas the development of a mobile appli-cation for urban transport.

To enable the transfer from research toinnovation, Systematic analyzes thebusiness value of the TravelDashboardsolution and identifies relevant business

models and strategies for the develop-ment and provisioning of personalizedmobility services in the urban crowd-sourcing context.

Impact: A Smart City for SmartTravellers The TravelDashboard framework will berealized by adhering to the principles ofService-Oriented Architectures. At itscore is the CHOReOS service-orientedmiddleware, atop which each partner isthen responsible for the development ofits own services. This core will bereleased as open-source software, thuspromoting the rapid and easy develop-ment of applications in support of highlypersonalized urban travel experiences.Both municipalities and transport opera-tors can use this platform to build smartapplications to help citizens navigatetheir cities. One demonstrator applicationwill be built and deployed during thisproject, catering to the London publictransport infrastructure, contributing tothe development of business models andstrategies for the provisioning of person-alized urban mobility services.

Link:TravelDashboard project: http://www-roc.inria.fr/arles/traveldashboard

References: [1] P. Bosnell: “Human Behaviours toMoving People More Intelligently”,2006[2] Rail Safety and Standards BoardLtd: “Topic Note on Travel Behaviourand Behavioural Change”, 2010[3] N. Lane, E. Miluzzo, H. Lu, D.Peebles, T. Choudhury, and A.Campbell: “A survey of mobile phonesensing”, Communications Magazine,IEEE, 2010.

Please contact:Pierre ChatelThales Group, FranceTel: +33 169415565E-mail: [email protected]



“Crowd-sensing” refers to the involve-ment of a large, diffuse group of partici-pants in the task of retrieving reliabledata from the field. This approach hasbeen used in multiple research studies,including traffic and road monitoring,social networking and environmentalmonitoring.

However, developing a dedicatedsensing application to collect a specificdataset over a target population presentsa real challenge [1]. Indeed, whenbuilding an efficient applicationinvolving a large community of partici-pants, a number of vital issues must betaken into account, including incentivemechanisms, recruitment models,energy limitations and privacy concerns.A thorough expertise in the area ofmobiles device technology is essential toaddress these issues. Consequently, it

can be difficult for scientists, inexpert inthis field, to collect realistic datasets fortheir studies. But more importantly, thead hoc applications that are developedmay neglect privacy and security con-cerns, resulting in the disclosure of sen-sitive user information, or even theadoption of unethical or illegalapproaches.

The state-of-the-art therefore lacksreusable solutions for safely collectingand exploiting crowd activity traces,which are usually difficult to set up andare tied to specific data representationsand device configurations.

In this context, the Inria ADAM projectteam has developed APISENSE [2], aplatform targeting multiple researchcommunities, and providing a light-weight solution to build and deploy

crowd-sensing applications for col-lecting experimental datasets.

The APISENSE PlatformThe APISENSE platform distinguishesbetween two roles. The first, called sci-entist, is typically a researcher whowants to define and deploy an experi-ment over a large population of mobileusers. The platform therefore providesan online environment (as a software asa service) allowing the researcher todescribe experimental requirements viaa scripting language, to deploy theexperiment scripts over a subset of par-ticipants and to connect other servicesto the platform in order to extract andprocess the collected datasets (eg, visu-alization, analysis, replay). Technically,the server-side infrastructure ofAPISENSE is built on the principles ofCloud computing in order to offer a

APISENSE: Crowd-Sensing Made Easy

by Nicolas Haderer, Romain Rouvoy, Christophe Ribeiro and Lionel Seinturier

The rapid emergence of mobile devices, such as TabletPC and smartphones, equipped with a rich array

of sensors, enables a new means of acquiring sensor data, known as crowd-sensing. Crowd-sensing is

currently receiving a lot of attention, not only from industry but also from various research communities

interested in collecting a new class of data over a much larger population than was previously possible.


modular service-oriented architecture,which can be customized to suit the sci-entist’s requirements. The second role isthe mobile phone user, identified as aparticipant. The APISENSE platformprovides the participant with a mobileapplication allowing experiments to bedownloaded and executed in a dedi-cated sandbox, and the collecteddatasets to be automatically uploadedon the APISENSE server.

The APISENSE LanguageTo facilitate its adoption by scientists,APISENSE adopts standard scriptinglanguages and provides an extension ofthe JavaScript, CoffeeScript and Pythonlanguages, as an efficient means ofdescribing an experiment without anyspecific knowledge of mobile deviceprogramming models (eg, AndroidSDK).

Figure 1 shows how simple it is todefine a sensing task to collect a spe-cific dataset in the wild and automati-cally build an open data map from col-lected data. The sensing task is definedusing CoffeeScript language, which istriggered whenever the location of aparticipant changes by a distance of 10m in a period of five minutes. Whenthese conditions are met, the script auto-matically builds a trace containing thelocation of the participant and attachesGSM networks characteristics. Thesensing task can be defined by the sci-entist using a web interface and a singleclick makes it available to participants.Figure 2 depicts the resulting open map

generated from datasets collected byonly three participants and displayingthe GSM signal strength of a given net-work operator in the area of theUniversity Lille 1.

The scripting library supports a widerange of features to define data col-lected during a sensing experiment,including traditional sensors proposedby smartphone technology, such asGPS, Bluetooth, accelerometer, com-pass, phone call, SMS, applicationstatus (installed, running), for oppor-tunistic sensing activities and also agraphical user interaction framework todescribe user surveys in the case of par-ticipatory sensing activities.

Ongoing WorkOur current work focuses on thedeployment of various large scalesensing experiments for building a col-laborative method i) to identify Androidmalware in the wild and ii) to deploymicro-seismic mobile stations for earth-quake monitoring.

Additionally, APISENSE collaborateswith the MetroScope initiative, sup-ported by Inria, to provide a new gener-ation of instruments for observingInternet usage. By providing tools,datasets, and analyses that make sensefor different disciplines, APISENSE notonly contributes to the theoreticalknowledge of the Internet, but alsointroduces a greater level of trans-parency by involving citizens andgiving them direct feedback.

Links:APISENSE website: http://apisense.frADAM project-team:http://adam.lille.inria.frMetroScope project:http://metroscope.eu

References:[1] N.D. Lane et al: “A survey ofmobile phone sensing”,Communications Magazine 48(9),IEEE. September 2010[2] N. Haderer et al: “A preliminaryinvestigation of user incentives toleverage crowd-sensing activities”, in2nd International IEEE Workshop onHot Topics in Pervasive Computing(PerHot), March 2013.

Please contact: Romain RouvoyADAM project-team, University Lille 1 - LIFL, Inria Lille - Nord Europe, FranceTel : +33 3 59 35 87 77E-mail: [email protected]

Figure 1: Description of a sensing task (in

CoffeScript) Figure 2: GSM open map generated from collected datasets

It is no surprise, given smartphones’ con-venience and utility, to see their wideadoption worldwide. Today, there are1.08 billion smartphone users out of atotal of five billion mobile phone usersworldwide, and the ratio is constantlyincreasing. Smartphones are used notonly to communicate but also to browsethe web and run various internet-enabledApps. As a result, they contain a lot ofinformation about the cyber activities oftheir owners, and therefore users’ inter-ests and behaviours. Furthermore, smart-phones are also equipped with GPS, NFCand Bluetooth units, along with a digitalcamera and are almost always connectedto the Internet; thereby revealing alot of information about the phys-ical activities of their owners. Ontop of this, smartphones are verypersonal to the user and arebarely turned off.

For the aforementioned reasons,combined with the fact that userstend to carry smartphones wher-ever they go, they are an idealtarget for marketers who want toprofile users to profit from theirpersonal data. Some studies evensuggest that the main businessmodel for some developers (inthe case of free Apps, forexample) is based on the collec-tion of personal data. As a result,many Apps might be leakingpersonal information to thirdparties, such as Analytics andAdvertising (A&A) companies.

A few insightsThe goals of the Mobiliticsproject are to investigate smartphoneOperating Systems (OSs) and Apps forpotential privacy leaks and to informtheir users about the privacy risks. Theproject currently targets two OSs,namely Android and iOS, because theycover almost 75% of the whole smart-phone OS market share.

As part of this project, we have devel-oped a software solution (an Androidversion with similar functionalities isunder development) for iOS to captureaccess to private information by var-ious Apps. When an App makes a callto the iOS API to access a broad list of auser's personal data, eg Contacts,Location, Device Name, UDID,Calendar, Reminders, Photos, Notesand Accounts, our software logs thisevent for later analysis. Note that someApps do actually need to access per-sonal data to provide the desiredservice. These applications do notbreach user privacy if they only process

and use the personal data to provide thedesired service and don’t transmit thedata to remote third parties. In order todetect personal information leakage,we also monitor whether the accessedpersonal data is sent to a third party, asin [1] and [2] but by using a differentapproach. Additionally, we are alsodeveloping a visualization tool to helppeople understand the privacy implica-tions by aggregating, interpreting anddisplaying all private data stored and/orsent by various Apps.

For instance, our iOS tool reveals thatmany Apps are accessing the Unique

Device ID multiple times (in theorder of hundreds), whichimplies that it is probably beingused for online tracking of theuser. Some Apps are also, sur-prisingly, accessing the user'sdevice name although theredoes not seem to be any obviousreason to do so. The name of thedevice is set during the initialdevice setup and often containsthe real name of the user.Moreover, even if the user doesnot set it to his or her real name,it might easily be used fortracking purposes since thedevice owner does not generallymodify it after the initial setup.

Our software solution wasdeveloped in 2012 for iOS 5.xbefore Apple launched iOS 6 inSeptember 2012. iOS 5.x didn’tseek the user’s permission forprivate data access except for

location information. In iOS 6,Apple decided to change its strategyand introduced a new privacy-specificsetting giving the user control overwhether an App can access private data:the user is prompted the first time anApp tries to access Contacts, Location,Reminders, Photos, Calendar andSocial Networking accounts and later,



Mobilitics: Analyzing Privacy Leaks

in Smartphones

by Jagdish Prasad Achara, Franck Baudot, Claude Castelluccia, Geoffrey Delcroix and Vincent Roca

Who, do you think, is aware of almost everything you do? Well, it’s probably right there in your pocket,

if you own a smartphone and carry it with you. In order to evaluate the actual privacy risks of

smartphones and to raise public awareness of these risks, the CNIL (French data protection authority)

and the Inria (French public science and technology institution dedicated to computational sciences)

Privatics team started working together in 2012 as part of the Mobilitics project.

Figure 1: Android and iOS currently don't

provide any mechanism to let users know how

their personal information is being used by

various Apps. Will it be used locally on the

device or sent to remote servers? Being aware

of it, users can probably make better decision

whether to allow/deny access to their personal

information for a particular App.

iOS remembers and follows the userpreferences.

In our opinion, this is a decent step byApple towards making iOS privacy-friendly. However, several questionsstill remain open: is the list of privatedata included in their privacy-settingssufficient? Is an authorization that doesnot consider any behavioural analysissufficient? For instance, accessing thedevice location upon App installation,to enable a per-country personalization,is not comparable to accessing the loca-tion every five minutes. Also, does theApp keep the personal informationlocally for internal purposes, or is itcommunicated to external servers? Ifthe latter, where exactly are theseservers? Moreover, A&A librariesincluded by the App developer alsohave access to the same set of user’s pri-vate data as the App itself. However, auser giving access to his or her Contacts

doesn’t necessarily indicate consent forthese data to be shared with A&A com-panies. Might this pave the path for pri-vacy invasion? These are the questionsthat Mobilitics will attempt to answer.

ConclusionOur preliminary results and the variousscandals that occured in 2012 show thatprivacy considerations are of utmostimportance if we want to continue usingthese devices with serenity. We believethat smartphones can’t be, in the longrun, black boxes to their owners becausenobody wants these great devices in ourpockets to be the ultimate spy.

Mobilitics is a CNIL-Inria project thatinvolved the following participants:Jagdish Prasad Achara, Franck Baudot,Claude Castelluccia, Geoffrey Delcroix,James Douglas Lefruit, Gwendal LeGrand, Stéphane Petitcolas, and VincentRoca.

Links:https://team.inria.fr/privatics/mobilitics/http://www.go-gulf.com/blog/smartphone/ http://kwz.me/04http://blogs.wsj.com/wtk-mobile/ http://kwz.me/00

References:[1] W. Enck et al: “TaintDroid: AnInformation-Flow Tracking System forRealtime Privacy Monitoring onSmartphones”, in proc. of OSDI 2010[2] M. Egele et al: “PiOS: DetectingPrivacy Leaks in iOS Applications”, inproc. of NDSS 2011.

Please contact: Jagdish Prasad Achara, ClaudeCastelluccia, Vincent RocaInria, FranceTel : +33476615215E-mail:{Jagdish.Achara,Claude.Castelluccia,Vincent.Roca}@inria.fr


Privacy-Preserving Interest-Cast

for Android Smartphones

by Gianpiero Costantino, Fabio Martinelli and Paolo Santi

We present an implementation of the FairPlay framework for secure two-party function

computation on Android smartphones, which we call MobileFairPlay. Our application was

developed to preserve the users’ privacy within opportunistic networks considering the

interest-casting model. Our tests show that the running times of the protocol on several

Android phones, are very reasonable (up to five seconds in the worst case).

Crowded places present an opportunityfor people to share personal information.In addition to sharing informationthrough traditional, web-based plat-forms and applications such asFacebook and Twitter, the availability ofshort range radio interfaces in smart-phones, tablet PCs, etc. allows individ-uals to share information with oneanother through direct, opportunisticcommunication (typically using theBluetooth or WiFi interface).

This model of store-carry and for-warding data to others is known asopportunistic networking (OppNets). Acommon feature of these approaches isthat, before making a decision aboutwhether to share information with anindividual, users have to exchange somesensitive information, such as history ofpast encounters [1], interest profiles, etc.

Given that the person encountered isgenerally a stranger, this exchange ofsensitive information (which occurs inplain text in the approaches mentioned)is likely to be deemed unacceptable bythe user in real-world scenarios, owingto privacy concerns.

To address this issue, we present a fea-sible implementation of a cryptographicframework for secure multi-party com-putation (the FairPlay framework pro-posed in [2]) targeted to the interest-castmodel and running on the Androidmobile platform. Our application,“Mobile-FairPlay” [3], has been devel-oped with the aims of: 1) finding peoplein the user’s (Alice) neighbourhoodthrough a Bluetooth scan operation, 2)connecting to another user (Bob) anddetermining whether Bob and Alicehave similar interest profiles without

disclosing sensitive information, and 3)sharing messages between Alice’s andBob’s devices in the event that their pro-files are similar.

When Alice and Bob have established anew connection, Bob, who received theconnection, randomly selects differenttopics to verify their similarity withrespect to these interests. Then, theystart matching interests using the secureframework implemented in the App.During this execution, both Bob andAlice use their own value for theselected topic, extracted from theinterest profile. However, these valuesare not sent to the other participants inplain, but are encoded in the garbledBoolean circuits exchanged throughMobileFairPlay. At the end of the hand-shaking phase, Alice and Bob onlyknow the result of the jointly computed



matching, without knowing the specificinterest values of the other party. In thecase of a positive comparison, the userwho received the incoming connectioncan start sending their own informationusing real files.

The execution time of the interestmatching performed with our applica-tion ranges between 2 to 2.5 seconds fora single topic comparison and between3.5 to 4.5 seconds for four topic compar-isons. The running time is dominated bythe cryptographic part run during thematching, and depends mainly on thesmartphone’s hardware. Times reportedhere are obtained using Samsung GalaxyS2 and Samsung Galaxy S and otherslower devices. However, by runningmore recent smartphones the computa-tion time can be even lower.

We are currently developing a differentversion of this application that uses pastusers’ locations to determine whetherthey have spent time in common places.

This matching is also performed whilstpreserving the users’ privacy. Commonpast locations are used to understandsimilarity in users’ behaviour, and takingadvantage of this can help users toincrease their social contacts.

The application was released late spring2012, and is available at the link below.

AcknowledgmentIt took over six months to develop thisapplication, and our thanks go to DarioAmoruso, who prepared his masterdegree thesis on this topic.

Most modern mobile OperatingSystems (OSs)rely on application mar-ketplaces, also called app stores, toallow users to download and installsoftware packages that extend theirdevices with new functionalities. Thesuccess of this software distributionparadigm is confirmed by the gargan-tuan number of mobile applicationsreleased and installed every day.

Despite the enormous advantages interms of customizability, access tomany third party software packagesraises several security issues. One majorissue is how to provide fine grained,strong security guarantees without com-promising usability and scalability.Most mobile OSs don’t apply security-relevant settings automatically, criticalchoices about security aspects have tobe made by the user. For instance,Android requires that users accept theapplication contract, or “manifest”, at

installation time. Also, the marketplacecollects positive and negative feedbackthat serves as a reputation system thatcustomers can consider beforeinstalling new software.

Nevertheless, these mechanisms do notoffer proper protection to users. Indeed,manifest descriptions are extremelycoarse grained, eg, “this app will usethe filesystem”, and even expert usersstruggle to find a relationship betweena manifest and their security needs.Recent studies have demonstrated thatusers generally disregard security-rele-vant warnings upon installation ofapplications, thus ignoring potentialsecurity risks.

Furthermore, once a user has installedan application, it can run freely andindependently of the context, ie thedevice configuration, under which it isused. For instance, current mobile OSs

have no mechanism to prevent a userfrom playing games or executing appli-cations that may spread sensitive infor-mation during business hours.Consequently, organizations that couldbenefit from their employees having asmartphone, eg, because the companycould avoid purchasing multiple spe-cific devices, may be reluctant to usethem.

To cope with these issues, we recentlyproposed (see [1,2]) a novel approachthat provides formal security guaran-tees to the users of mobile devices.Furthermore, the system is built on topof the existing mobile applications dis-tribution paradigm, which makes thewhole system scalable and reusable.

In practice, a mobile application mar-ketplace works as follows: Owners ofmobile devices register to a market-place by means of access credentials.

Market-Based Security for Mobile devices

by Gabriele Costa, Alessio Merlo and Luca Verderame

We present a security-enabled application marketplace that provides formal security guarantees to the

existing mobile software distribution paradigm. Our proposal allows users and organizations to apply

fine-grained security policies on top of the existing market-based software deployment with no need for

invasive customization of devices and without compromising the system scalability and usability.

Link:http://www.iit.cnr.it/staff/gianpiero.costantino/

References: [1] E. Daly and M. Haahr: “Socialnetwork analysis for routing indisconnected delay-tolerant manets”, inACM Mo- biHoc [2] D. Malkhi et al: “Fairplay: a securetwo-party computation system”, inproc. of the 13th conf. on USENIXSec. Symp. [3] G. Costantino et al: “Animplementation of secure two-partycomputation for smartphones withapplication to privacy-preservinginterest-cast”, PST 2012.

Please contact:Gianpiero Costantino, Fabio Martinelli and Paolo Santi IIT-CNR, Pisa, ItalyE-mail: [email protected],[email protected],[email protected]

http://www.iit.cnr.it/staff/gianpiero.costantino/


The marketplace stores device informa-tion and configuration, eg, hardwareprofile, OS version and installed appli-cations. When the user decides toinstall a new application, the market-place checks the compatibility with thedevice configuration and requests theuser to provide the application with therequested access rights. If the user dis-agrees, the installation is cancelled,otherwise the marketplace sends thesoftware package and the deviceinstalls it.

We propose the use of a securityenabled marketplace that acts as a secu-rity proxy, providing a formal securityguarantee. A secure marketplaceexploits a security policies manage-ment system and verifies whether a cer-tain application can be installed on adevice without affecting the securityconfiguration. Application packagesare analyzed to extract security con-tracts that, in turn, extend and enrichthe manifest. Contracts are safe repre-sentations of the application’s behav-iour, and denote all the possiblesequences of system access operationsthat the software can perform(including possible synchronizationswith other installed applications).

The secure marketplace then applies amodel checking procedure to verifywhether the device configuration com-posed with the application contract isstill policy-compliant. If it is, the appli-cation is labeled as safe. Otherwise, the

application can be modified by instru-menting its code with security checks,guaranteeing that it respects the policy.In both cases, the secure marketplacegenerates a safety proof that the usercan verify, through automatic proofchecking, before installation.

After installation confirmation, thesecure marketplace updates the policymanagement system with the new secu-rity state of the device.

Our proposed system could work withmost of the common usage contexts formobile devices. Indeed, it appliesequally to both private customers, ie,users wanting to apply security controlsto their own devices, and companies,ie, organizations in which employeesand affiliated persons must respect pre-cise security restrictions.

We are currently developing a realimplementation of our secure market-place that customers can use to mediatethe access to standard marketplaces, eg,Google play.

We plan to release a secure marketplaceimplementation for Android OS in mid-2013. Moreover, we are currentlyinvestigating potential extensions ofour proposal; including the adoption ofsecurity monitoring to allow the user toinstall applications that statically vio-late the policy. In this way, applicationexecutions are guaranteed to respectthe security policy.

References:[1] A. Armando, G. Costa, A. Merlo:“Formal Modeling and Reasoningabout the Android SecurityFramework”, in proc. of TGC 2012,Newcastle upon Tyne, UK, 2012http://www.ai-lab.it/merlo/publications/AndroidModel.pdf

[2] A. Armando, G. Costa, A. Merlo, L.Verderame: “Bring Your Own Device,Securely”, in proc. of SAC 2013,Coimbra, Portugal, 2013http://www.ai-lab.it/merlo/publications/BYODroid-extended.pdf

Please contact:Gabriele CostaUniversity of Genova, ItalyTel: +39 0103536545 E-mail: [email protected]

Alessio Merlo E-Campus University, ItalyTel: +39 0103532344E-mail: [email protected]

Figure 1: Architecture of the secure marketplace

http://www.ai-lab.it/merlo/publications/AndroidModel.pdf

http://www.ai-lab.it/merlo/publications/ BYODroid-extended.pdf



Wi-Fi protocol includes potentialsources of personal information leaks.Wi-Fi enabled devices commonly useactive discovery mode to find the avail-able Wi-Fi access points (APs). Thismechanism includes broadcast of theWi-Fi network’s names to which themobile device has previously been con-nected, in plain text, which may beeasily observed and captured by any Wi-Fi device monitoring the control traffic.The combination of the network namesassociated with any single mobile devicecan be considered as a Wi-Fi fingerprintwhich can be used to identify the user towhom the mobile device belongs. Ourresearch [1] investigates how it is pos-sible to exploit these Wi-Fi fingerprintsto identify links between users, ieowners of the mobile devices broad-casting such network names.

From June to October 2012, we col-lected data using a laptop running Wi-Fimonitoring tools in the city of Sydney.Overall, Wi-Fi fingerprints of more than8,000 devices were collected and wefound that some devices were revealingtheir associations to more than 80 Wi-Finetworks. In order to test our hypothesis(social links can be inferred based on theWi-Fi fingerprint) we collected the Wi-Fi fingerprint, as well as the existingsocial links, from a group of volunteers.

Identifying linked individualsOur approach is based on the similaritybetween Wi-Fi fingerprints, which isequated to the likelihood of the corre-sponding users being linked. Whencomputing the similarity between twoWi-Fi fingerprints, two dimensions needto be considered: • The number of network names in

common. Indeed, sharing a network isan indication of the existence of alink, eg friends and family that sharemultiple Wi-Fi networks.

• The rarity of the network names incommon. Some network names arevery common and sharing them doesnot imply a link between the users.This is the case for public networknames, for instance McDonalds FreeWi-Fi, or default network names suchas NETGEAR and linksys. On theother hand, uncommon networknames such as Griffin Family Net-work or Orange-3EF50 are likely tobe associated with a strong linkbetween the users of these networks.

Utilizing a carefully designed similaritymetric, we have been able to infer theexistence of social links with a highconfidence: 80% of the links weredetected with an error rate of 7%.

Who should worry about it?Owners of smartphones are particularlyexposed to this threat, as these devicesare carried on persons throughout theday, connecting to multiple Wi-Fi net-works and also broadcasting their con-nection history.

What can be done to prevent thelinking?There are a number of industry andresearch initiatives aiming to addressthe Wi-Fi related privacy issues. The

deployment of new technology, ie pri-vacy preserving discovery services,would necessitate software modifica-tions in currently deployed APs anddevices. The obvious solution - to dis-able active discovery mode - comes atthe expense of performance andusability, ie it would take longer for theWi-Fi capable device to find and con-nect to an available AP. As a possiblefirst step, users should be encouraged toremove obsolete connection historyentries, which may lower the similaritymetric and thus reduce the ease oflinkage.

Links: http://www.inria.fr/en/teams/privaticshttp://www.nicta.com.au/research/projects/trusted_networking

Reference:[1] M. Cunche, M.A. Kaafar, and R.Boreli: “I know who you will meet thisevening! linking wireless devices usingwi-fi probe requests”, WoWMoM 2012IEEE International Symposium, p. 1-9,June 2012.

Please contact: Mathieu CuncheINSA-Lyon and Inria, FranceE-mail: [email protected]

Revealing Social Links Between Owners

of Wi-fi Enabled Smartphones

by Mathieu Cunche, Mohamed Ali Kaafar and Roksana Boreli

Wi-Fi technology, available in the vast majority of mobile phones, tablets, laptops and other

computing devices that we use in our daily lives, has enabled widespread use of new

applications and services. This technology, however, has a number of issues related to privacy

loss, exacerbated by its ubiquity. Our research shows how the information freely transmitted by

the Wi-Fi protocol can be used to identify links between people, ie whether they are family,

friends, colleagues etc.

Figure 1: A smartphone

broadcasting in plaintext

its Wi-Fi fingerprint, ie

names of networks to

which it has previously

been connected.

© In

ria

- Pho

to H

.Rag

uet

http://www.nicta.com.au/research/projects/trusted_networking


Smartphones (or mobile computingdevices) are proliferating globally.According to a study by comScore [1],by December 2012, more than 125 mil-lion people in the USA alone ownedsmartphones. These mobile devices pro-vide a variety of networking options suchas GSM, GPRS, Bluetooth and Wi-Fi.Therefore, smartphones are becomingthe device of choice for people to collab-orate with family members, friends andbusiness colleagues/customers. Theyenable us to communicate not only bytelephone, by email and short messages,to play games, to share information,organize videoconferences, and coordi-nate business events.

The adoption of collaboration viasmartphones relies on critical factorssuch as network coverage and securityconcerns. Collaboration may involvemobile users located in the same limitedgeographical area, yet connected viawireless channels, which can be lessreliable and offer lower data rates thantraditional wired ones. Moreover, theamount of personal/corporate datastored in mobile devices has increased.Sharing such data requires preventivesecurity and privacy mechanisms toprotect and regulate access to thisshared data by other users.

The CASSIS project team at InriaNancy Grand-Est has designed anddeveloped a decentralized and secureshared calendar, which can be easilydeployed on ad-hoc networks of smart-phones [2]. Unlike shared calendarssuch as Google Calendar, our collabo-rative application ensures privacy andconfidentiality of calendar events sinceit is independent of any third-party orbig company servers.

The calendar enables users to sharetheir calendar events with selected usersin a dynamic group (users can join orleave the group at any time), and is asresponsive as a personal calendar. Thishigh responsiveness is achieved byreplicating the shared calendar at eachparticipating user. The consistency ofthese replicas is maintained in a decen-tralized way (without a central coordi-nation point) using OperationalTransformation (OT) approach.

To prevent unauthorized access byillegal users, our shared calendar isendowed with a flexible access controlmechanism where each user can definean authorization policy for controllingaccess to its proper calendar events.Furthermore, all authorization policiesare replicated at the local memory of

each user in order to ensure low latencyfor access rights checking. Thus, a userwill own two copies: the shared cal-endar and the access data-structure. It isclear that this replication allows forhigh data availability since when userswant to read or update the shared cal-endar, this manipulation will be grantedor denied by controlling only the localcopy of the access data-structure. Dueto the out-of-order execution of theshared calendar's updates and theauthorization policy's updates, we haveused an optimistic approach that toler-ates momentary violation of accessrights but then ensures the copies to berestored in valid states with respect tothe stabilized access control policy.

Figure 1 shows the components of ourdecentralized and secure shared cal-endar: a user locally manipulates his orher copies (the shared calendar and itsauthorization policy), and the remoteupdates from other users (arriving bymeans of ad-hoc networks) areprocessed. When a user intends toupdate his or her local copy of theshared calendar, this update will begranted or denied by checking the localcopy of the policy. Once granted andexecuted, the local update is processedby coordination and access control

Secure Collaboration for Smartphones

by Abdessamad Imine and Michaël Rusinowitch

Designing secure collaborative applications has become a hot topic in the area of mobile devices,

posing challenging problems such as data management and security.

Figure 1:

Architecture of our

shared and secure

calendar



modules in order to be wrapped bysome meta-information, which are nec-essary to enforce the calendar consis-tency and its authorization policy, andnext it is broadcast to other users. As forthe remote calendar update, the accesscontrol module checks whether or notthis update is authorized and, if granted,sends it to the coordination module forfurther processing to maintain the cal-endar consistency before its immediateexecution on the local copy of theshared calendar. Note that individualusers administer their own proper cal-endar events. Thus, a user can define anauthorization policy on which accessrights for other users can be specified.

When an administrator modifies a localpolicy by adding or removing authori-zations, this modification is sent toother users in order to update their localpolicy copies.

Our research work is funded by theEuropean project FP7 NESSoS.

Links:http://www.loria.fr/~imine/tools/home.htmNESSoS: http://www.nessos-project.eu

References:[1] comScore, comScore reportsDecember 2012, U. S. smartphonesubscriber market share, press release,February 6, 2013[2] J. Achara, A. Imine and M.Rusinowitch: “DeSCal - DecentralizedShared Calendar for P2P and Ad-HocNetworks”, in IEEE ISPDC 2011,Cluj-Napoca, Romania, 2011, pages223-231.

Please contact:Abdessamad ImineUniversity of Lorraine and InriaNancy-Grand EstE-mail: [email protected]

Navigation throughout vast complexbuildings, eg, hospitals or school build-ings, can be difficult even if the buildingis densely equipped with navigationsigns. For acquiring location coordi-nates, the GPS can be used. This tech-nique is commonly used in transporta-tion and these days also in mobilephones to calculate coordinates on a vir-tual map. Although GPS tracking can beuseful outdoors, it cannot be usedindoors, because there is no direct visi-bility to GPS satellites.

Fortunately, there are other means forcurrent smartphones to perform local-

ization. They are equipped with varioussensors such as magnetometer, linearaccelerometer, gyroscope, and wirelessnetwork adapters. Even a low-endsmartphone contains at least a magne-tometer and an accelerometer.Therefore, there is now a unique oppor-tunity to realize a solution for indoorlocalization.

At Masaryk University in Brno we rec-ognized that opportunity, and have beendeveloping an indoor localization formobile devices. For our project we haveselected the Google Android platform,which offers well documented platform

tools enabling easy access to sensordata. In addition, developments forAndroid are mainly based on the Javaprogramming language. Therefore, var-ious already programmed libraries canbe used.

Our localization system is based onalgorithms published in [1, 2, 3]. Thedeveloped solution consists of threelocalization techniques merged togetherto provide more accurate results. Thefirst technique is indoor tracking basedon wireless networks. As presented in[1], we use the Received SignalStrength (RSS) fingerprinting to create

Indoor User Localization Using Mobile devices

by Jonáš Ševčík

The ability to easily access your location along with additional relevant information has emerged

over the last six years with the rapid development of the smartphone industry. The trend of using

mobile phones to get a sense of location is now one of the main features of these devices. Most

outdoor tracking is done by the Global Positioning System (GPS), but what are one’s options if one

is indoors, where you cannot use the GPS signal?

Figure 1: Recalibrating location in a library

Figure 2: Map view with

filter data


a database of unified access point iden-tifiers and their received signal strengthwhich is mapped to location coordi-nates. RSS measurements at anunknown location are then compared tothe signal strength maps to estimate thereceiver’s location.

The second technique is dead reckoning– a process of calculating the currentposition by using previously deter-mined coordinates, which are advancedby known speed and course. This tech-nique is well known from air naviga-tion. We use dead reckoning in an iner-tial navigation system, where the initiallocation is determined via the formertechnique. Alternatively, coordinatescan be encoded into a barcode, whichwhen placed into the position of thosecoordinates, can be scanned by acamera and then decoded. Positiontracking is done by calculating anumber of steps. Step length is esti-mated using a neural network as pre-sented in [2]. The course of steps isdetermined by a gyrocompass. Sincedead reckoning is subject to cumulativeerrors, we use barcodes for recalibratinga position (Figure 1).

The third technique is Sequential MonteCarlo filtering [3], which uses particlesevenly spread in the probable location

(as shown in Figure 2 by the blue rec-tangle) determined by a RSS fingerprintdatabase. These particles are set inmotion with events generated by stepdetection. Then these particles whosehypothetical motion leads throughimpassable obstacles, eg walls, areeliminated. This improves location esti-mation.

The indoor localization prototype iscapable of displaying the position of auser on a floor map. The basic viewoffers display of walls, doors, stair-cases, and elevators. Additional infor-mation can be displayed using a set oflayers, which contains points of interestsuch as room numbers, positions ofvending machines, ATMs etc. Layerscan be applied according to the needs ofthe user. This system is useful inunknown environments, providingusers with crucial information abouttheir location and helping them tracktheir movement.

In a further development, we are plan-ning to implement navigation to aselected point on the map. We will alsoextend view modes by introducing anaugmented reality, which will allowusers to perceive camera imagesenhanced by information about theirsurroundings.

The indoor localization prototypedemonstrates that current smartphonesare capable of difficult real-time calcula-tions and that it is possible to implementa sophisticated localization systemwithout using any additional hardware.This system provides guidance crucialfor orientation in unknown buildings,where standard maps cannot be used.

References:[1] A. Chen et al: “An algorithm forfast, model-free tracking indoors”,ACM SIGMOBILE MobileComputing and CommunicationsReview, ISSN 1559-1662, vol. 11 no.3, 2006[2] S. Cho: “MEMS Based PedestrianNavigation System”, Journal ofNavigation, ISSN 0373-4633, vol. 59,pp. 135–153, Royal Institute ofNavigation, 2006[3] M. Arulampalam et al: “A Tutorialon Particle Filters for OnlineNonlinear/Non-Gaussian BayesianTracking”, IEEE Transactions onSignal Processing, ISSN 1053-587X,vol. 50 no. 2, 2002.

Please contact:Jonáš ŠevčíkMasaryk University - CRCIM, CzechRepublicE-mail: [email protected]

Location information has become anintegral part of the mobile user’s dailylife and Google is already reporting thata third of searches on portable electronicdevices refer to spatially relevant con-tent, while 94% of smartphone usershave searched for local information.This is mainly due to the proliferation ofpowerful smartphones featuring on-board GPS sensors. According to TomiAhonen Consulting, there are already1.2 billion smartphone devices in useworldwide, and 770 million of these are

equipped with GPS. As one mightexpect, the most successful Location-Based Service (LBS) is vehicle naviga-tion and pedestrian guidance to pointsof interest in outdoor areas.

There has, however, been a recentincrease in interest in location-awarecontent specific to large indoor environ-ments, such as shopping malls,museums, exhibition centres, confer-ence venues and airports. Statistics byStrategy Analytics indicate that people

spend 80-90% of their time insidebuildings, while 70% of cellular callsand 80% of data connections originatefrom indoors. This has triggered anincreasing interest in indoor applica-tions, such as in-building navigation,asset tracking, elderly support forAmbient and Assisted Living (AAL)and others. Since GPS has low avail-ability indoors due to the blockage orattenuation of the satellite signals, alter-native geolocation solutions arerequired.

Airplace: Indoor Geolocation on Smartphones

Through Wifi fingerprinting

by Christos Laoudias, Georgios Larkou, Demetrios Zeinalipour-Yazti and Christos G. Panayiotou

The wide availability of location data is undoubtedly reshaping the entire smartphone ecosystem

with the advent of innovative location-aware services and applications. To accelerate their adoption

by the public, the Airplace positioning system is a location-enabling platform that aims to deliver

highly reliable and accurate location information right where the highest demand is anticipated in

the near future: inside buildings.



At the University of Cyprus, we appre-ciate the benefit of indoor LBS and ourgoal is to facilitate their wide accept-ance. Thus, we focus on the provision ofGPS-free location data inside buildingsby leveraging location-oriented infor-mation collected with the smartphone’sbuilt-in sensors, such as WiFi,accelerometer, gyroscope and digitalcompass. The Airplace indoor posi-tioning system [1] is a collaborativeresearch effort between the KIOSResearch Center and the Department ofComputer Science at the University ofCyprus that was initiated a few yearsago. Airplace is developed on popularAndroid smartphones and the unknownuser location is determined with the useof Received Signal Strength (RSS) datafrom the surrounding WiFi infrastruc-ture, ie, wireless Access Points (AP).

This approach is very convenient andextremely suitable for indoor applica-tions running on the users’ smartphones.Firstly, the existing WiFi APs, alreadydeployed in indoor environments forwireless connectivity, are exploited thusavoiding the installation of costly dedi-cated positioning equipment, such ascustom transmitters and antennas.Secondly, we rely on RSS dataextracted through passive scanning ofthe WiFi beacon packets transmitted byneighbouring APs as part of the stan-dard network functionality. Thus, com-mercial mobile devices, which are usu-ally equipped with WiFi adapters, canbe used without any hardware modifica-tions and only a software agent isneeded for monitoring the RSS values.

Airplace is based on the RSS finger-printing approach to address the chal-lenging indoor signal propagation con-ditions. A number of RSS fingerprints(ie, vectors of RSS measurementsrecorded from APs in the vicinity of theuser) are collected a priori to build the“radiomap”. Location is then estimatedby finding the best match between theobserved fingerprint and fingerprints inthe radiomap using state-of-the-artalgorithms developed in-house [2]. Theprototype system follows a mobile-based network-assisted architecture inorder to reduce the communicationoverhead, while at the same timeaddresses security concerns andrespects user privacy [3]. In a typicalpositioning scenario, when a userenters an indoor environment, covered

by several WLAN APs, the user ’ssmartphone obtains the RSS radiomapfrom the local distribution server in asingle communication round and isthereafter able to self-locate independ-ently using only local knowledge and,more importantly, without revealing itspersonal state.

The Airplace system won the BestDemo award at the 13th IEEEInternational Conference on MobileData Management (MDM'12) and theresults of our research are also beingexploited for the development of anenhanced prototype system via a tech-nology transfer agreement with a largehardware and software company inTaiwan. In particular, we recently builtand demonstrated a hybrid positioningsolution that fuses the WiFi locationwith multi-sensory inertial data that arewidely available on Android smart-phones to deliver fine-grain locationinformation. Releasing a "big-data"Web2.0-oriented geolocation servicethat will be scalable to large indoorenvironments is another direction ofthis research. In this context, some ofour future activities include the investi-gation of algorithms for reliable floordetermination in multi-storey buildings,supporting diverse mobile deviceswithout compromising the positioningaccuracy, developing an efficient map-matching component and building auser-friendly indoor navigation applica-tion using Google Maps.

Links:http://www2.ucy.ac.cy/~laoudias/pages/platform.htmlhttp://www.kios.ucy.ac.cy/http://dmsl.cs.ucy.ac.cy/

References:[1] C. Laoudias et al: “The AirplaceIndoor Positioning Platform forAndroid Smartphones”, in proc. IEEEMDM’12[2] C. Laoudias et al: “Localizationusing radial basis function networksand signal strength fingerprints inWLAN,” in proc. IEEE GLOBECOM,2009[3] G. Chatzimiloudis et al:“Crowdsourcing with Smartphones”,IEEE Internet Computing, Volume 16,2012.

Please contact:Christos G. PanayiotouKIOS Research Center for IntelligentSystems and NetworksUniversity of CyprusTel: +357 22 892298E-mail: [email protected]

Demetrios Zeinalipour-Yazti Data Management Systems Laboratory(DMSL), Department of ComputerScience, University of CyprusTel: +357 22 892755E-mail: [email protected]

Figure 1: Screenshot of the Airplace indoor

positioning interface for Android

Figure 2: Screenshot of the Airplace indoor

mapping interface for Android

http://www2.ucy.ac.cy/~laoudias/pages/platform.html


A key enabler of mobile computing,apart from the networking infrastruc-ture, is the successful integration ofsemantics and advanced reasoning tech-niques into mobile information systems.These features have been recently char-acterized under the Ambient Intelligence(AmI) technologies. The range of appli-cations in ambient information systemsis progressively evolving from large tosmall scale environments. In particular,there is a growing need for applicationsthat assist humans in their navigation-related activities in indoor spaces (eg,airports, museums, office buildings). Asuccessful integration of indoor knowl-edge representation and ambient sys-tems still requires the development ofappropriate spatial data structures anddata management facilities. We believethat this is mandatory for the delivery ofintelligent and context-aware systemsapplied to indoor spaces.

Location-dependent queries and moregenerally context-dependent queriesappear to have considerable impact forthe development of different categories

of such location-based and context-aware services. The context-dependentcharacter of these queries means thatany change in the context (eg, changesin the locations and profiles of theobjects that are involved in the query)may significantly affect the answer. Asthese queries are time-sensitive andlocation-dependent, they may be validonly for a given period of time andwithin a given area. Therefore, thesequeries are expected to be processed ascontinuous queries, meaning that thesystem should continually keep theanswers up-to-date, until the query isexplicitly cancelled by the user. Aunique combination of challenges ariseswith location- and context-aware serv-ices and queries in indoor environ-ments, as researchers must be able torepresent different kinds of location-dependent and user-centred queries(typically path search and rangequeries) in a flexible manner, and totake into account additional contextinformation, time-dependency, and thehierarchical layout of the indoor envi-ronment.

Indoor space modelling is the core ele-ment to be investigated for advancedservices [1]. We postulate that model-ling and designing a hierarchicallyorganized context-dependent indoordata model is the best way to satisfy awide range of location-based services.Such a data model can be viewed as atree structure in which location infor-mation is represented at different levelsof abstraction: (1) a fine-grained graphembedded within an occupancy grid(Figure 1); (2) an exit hierarchy (Figure2); and (3) a location hierarchy (Figure3). Moreover, such a model is able torepresent: (i) static/moving features ofinterest, (ii) their spatial properties, and(iii) the behaviours or actions thatemerge from them. The hierarchicaldesign alleviates performance and scal-ability issues in location dependentquery processing. In addition, time-dependent functions that compute thenetwork distance and the travel time areintroduced. Furthermore, a classifica-tion of user profiles is possible in orderto perform an offline filtering of themultilevel data model, thus reducing the

Knowledge Representation and Management

in Indoor Mobile Environments

by Imad Afyouni, Cyril Ray and Christophe Claramunt

Despite the continuous development and improvements made in mobile computing and the variety of

technologies that can be used to enable ambient indoor environments, there are still many research

challenges in this area that need to be addressed. Spatial data representation and management as well

as location-dependent query processing are among current issues to be coped with so that an efficient

and sufficiently flexible indoor context-aware navigation system can be designed.

Figure 1: A fine-grained graph of an indoor network with exits



amount of data that need to beprocessed in real-time.

The semantics of a query grammar tiedto the indoor data model have also beendeveloped [2]. Such a grammar isrequired to express location-dependentqueries. It supports navigation queriesand incorporates some other prefer-ences and semantics in the query model.It also supports the hierarchical datamodel by using the concept of locationgranules to represent the different levelsof abstraction. The use of the locationgranules concept allows formulation ofqueries using the location terminologyrequired by the user (eg, vertices at thefine-grained level, rooms, floors, build-ings, etc.).

We designed algorithms and a genericarchitecture usable for the continuousprocessing of location-dependentqueries in indoor environments [3].Navigation-related queries areprocessed in accordance with this archi-tecture, and are executed continuouslywhile the request is not explicitly can-celled by the user. Particularly, algo-rithms for hierarchical path searchesand range queries applied to both staticand moving objects have been devel-oped [3]. These algorithms take advan-tage of the hierarchical data model ofthe indoor environment, and employ anincremental approach in order to effi-

ciently execute continuous location-dependent queries, thus avoidingsolving each search problem independ-ently from scratch. Our design andimplementation rely on a databaseextension based on the open sourceDBMS PostgreSQL. The main parts ofthe prototype developed are: (i) thehierarchical network-based data modelof the indoor environments; (ii) theoperators and location-dependent con-straints introduced in the querygrammar; and (iii) the algorithms toprocess continuous location-dependentqueries over moving objects.

To conclude, a successful integration ofindoor knowledge representation andmobile information systems requiresthe development of hierarchical spatialdata models. These models should sup-port continuous processing of location-dependent queries applied to movingobjects acting in these environments.This is exactly the scope of theapproach that has been developed byour research. Application perspectivesare very large, from the development ofinteractive systems for built environ-ments, to additional professional- anduser-oriented services. Not only are theapplication perspectives promising, butthere is also great potential for businessas the range of possibilities will surelyopen many opportunities.

Link: Naval Academy Research Institute:http://irenav.fr

References:[1] I. Afyouni et al: “Spatial Modelsfor Indoor Context-Aware NavigationSystems: A Survey”, JOSIS, 4(1), 85-123, 2012[2] I. Afyouni et al: “Context-awaremodelling of continuous location-dependent queries in indoorenvironments”, JAISE, IOS Press,5(1):65-88, 2013[3] I. Afyouni et al: “Algorithms forcontinuous location-dependent andcontext-aware queries in indoorenvironments”, ACM SIGSPATIAL2012, 329-38.

Please contact: Imad Afyouni and Cyril Ray andChristophe Claramunt Naval Academy Research Institute,FranceE-mail: {imad.afyouni, cyril.ray,christophe.claramunt}@ecole-navale.fr

Figure 3: Part of the location hierarchy derived from the fine-grained graph (Figure 1). “HW” means hallway,

“MR” meeting room, “BT” bathroom.

Figure 2: Part of the exit hierarchy derived from the fine-grained graph of the indoor network.


Pervasive mobile smart devices andsensor networks in public transport vehi-cles are enabling a new approach forenhancing the experience of publictransport customers. The experiencemay actually start before a journey takesplace, by planning a trip, and travellerscould still be providing relevant infor-mation after reaching their destinations.Transport services and associated infor-mation services are closely coupled, andthis relationship needs to be betterunderstood. We claim that pervasivemobile services can be used forenhancing user experience in the trans-port services, and also for enhancingoverall public transport services.

Research being conducted in publictransport information services at FEUP,the School of Engineering of theUniversity of Porto for the past tenyears, and at Imperial College Londonfor the past three years, has been aimingat providing users with real time infor-mation, but also finding ways to use theuser feedback for benefiting all othertransport service stakeholders. Mobilecomputing is a key enabler for makingsuch a vision possible.

The MOVE-ME serviceThe MOVE-ME project has developedan infrastructure and a mobile applica-tion enabling users to access publictransport information in real time. From2012, this application enables travellersto plan their journeys based on real timeor planned data from metro, bus, coach,and train schedules. The infrastructurebrings together geo-referenced datafrom different transport companies,Google map data, and also relevant loca-tions from tourist offices enabling multi-modal journey planning. Real time mul-timodal travel planning can be done in a60 minute time horizon. When real timedata is not available or when the timewindow is larger than 60 minutes andshorter than three days, travel planningis done based on available publishedschedules over a three day horizon. Withsuch time horizons the available infra-

structure has been able to guaranteeacceptable response time for a largenumber of simultaneous users.

The most advanced MOVE-ME servicewas launched in May 2012 in theMetropolitan Area of Porto, and in otherregions of Portugal. Over 25 distinctmetro, bus, coach, and train companiesshare information on their service, andover 20,000 Android and iPhone usersare now benefiting from this service.

The MOVE-ME mobile service andinfrastructure was developed by OPT(www.opt.pt) with research supportfrom FEUP, in the context of the CIV-ITAS Elan European project. MOVE-ME won the 2012 CIVITAS EuropeanTechnical Innovation Award.

Research on mobile advancedtraveller information servicesMore advanced mobile computingresearch with the objective of

enhancing public transport user experi-ence has been under way at FEUP, incollaboration with Imperial College,London, for the past three years.

Current research is also addressingways of extending MOVE-ME toenable travellers to pay for their jour-neys using their mobile devices. TheMetropolitan Area of Porto has an opentransport network with no barriers. Inthe future it is possible to envision anentirely pay-as-you-go transport servicebased on users’ mobile devices. As Webenabled smartphones become perva-sive, ticket selling machines and vali-dating machines could become redun-dant and unnecessary.

Mobile devices are also being used tomeasure users’ affective state in realpublic transport contexts, and how toshare and disseminate real time infor-mation through social networks for thebenefit of other travellers and also forthe benefit of the control rooms of trans-port companies [1]. Preliminary experi-ments have been conducted in Porto andin London transport services.

A prototype cloud-based mobile serviceto assess the relationship between affec-tive state and travelling context hasbeen developed and then tested withcommuters of the Porto transport net-work [2]. User’s affective state was cap-tured using a simple emotional model oftravelling mood, with cognitivepleasure and physical arousal dimen-sions based on Russell’s circumplex ofemotion [3]. Travelling context includednoise, saturation, smoothness, ambi-ence, speed and reliability. The findingsshow a strong correlation betweenmood and context, dependant on theuser.

A prototype crowdsourcing applicationto share information has been devel-oped and tested with commuters of theLondon transport network. The modelborrows key principles from Internetbased services. It strives to intensify

Mobile Real Time Applications for Enhancing Public

Transport User Experience - The MOVE-ME Project

by João Falcão e Cunha, Teresa Galvão and Jeremy Pitt

Research at the School of Engineering of the University of Porto and Imperial College London has been

investigating how smartphones will enable you to enjoy the experience of travelling in a bus.

Figure 1: The map shows a multimodal public

transport route from Vila do Conde to Porto

(in blue) and its stops/stations (red and white

poles). Users can follow a trip in real time.



win-win relationships between publictransport passengers and operators. Thestructured exchange of information issustained by a validation mechanism fordata reliability, and an incentive mecha-nism to encourage passenger participa-tion. Passengers benefit from rich real-time data to ease their journeys andimprove travel experience, in exchangefor their own participation providingand validating information. Operatorsgain access to rich customer generateddata, which in an aggregated formatmay provide a real-time assessment ofcustomer experience and of local per-formance across the entire networkoperation.

It is expected that ubiquitous avail-ability of high quality Web enabledmobile devices and services willimprove public transport user experi-ence, both functionally and emotionally.Collaboration with users may also ben-efit public transport operators, withuser ’s feedback enabling them toenhance service levels.

Better experience leads to increasedusage of shared mobility modes, andtherefore to more sustainable cities inthe future.

AcknowledgmentsThe authors would like to acknowledgethe work of colleagues, PhD students,and collaborators at OPT. Funding fromthe following projects is also acknowl-edged: OneStopTransport QREN TICEno. 13843, MOBIPAG QREN TICE13847, EU CIVITAS Elan, IBM CASPortugal, INEGI, IDMEC Polo FEUP,UGEI - INESC TEC, PhD grants fromFCT, Portugal.

Link: http://www.move-me.mobi

References:[1] A. A. Nunes, T. Galvão Dias, J.Falcão e Cunha, J. V. Pitt: “Using socialnetworks for exchanging valuable realtime public transport informationamong travellers”, in proc. of the IEEEConference on Commerce andEnterprise Computing, 365-370, 2011

[2] P. M. Costa, J. V. Pitt, J. Falcão eCunha, T. Galvão Dias:“Cloud2Bubble: Enhancing quality ofexperience in mobile cloud computingsettings - A framework for systemdesign and development in smartenvironments”, in proc. of the 3rdACM Workshop on Mobile CloudComputing and Services (MCS’12),45-52, 2012

[3] J. A Russell: “Circumplex Model ofAffect”, journal of personality and socialpsychology 9, 6 (1980), 1161–1178.

Please contact: by João Falcão e Cunha and TeresaGalvãoUniversidade do Porto, PortugalE-mail: [email protected],[email protected]

Jeremy PittImperial College London, UKE-mail: [email protected]

U-AirPoll: Mobile distributed and Collaborative

Air Pollution Measurement

by Marino Linaje and Luis Miguel Dominguez-Peinado

Current technology, including smartphones, sensors and the Cloud can revolutionize traditional air

pollution measurement. This article summarizes a novel approach focused on cost optimization,

collaborative data capture, use of standards and public data distribution.

Generally, pollution measurement proj-ects [1] are location static, and data maybe controlled by entities subjugated toeconomic or politic factors.Consequently, environmental data maybe subjective and valid only for a spe-cific location (ie, the area covered by theenvironmental monitoring station).

U-AirPoll is a small, fully open (hard-ware and software), web standard-basedand autonomous system of air pollutionmeasurement. It can be carried bypeople or installed in vehicles such asbicycles, motorcycles or buses. To mini-mize cost, the project relies on the factthat increasing numbers of people arecarrying smartphones. Thus, smart-phone native sensorial capacities areextended with an external hardwarecapable to measure air pollution. Thishardware base is cheap (around 70€)

when compared to the environmentalstation’s price. Data are captured andsent from the air pollution sensors to thesmartphone and uploaded from thesmartphone to the Cloud. Thus, data canbe collaboratively collected by manynodes (users or vehicles) from thepublic or private sectors. Data are openpublished and shared in the cloud to beeasily consumed in XML. Thisapproach thus avoids the gaps of tradi-tional pollution measurement projects.

U-AirPoll communications architectureis composed of three parts: the drone,the droid and the Cloud services (Figure1). In contrast to other solutions, theapproach fully exploits the Web ofThings paradigm.

The drone is a device that extends thephone sensorial capacities. Two ver-

sions are available: Firstly, the proto-type version (Figure 2) including acustom CO2 concentration sensor andan already existing general air qualitysensor. Secondly, the ready-to-use ver-sion (Figure 2, top-left) that is within anenclosure and avoids unnecessary hard-ware (eg, grove nest). The platform isextensible with up to 22 environmentalsensors. It is built using Flyport byopenPicus. Flyport was selectedbecause it is a low cost open hardwarewith IEEE 802.11b interface and amicro-webserver embedded supportingREST (Representational StateTransfer).

Drone hardware could be replaced byFlyport minified and/or washable ver-sions to be embedded into clothes. Oneaspect of U-AirPoll’s versatility is itsmobility: a drone can be placed wher-


ever the user wants within the range ofthe user’s Smartphone Wi-Fi connec-tion (eg, the roof of a bus).

From the droid perspective, the drone isjust a service that is used to acquiredata. The droid is a mobile applicationthat uses the smartphone geolocation tocomplete air pollution data from thedrone. It is implemented using ApacheCordova multi-device developmentframework that can run in up to ninedifferent mobile platforms such asAndroid, iOS or Windows Phone.Apache Cordova was also selectedbecause its applications are coded usingstandard web development technologies(ie, HTML, CSS and Javascript).

When not connected, the droid tries tolocate the drone and checks server con-nectivity. When the drone is detected,data capturing starts and the droid savesall measurements into a local databaseon the smartphone, which is cheaperthan storing them in the drone (ie, addi-tional storage hardware is not required).In order to upload data, login isrequired. Once logged in, the droidbegins to send data (ie, the pollutiondata from the Drone augmented withthe smartphone geolocation). In par-

allel, the droid requests data from aserver in a range of coordinates to showa map in the smartphone. The trend inthe Web of Things field is to use RESTfor the Web services provided.However, to prove that our approachdoes not fix the Web service technologyused, REST and SOAP (Simple ObjectAccess Protocol) are mixed in the finalsolution.

All the cloud services have been speci-fied using a model driven developmentenvironment called WebRatio. All theJava code generated by this tool is opensource. The cloud services provideanonymous data pollution concentra-tions and provide an open access todata. To avoid problems with multipleconnections (also known as the c10kproblem) a JSP and c10k capable server,such as JBoss, is required.

The main idea behind U-AirPoll is to setthe foundations for an open, distributedand mobile air pollution measurementsystem which allows data to be easilyconsumed by custom end-user applica-tions and services (eg, a user or organi-zation that wants to create green trafficroutes to run or to cycle). Since U-AirPoll is an open hardware project,

other people can add more pollutionsensors, design their own enclosures ormodify any piece of the project to bettersuit their needs.

Currently, U-AirPoll supports only 1:1droid-drone connections. Our researchis currently investigating N:N supports(multiple drones and droids sharedamong users).

This work has been funded by theSpanish Ministry of Science andInnovation (TIN2011-27340) and theEuropean Regional Development Fund(ERDF).

Links:http://sites.google.com/site/U-AirPoll/http://www.openpicus.comhttp://cordova.apache.orghttp://www.webratio.com

Reference:[1] D. Vallero: “Fundamentals of airpollution”, Academic press, 2007.

Please contact:Marino Linaje TriguerosUniversidad de Extremadura, Spain Tel: +34 927257258 E-mail: [email protected]

Figure 1: U-AirPoll communications architecture Figure 2: Drone prototype

ARGO Sentinel: The Mobile App for Reporting

Oil Spillages at Sea

by Massimo Martinelli, Davide Moroni and Ovidio Salvetti

We believe that the contribution of volunteers could play a fundamental role in monitoring and

protecting the environment. People at sea sighting pollution caused by oil or hydrocarbon spillages

can now immediately report this using a freely downloadable mobile application.

At the Signals and Images Laboratory(SI-LAB) of the Institute of InformationScience and Technology of the National

Research Council of Pisa (ISTI-CNR),we have developed a MarineInformation System (MIS) for moni-

toring vessel traffic and oil spills withinthe Mediterranean basin. The MIS col-lects and integrates geotagged data



related to safety and health issues of thesea from various sources (satellites,optical sensors, electronic noses,autonomous underwater vehicle sys-tems) and provides predictive models toassist the authorities in the managementof emergencies at sea.

An integrant part of the MIS is repre-sented by the data that can be collectedand shared by volunteers who want tocollaborate in monitoring the status ofthe sea.

For this purpose, we have developedARGO Sentinel, a free application forsmartphones with a GeolocationSystem (GPS)[1]. Two versions of theapp have been implemented: anHTML5 version, running as an Operawidget distributed only to our partnersof the ArgoMarine project, and a nativeAndroid (v2.2 or higher) publicly dis-tributed because of its level of stability,and downloadable by anyone fromGoogle Play (see Link below).

The app was conceived on the intuitionthat the contribution of volunteers couldplay a fundamental role in monitoringand protecting the environment. Usingthe app, whenever someone at sea sightssigns of oil or hydrocarbon pollutionthey can immediately report this to theSI-LAB in Pisa. The information isrecorded in the MIS and complementsthe data obtained from more traditionalsources (eg satellites), improving thequality and coverage of marine moni-toring, especially in protected areas. Inthis way we can build up a detailed mapof the status of our seas.

The application – distributed in Italian,English and Greek –sends reports ofsuspected spills to our Lab, providing adescription of the spill and specifyingthe precise point and severity.

The main screen of ARGO Sentinelshows: a "Message" area where adescription of the sighting is entered; a"Red Alert" button to signal a major oil-spill with a diameter greater than or equalto 20 meters approx.; a "Yellow Alert"button: to report a mild/moderate amountof oil spill with a diameter less than 20meters approx.; an "i" button: for infor-mation on the use of the app; an "exit"button to close the app (See Figure 1).

By clicking on the red or yellow alarmbuttons, an SMS is sent to the CNR

headquarters, which will process thereceived data.

The app has been installed by about 700individuals all over the world. By inte-grating the alerts sent by the volunteerswith all the other information sourcescollected in the MIS, a semi-automaticanalysis eliminated the alerts identifiedas false positive. Even though we haveno formal obligations as the project isfor research purposes only, we forward

all the significant information to theGeneral Command of the Italian CoastGuards in Rome, Italy.

Our field tests demonstrate that the useof this new technology could be reallyimportant in combating pollution. Ourexperience suggests that this kind oftechnology can be applied to manyother fields where environmental moni-toring and safety is crucial.

An important result is that, during theperiod of the project’s activity, we haveacquired a more detailed and immediateknowledge of the conditions of the sea.The use of this application represents astep forward in marine environmentalmonitoring, because, in addition to theother technologies that are used by theArgomarine project, it also adds thecontribution of volunteers who caneasily communicate the sighting of a

spill. Knowing that deliberate spills canbe detected in a timely fashion is initself a deterrent to malicious actions.Possible developments may allow amore effective intervention by theauthorities.

A new version of the ARGO Sentinelapp is now under development and willalso be released for iOS user.

ARGO Sentinel is has been developedin the framework of the Europeanresearch project ArgoMarine,("Automatic Oil spill Recognition andGeopositioning integrated in a MarineMonitoring Network") which aims attraffic and marine pollution monitoring.Coordinated by the Tuscan ArchipelagoNational Park, the partners ofArgoMarine include ISTI-CNR, theNational Technical University ofAthens, the Nansen Environmental andRemote Sensing Center, the Centro deInvestigação Marinha and Ambiental,Universidade do Algarve, the NationalMaritime Park of Zakinthos, the JointResearch Center and the NATOUndersea Research Center.

Links: ARGO Sentinel:http://tinyurl.com/argosentinelArgomarine Project:http://www.argomarine.eu

Reference:[1] “ARGO Sentinel: the applicationfor reporting oil spillages at sea”,http://www.cnr.it/istituti/FocusByN_eng.html?cds=074&nfocus=22, CNRFocus, National Research Council,2012

Please contact:Massimo MartinelliISTI-CNR, ItalyTel: +39 050 621 28 03E-mail: [email protected]

Figure1: the main screen of the "ARGO

Sentinel" mobile app


The exploitation of medicaldata through distributed com-puting infrastructures (DCIs)has led to new research linesfocused on the extraction ofknowledge and the organizationof the information to assist radi-ologists and researchers toretrieve pertinent data fromexisting medical image reposi-tories. This development notonly offers great benefits totraining and research, but canalso help to improve the clinicalmanagement of patients. Onesignificant case is breast cancerscreening. Specifically, effec-tive content-based retrieval isbased on: (i) the structured repre-sentation of the knowledge relative tothe images, and (ii) the integration of ameans to create and exploit a knowledgedatabase with structured data. However,content-based retrieval is stronglyaffected by the quality of input data,especially in the medical area. Theacquisition of accurate and completedata is a key challenge in clinical health-care [1].

On the other side, the inherent com-plexity in the use of DCIs prevents theirusage within the wider community.Scientific gateways are being widelyused to facilitate the access of scientiststo these tools, simplifying the organiza-tion of data repositories and the execu-tion of experiments. In the particularcase of biomedicine, usability evalua-tion is widely recognized as critical tothe success of any software system withwhich end users interact [2].

The introduction of mobile computingconstitutes a great opportunity forincreasing both the quality of introduceddata and the content-based retrieval ofexisting data. Radiology reports areintroduced in plain text using voicerecognition systems. Coding is limitedto conclusions in the best case. Plain

text-based retrieval is inefficient as doc-umented in the literature due to theambiguities of terms and the difficultiesin dealing with negative statements.Structured reports are playing a key rolein the definition of unequivocal piecesof information for comparisons and datamining.

For this purpose, TRENCADIS [3](Towards a Grid Environment toProcess and Share DICOM Objects) is atechnology, based on Grid and Cloudcomputing, created for securelysharing, organizing and searching med-ical images. Moreover, it provides atoolkit to support the development ofapplications that are capable of usingheterogeneous, distributed computingand storage resources, made availablethrough TRENCADIS services. TREN-CADIS has incorporated a mobile-com-patible interface to ease the introductionof structured reports. A prototype hasbeen developed in cooperation with theHospital Universitario Dr. Peset fromValencia for the diagnosis, follow-upand response analysis of breast cancer.

The complete process of breast cancermanagement has been coded into ninetemplates that allow the radiologists to

fill in the report faster thanusing keyboard and mousewithout compromising theaccuracy and guidance of struc-tured terms. A mobile interfacealso enables writing the reportat the image acquisition point.These components are beingintegrated in an existing proto-type being developed for diag-nosis support and research inbreast cancer. A usability studywill follow the validation of theprototype.

Currently, TRENCADIS soft-ware and support is provided tothe collaborating centres within

research projects and contracts.An open-source distribution of thetoolkit is planned for the first term of2014.

References:[1] B. Zheng et al: “Interactivecomputer-aided diagnosis of breastmasses: computerized selection ofvisually similar image sets from areference library”, AcademicRadiology 14(8), pp. 917–927, 2007[2] C. Maestre et al: “Assessing theUsability of a Science Gateway forMedical Knowledge Bases withTRENCADIS”, J Grid Computing, 10,pp. 665–688, 2012[3] I. Blanquer, V. Hernández, J.E.Meseguer: “Content-BasedOrganisation of Virtual Repositories ofDICOM Objects”, Future GenerComput System, 25(6), pp.627–637,2009.

Please contact: Ignacio BlanquerUniversitat Politècnica de València,SpainE-mail: [email protected]

Mobile devices to Improve Breast Cancer

Information Management

by Damià Segrelles, Maite Giménez and Ignacio Blanquer

Intuitive interfaces of mobile devices facilitate the introduction of structured data in breast cancer

management, leading to an increase of completeness and accuracy of diagnosis and follow-up

evaluations, as well as opening the door for more effective content-based retrieval techniques for

clinical decision support.

Figure 1: Filling in structured reports for a mammography from a

tablet

Research and Innovation


Secure and Privacy-Aware

Mobile Identity

Management

by Fabio Martinelli

As evidenced in our special theme, mobile devices play

an increasingly important role in our everyday lives, not

only by enabling us to communicate but also by

providing access to a large variety of pervasive services.

The global mobile security market [1] is steadily

increasing in value (about $1.6bn in 2012), as

companies and organisations seek to secure their

smart devices against the dangers of mobile malware.

Our software products aim at targeting both this market

and that interested in the integration of mobile devices

into legacy applications.

We use mobile devices in many roles, eg as employee in rela-tion to our employers, as consumers in relation to commer-cial service providers, and as citizens in relation to our gov-ernments. There are now more mobile devices than desktopcomputers accessing the Internet. Extrapolating this trend,the number of vulnerabilities affecting these mobiles devicesand their typologies is increasing with the growing perva-siveness of the services. These devices thus represent anattractive target for attackers.

Mobile devices are not only taking over typical PC activities(social networking, browsing, e-mailing, online shopping)but also offer more sensitive applications in areas such asmobile payment transactions, access to health services, etc.The frequent press coverage of hidden tracking, open pro-files, and fraud is one of the factors that has convinced us thatsecurity and privacy threats are realistic. The ever-growingsophistication in data mining can easily be misused for com-mercial or criminal exploitation. A particular concern is thatmobile devices are available to a wide range of users (fromteenagers to seniors), who are not necessarily experts or edu-cated to these risks. This makes successful attacks on thesedevices even more dangerous, especially when the devicesare then used to access services.

In the framework of the EIT ICT Labs (an initiative of theEuropean Union), several research organizations (CNR,Novay , TU/e, TU Berlin,) and leading European industries(Engineering and SAP) are working together to address thechallenge of ensuring the security of mobile devices guaran-teeing service access in a privacy preserving way.

The partners in this activity have a significant track recordfor research and innovation in many mobile security relatedareas and bring together a significant set of experienceranging from European projects to industry based pilots andexperimentations . The field is far from mature, neither interms of providing solutions nor in terms of fully under-standing the threats and the potential on new protectionmethods. Nonetheless many specific activities have beenalready performed and are amenable to exploitation. Thepotential adoption of these technologies –if deployed prop-

erly - is extremely high, while on the other hand the failure toproperly secure our mobile devices can be harmful to societyat large.

We are thus currently developing a software suite of thesemobile devices to secure access to data-sensitive platformssuch as financial and e-government services or pervasiveones such as e-health for personal monitoring. The data andprotocols used to store and (sometimes even autonomously)use personal identity information must be secure. This entailsthe protection of the devices themselves, in order to avoidvarious types of misuse, ranging from the security of com-munications, to privacy considerations on data disclosed toget access in these communications.

One of our main challenges is the use of security solutionswithin mobile applications. To address this problem, SAPaims at facilitating the development of secure mobile appli-cations with its SAP Mobile Platform (SMP). SMP offers aunique opportunity for developers to quickly develop busi-ness applications interacting with critical systems – typicallythe company ERP. While SMP is already providing state ofthe art security, SAP intends to extend the security function-ality spectrum and to provide new sensitive services devel-oped within the project.

Besides protecting the private data on the device, we alsoconsider the protection of data used by the services thedevice connects to. In authenticating to the services, wedeploy privacy-friendly biometrics with biometrics tem-plates which are protected, also from `insider’ threats.Whenusing the services, a lot of personal information is shared. Weprovide tools to analyze the amount of personal data revealedand to validate the privacy-protecting properties of commu-nication protocols used.

Our applications will be tested using the Experience andLiving Labs facility offered by EIT ICT. This will allow totest the usability and the social acceptance of the solutionsdeveloped. Several business and transfer technology actionsare planned in order to make our envisaged mobile centricsecure identity managing available on the market place. Theactivity described above is in collaboration with manygroups. As main partners we mention Silvia Boi, LaurentGomez, Niklas Kirschnick , Jerry Den Hartog, MartijnOostdijk, Jean-Christophe Pazzaglia and Daniele Sgandurra.

Link:SAP Mobile Platform:www.sap.com/solutions/tech/mobile.html

Reference:[1] The Mobile Security (mSecurity) Market 2012-2017March 2012, https://www.asdreports.com/

Please contact:Fabio Martinelli, IIT-CNR, ItalyE-mail: [email protected]


Social Electricity: When

Awareness About

Electricity Becomes Social

by Andreas Kamilaris, George Taliadoros and AndreasPitsillides

Although domestic smart metering has been introduced

into our lives, it is still not easy for consumers to know

how much electricity they are using. Consumers are

unable to assess whether their consumption is low,

average or high. “Social Electricity” is a Facebook

application that aims to make individuals aware of their

energy consumption by means of comparisons with the

corresponding electrical consumption of their friends,

as well as with the total consumption in the area where

they live. Effective and realistic social comparisons

could raise consumers’ awareness of their consumption

behaviour, enabling them to take steps to reduce their

electricity use and carbon footprint.

Energy conservation is a global issue with huge environ-mental, social and political implications. Smart metering hasbeen introduced into homes to provide timely energy con-sumption feedback, helping consumers to more effectivelymanage their power use. Nevertheless, it is still difficult forpeople to quantitatively assimilate how much energy theyconsume. Consumers lack the required metrics to determinewhether their total consumption is low, average or high. Apromising way to understand the "semantics" of consumedenergy is to compare it with the amount consumed by rela-tives, friends and neighbours. The increasing popularity ofonline social networking sites (eg Facebook, Twitter), allowsthe proliferation of social ICT applications targeting energyawareness.

Social Norms for Energy ConservationPeople are willing and able to adapt their behaviour toenergy-saving lifestyles if given the necessary feedback,support, and incentives [1]. While detailed feedback can

Figure 1: A snapshot of Social Electricity application showing a

map of Cyprus with a user’s friends appearing in green if their

consumption is low and red if their consumption is relatively high.

raise people’s awareness of their consumption, communityinfluence, which is recognized as an important factor inenergy-saving initiatives, has the potential to drive residentstowards a more persistent behavioural change [2]. . Socialnorms can motivate people to question their behaviour ifthey discover it is not "normal" [3].

Social Electricity: Energy Conservation through SocialComparison “Social Electricity” is a Facebook application that aims to

make consumers aware of their electrical consumption, bymeans of comparisons with the corresponding consumptionof their friends, as well as with the total consumption in theirstreet/neighbourhood/village/city/country. By effective andrealistic comparisons, consumers may perceive their energybehaviour and take steps to reduce their electricity footprint.

The project has been developed by the Networks ResearchLaboratory, part of the Computer Science Department,University of Cyprus. It is supported by the ElectricityAuthority of Cyprus (EAC), which provides data on thedomestic consumption in Cyprus every two months.

These data are anonymous to protect the privacy of Cypriots,and are aggregated at street level, to further ensureanonymity. Through Facebook, people can compare theirconsumption with their online friends and view statisticsrelating to the areas with the lowest/highest energy consump-tion. Users are also encouraged to add their own consump-tion, and compare it with that consumed on their street,neighbourhood, village or the whole of Cyprus. Moreover,users may observe the electrical consumption of their street(or their friends’ streets) over previous months (or the samemonth in previous years) and compare it with their presentstreet consumption.

Social Electricity aims to encourage healthy competition, inwhich users compete with each other to reduce energy con-sumption. People from the same area are encouraged tocooperate in order to reduce the total aggregated consump-tion and improve the overall "green ranking" of the area.

This social application is an innovative initiative worldwideand has recently been awarded first prize in the 2nd GreenICT Application Challenge, organized by the InternationalTelecommunication Union (ITU).



The project only started two years ago, but was officiallylaunched recently and has already amassed over 1,000 users.This is significant, considering that Cyprus has a populationof less than a million.

EvaluationThe project was recently evaluated with the use of onlinequestionnaires completed by 190 users of the application.Eighty percent of respondents reported the application to bevery useful. Half said they were positively influenced tobecome more energy-aware. Around 16% realized theirenergy consumption was high, while 78% said that the appli-cation helped them to better perceive and understand theirenergy behaviour and consumption. Sixty-two percentclaimed their consumption was reduced relative to the sameperiod last year. Our measurements in the last 6 months rein-force this claim, since in all these months, the streets of ourusers consume less energy in comparison to the streets oftheir postal code (4.3-6.5%) or to the whole of the country(2.5-4.8%).

The primary incentives for using the application are financialand environmental, and secondary incentives are curiosityand social responsibility. The most popular incentive forenergy reduction would be a discount on the electricity billsof energy-aware citizens.

Thirty-nine percent of users believe the application willbecome more useful in coming years, in combination withthe massive deployment of domestic smart meters.

Impressively, 64% believe that Social Electricity will helpthem to reduce their electrical consumption by more than10% in the long-term.

ConclusionSocial norms may improve people’s environmental aware-ness, while comparison between friends may help them toreduce their electricity footprint. Social Electricity focuseson investigating the effectiveness of these factors on energyawareness and conservation of electricity.

Future work involves enhancing the features of the applica-tion, including developing more effective comparisons forelectrical consumption between households that share sim-ilar physical characteristics (eg size, number of occupants,type of heating), as well as online, real-time competitions forenergy savings between friends and neighbours.Accompanying the smart grid is the massive deployment ofsmart meters for real-time acquisition of household elec-tricity consumption. The concept of Social Electricity iscompatible with the future smart grid, since smart meterscould enable real-time energy comparisons and promotecompetition between friends and neighbours, related toenergy savings. Evidence indicates that the future Web willbe social and pervasive, connected to the physical world. Weaim to exploit this trend to promote energy awareness andconservation.

Links:Social Electricity Application:https://apps.facebook.com/socialelectricity/Social Electricity Facebook Page:https://www.facebook.com/SocialElectricity2nd Green ICT Application Challenge:http://www.itu.int/ITU-T/climatechange/greenict/201206/index.htmlNetworks Research Laboratory:http://www.netrl.cs.ucy.ac.cy/

References:[1] A. Kamilaris et al: “Energy Conservation throughSocial Competitions in Blocks of Flats”, in proc. ofSMARTGREENS 2012[2] R Cialdini: “Influence: science and practice”, Allyn andBacon, 2001[3] H. Allcott: “Social norms and energy conservation”,Journal of Public Economics, 95(9-10):1082–1095, 2011.

Please contact:Andreas Kamilaris, University of CyprusTel: +357 99 551016E-mail: [email protected]

A snapshot of Social Electricity

application showing a menu in which the

user is encouraged to add his own,

personal consumption and compare it

with his local neighbourhood,

town/village or the whole of Cyprus.


New Interaction

Paradigms in Energy

Management

by Paulo Carreira and Alfredo Ferreira

Information technology tools for energy management

have, to date, failed to take advantage of the recent

advances in human-computer interaction. In fact,

considerable expertise is still required to interpret and

manage energy data. Other domains, in contrast,

including architecture, engineering and construction,

have embraced new interaction paradigms aiming at

more natural ways to interact with and analyse data.

We foresee that integrating natural user interfaces with

virtual and augmented reality environments in an

energy management system will provide a powerful

tool, enabling unskilled users to perform complex

energy management activities.

Extracting meaningful information from energy datarequires integrating attributes and arranging multiple ele-ments, not only regarding consumption but also buildingenvelope, space function, occupancy history, environmentalconditions and equipment status. Traditionally, distinct toolshave been used to manage these data, such as CAFM andCAMM tools, as well as energy management (EM) andbuilding control (BCS) systems tools. Since there is no stan-dard integration methodology and framework for energydata, as there is for business data, practitioners must interactwith multiple tools to obtain relevant information [1].Consequently, integration is improvised and largely per-formed mentally. In a professional setting, the amount ofdata is overwhelming, hampering meaningful analysis.

In addition, energy data is a type of space related informationthat is difficult to explore and visualize with traditional tools.Activities such as inspecting the consumption profile of spe-cific types of space, groups of occupants, or specific equip-ment all have an underlying spatial dimension. Integratedrendering of spatial information is important for visualizingcomplex phenomena that arise when combining data frommultiple sources in order to achieve new insights. Therefore,

most EM tools have to manage spatial information, whichcan be visualized more effectively when rendered in a graph-ical representation.

Graphical rendering accomplishes instantaneous identifica-tion of the space reality along with the relationships of theelements therein to encourage a fast response. Historically,planimetric CAD drawings and Geographical InformationSystems have been used as an effective way to display andmanage spatial information related to facilities.

On closely related domains, virtual and augmented realityenvironments are emerging as a sophisticated and effectiveway of rendering spatial information [2]. The domains ofarchitecture, engineering and construction (AEC), alreadyhave tools that combine these environments with new inter-action paradigms and natural user interfaces (NUIs) toimprove productivity, such as the tool for HVAC mainte-nance presented in [3]. However, user interfaces for EMtools are not yet taking advantage of the recent advances inhuman-computer interaction. In fact, we believe that they arenot even close to approaching the spatial dynamism and easeof use offered by virtual and augmented reality environmentswith NUIs.

Devising a solution of this type presents a twofold researchchallenge. On the one hand, integration of energy datarequires the collection and analysis of large amounts ofsensor and meter data. Furthermore, the database workloadinvolves a large number of hybrid queries combining largepersisted data with live data posed against an integratedschema - a problem that is not well handled by current data-base systems. On the other hand, we need to design adequatevirtual and augmented reality environments to analysehighly-complex, multi-dimensional and space-related energydata. These challenges are highly interdependent. The moredata can be integrated, the richer the interactions can be.Richer interactions, however, require answering even moredata integration queries that combine persisted and live data.

Tackling these research challenges calls for a multi-discipli-nary approach. At INESC-ID Lisboa, a group of specialistsin data management, building automation and energy man-agement have been collaborating with a team of human-com-puter interaction researchers on a novel approach to EM. Wehave been designing and experimenting with tools and proto-

Figure 1: Screenshot of the VRE

prototype application displaying

energy data and rendering status

of devices at INESC-ID Taguspark

building



types that augment a virtual facility with energy consump-tion information along with spacial characteristics, locationand status of equipment, while providing simple ways tocontrol them. Unlike previous work that has largely relied onCAD or VRML for scene generation, this work takes advan-tage of recent game engine technologies for fast and real-time rendering of feature-rich representations of the facility,along with spacial information, equipment conditions anddevice status.

One of the above-mentioned tools has been rolled out com-mercially. Our vision is to create a platform for data integra-tion and visualization that enables easy and efficient ways forunskilled users to explore energy data. We believe that abetter, integrated ICT solution for handling energy data maytranslate to faster and more effective energy managementthrough more accurate energy usage diagnostics and just-in-time corrective action that will eventually translate to a morerational usage of energy.

References:[1] J. Granderson, M. A. Piette, G. Ghatikar, P. Price:“Building Energy Information Systems: State of the Tech-nology and User Case Studies”, technical report, LBNL-2899E, Environmental Energy Technologies Division(2009)[2] P. Fite-Georgel: “Is there a reality in Industrial Aug-mented Reality?”, in proc. of ISMAR’11, IEEE ComputerSociety, Washington, DC, USA, 201-210 (2011)[3] Viet Toan Phan and Seung Yeon Choo: “Using AR forreal-time crosscheck of ventilator ducts at worksite”, inproc. of VRCAI’10, ACM, New York, 293-298, 2010.

Link:http://www.inesc-id.pt

Please contact:Paulo Carreira, Alfredo FerreiraINESC-ID Lisboa and IST/Technical University of Lisbon,PortugalTel: +351 214 233 287, +351 214 233 512 E-mail: [email protected], [email protected]

VMC: A Tool for the

Analysis of Variability in

Software Product Lines

by Maurice ter Beek, Stefania Gnesi and FrancoMazzanti

Researchers from the Formal Methods and Tools group

of ISTI-CNR have developed a tool for the computer-

aided verification of behavioural variability in product

families.

During the last decades, we have witnessed a paradigm shiftfrom mass production to mass customisation in an attempt toserve as many individual customer's needs as possible. A typ-ical example is the production of mobile phones. SoftwareProduct Line Engineering (SPLE) translates this paradigminto a software engineering approach aimed at developing, ina cost effective way, a variety of software-intensive productsthat share an overall reference model, ie that together form aproduct family. Usually, commonality and variability aredefined in terms of features, and managing variability isabout identifying variation points in a common family designand deciding which combinations of features are to be con-sidered valid products.

Feature models are widely used for variability managementin (S)PLE. A feature model provides a compact representa-tion of all the products of a product family in terms of theirfeatures. Figure 1 shows an example feature model inspiredby the mobile phone industry, adapted from [1]. The featuresare represented as nodes in a tree, with different relationshipsbetween them. We see that all mobile phones must run soft-ware to support calls and to display information on either abasic, a colour or a high resolution screen. The software mayinclude support for a GPS and for media devices such as acamera, an MP3 player or both. Finally, mobile phonesincluding software for a camera must also include softwareto support a high resolution screen, whereas software for aGPS cannot run on a basic screen.

There is a large body of literature on the computer-aidedanalysis of feature models to extract valid products and todetect anomalies, ie, undesirable properties such as super-fluous or - worse - contradictory variability information (forinstance so-called false optional and dead features). Theseanalyses however do not take into account any behaviouralvariability as only the presence of the software implementingthe features is considered, not their temporal ordering. Sincesoftware products are often large and complex, and many areused in safety-critical applications in the avionics, railways,or automotive industries, it is unrealistic and practicallyinfeasible to specify - let alone verify - the behaviour of eachproduct individually. Note that our mobile phone examplewould already require behavioural models for 14 largelyidentical products.

For this reason, we recently launched a research effort to1) develop a formal modelling and analysis frameworkcapable of dealing with behavioural variability and 2) pro-


vide tools to support this framework with automated verifi-cation. This has resulted in the use of Modal TransitionSystems (MTSs) as a formal method for describing thecombined operational behaviour of an entire productfamily. An MTS is a Labelled Transition System (LTS) thatdistinguishes between optional (may) and mandatory(must) transitions. Since MTSs cannot model all variability,we enrich them with an additional set of variability con-straints.

We have also developed a tool for the computer-aided verifi-cation of behavioural variability in product families, namelyVMC (Variability Model Checker). VMC takes as input ahigh-level description of an MTS together with a set of tex-tual constraints. In our mobile phone example the latterwould include “GPS EXC Basic” and “Camera REQ Highresolution”.

VMC allows the user to interactively explore an MTS of aproduct family; model check properties (branching-timetemporal logic formulae) over an MTS; visualise the (inter-active) explanations of a verification result; generate one,some, or all of the family's valid products (represented asLTSs); browse and explore these; model check whether ornot products (one, some, or all) satisfy certain properties;and, finally, help the user to understand why a certain validproduct does or does not satisfy specific verified properties

by allowing such a product to be inspected individually.Figure 2 shows VMC's capability to model check a temporallogic formula over all valid products of a family.

The core of VMC consists of a command-line version of themodel checker and of a product generation procedure. Theseprograms are stand-alone executables written in Ada that caneasily be compiled for the Windows, Linux, Solaris, and MacOS X platforms. These core executables are wrapped with aset of CGI scripts handled by a web server, making it easy tobuild an html-oriented GUI as well as to integrate graphdrawing tools. The development of VMC is ongoing, but aprototypical version is publicly usable online (see the linkbelow) while its executables are available upon request.

Currently, VMC is not targeted at very large families. Itsmain limitation, however, is the generation of the modelfrom its input language, while its on-the-fly verificationengine and advanced explanation techniques are those of thehighly optimized family of on-the-fly model checkers devel-oped at ISTI-CNR over the last few decades for verifyingformulae in action and state-based branching-time temporallogics derived from the CTL family of logics. The on-the-flynature of VMC means that in general it is not necessary togenerate and explore the whole state space. This featureimproves its performance and allows it to deal with infinite-state systems.

Link:VMC: http://fmt.isti.cnr.it/vmc/

References:[1] D. Benavides, S. Segura, and A. Ruiz-Cortés: “Automated analysis of featuremodels 20 years later: A literaturereview.” Information Systems 35(6) pp.615-636, 2010.http://dx.doi.org/10.1016/j.is.2010.01.001

Please contact:Maurice ter Beek ISTI-CNR, ItalyE-mail: [email protected]

Figure 1: A feature model of a

mobile phone product family

Figure 2: Verification result of a property for all products of a family of coffee machines



Wood Variety Recognition

on Mobile devices

by Pavel Vácha and Michal Haindl

Each type of wood has its own specific physical,

aesthetic and economic properties; thus correct

identification of wood species is required in numerous

practical applications, from construction industry,

manufacturing, furniture design, and restoration to

pricing evaluation of wooden items. Fast, reliable, and

practical recognition of wood species is therefore

important, having potential impacts in a range of areas,

including: the intended application, construction safety,

and detecting illegal logging of endangered species. A

mobile Android application that can automatically

recognize wood species from a phone photo has been

developed at the Institute of Information Theory and

Automation AS CR in Prague..

The traditional method of identifying wood species involvesmanual browsing through digital wooden veneer cataloguesand making a subjective judgement. This is labour intensive,and concentration problems can lead to errors. Additionally,gradual changes and changing shades due to variable lightconditions are difficult for humans to detect. As an alterna-tive, we aimed to develop an application to identify wood

species using a smartphone camera, which returns theresulting species name and a corresponding high qualitydatabase wood specimen image. This computer-aided woodidentification system retrieves a wood template from a dig-ital wood database, selecting that which most closely resem-bles the query sample.

A wooden surface is captured by a smartphone camera withthe developed Android application, and the image is trans-mitted to the server side which computes the advanced multi-spectral Markovian textural features [1] and finds the mostsimilar wood species from its database. Markovian featuresare not only very efficient, compacting the representation ofvisual properties of wood, but are also invariant to illumina-tion colour and robust to illumination heterogeneity and illu-mination direction, therefore the retrieval result is not influ-enced by illumination properties. The recognized woodspecies together with its high quality database pattern is sentback to the user so he or she can verify the classifier’s result.The challenging part of the method is to compare poorquality smartphone images taken under variable illuminationand resolution conditions with high quality high resolutionmatte wooden textures stored in the wood database.

The performance of our application was verified on thewood database, which contains about sixty differentEuropean as well as exotic wood species, each with twosample images. Our Markovian textural features (2D CAR)[1] compared favourably with alternative Local BinaryPatterns (LBP) [2] and Gabor textural features [3]. The testimages were photographed from approximately the samedistance and with camera oriented along tree rings to reducevariation in scale and rotation. These simple non-restrictiveconditions were enough to raise the performance of the fea-tures above their rotation invariant versions. Our correctrecognition rate was 66 %; LBP features 22 %, while theGabor features completely failed due to non-homogeneoussmartphone flash illumination. The wood recognition ratesignificantly improves (ie, Markovian features 81 %, LBP55 %, Gabor features 48 %) if the input samples are meas-ured with a high-end consumer digital camera. Althoughsmartphone cameras have sufficient resolution (up to 10mega pixels), their poor quality lenses and aggressive JPEGcompression result in overall inferior image quality and thusa more demanding recognition task. Figures 1 and 2 illus-trate the system’s typical performance applied to Europeanand exotic wood samples.

References: [1] P. Vácha, M. Haindl, T. Suk: “Colour and rotationinvariant textural features based on Markov random fields”,Pattern Recognition Letters, vol.32, no.6, pp.771-779, 2011[2] T. Ojala, M. Pietikäinen, T. Mäenpää: “Multiresolutiongray-scale and rotation invariant texture classification withlocal binary patterns”, IEEE Trans. Pattern Anal. Mach.Intell. 24(7), 971–987 (2002)[3] B. S. Manjunath and W. Y. Ma: “Texture Features forBrowsing and Retrieval of Image Data”, IEEE Trans. onPattern Anal. and Mach. Intelligence, 18(8), 1996.

Please contact:Pavel Vácha, CRCIM (UTIA), Czech RepublicTel: +420 2 6605 2365, E-mail: [email protected]

Figure 1: Apple wood sample taken by a smartphone camera and the

three closest query results using Markovian and LBP textural

features

Figure 2: Comparison of results for Palisander wood smartphone

retrieval


The New SHIELd

Architectural framework

by Mariana Esposito, Andrea Fiaschetti, FrancescoFlammini

New SHIELD (nSHIELD) is a European research project

co-funded by the Artemis Joint Undertaking (Sub-

programme SP6) focused on the research of innovative

solutions for security, privacy, dependability (SPD) in the

context of embedded systems (ES), including those

employed in railway security applications.

The nSHIELD project aims at addressing SPD issues as“built in” rather than as “add-on” functionalities, by adoptingan innovative holistic approach. We perceive this strategy asbeing the first step towards SPD certification for future ES.

The leading ideas at the basis ofthis research are: i) to enrich thestate-of-the-art with new SPDsolutions and ii) to enable thecomposability of these (new oralready existing) solutions. Thiswill be achieved in two steps.First, starting from current SPDsolutions, the project will developnew technologies and consolidatethose already explored inpSHIELD (a SHIELD pilotproject) in a solid basement thatwill become the reference mile-stone for a new generation of“SPD-ready” ES. Second, thesetechnologies will be thenenhanced with the “composability” functionality that isbeing studied and formalized.

In a nutshell, composability is the possibility of dynamicallyactivating one or more SPD functionalities in order toachieve a desired SPD level. This is possible with the imple-mentation of the following enabling mechanisms and tech-nologies:• Semantic description of security domain and system com-

ponents, in order to have a machine-understandable lan-guage to drive the automatic composition.

• SPD Metrics, in order to quantify the security needs and theachieved security level over heterogeneous environments

• Security Agent, the engine is in charge of continuouslymonitoring the environment to look for new componentsor new security needs

• Policies and control algorithms to provide a solution for the“composition problem”, ie how to put together the availableSPD technologies in order to achieve the security target.

nSHIELD will approach SPD at 4 different levels: node, net-work, middleware and overlay (see Figure 1). For each level,the state of the art in SPD of individual technologies andsolutions (ranging from hardware and communication tech-nologies to cryptography, middleware, smart SPD applica-tions, etc.) is expected to be significantly improved and inte-

grated into the so-called SHIELD architectural framework,which will represent the breakthrough result of the project.

The nSHIELD consortium comprises 6 manufacturers andsystem integrators, 7 universities, 9 SMEs and 2 IndustrialR&D organizations, mostly members of ARTEMISIA, theEuropean community of experts in the Embedded Securitydomain. The project is led by an industrial partnership (70%of the effort) although the universities and research centresinvolved in the project have an important role (30%) in con-tributing the needed innovation.

The main objective of the project is to conceive and designan innovative, modular, composable, expandable and high-dependable architectural framework. nSHIELD will achievethe desired SPD level in the context of integrated and inter-operating heterogeneous services, applications, systems anddevices, and will develop concrete solutions capable ofachieving this objective in specific application scenarios

with minimum engineering effort. Four scenarios have beencarefully selected in order to cover a wide and significantrange of expected industrial needs.

One of these scenarios addresses dependable surveillancesystems for rail-based transit security, but the aim is toextend applicability also to safety-critical (the so-called“vital”) subsystems in railway signalling, control and super-vision. In these contexts, the composability of the SHIELDarchitectural framework will have great impact on the systemdesign costs and time to market of new products and solu-tions. At the same time, the integrated use of SPD metrics inthe framework will impact on the development cycles ofSPD in ES because the qualification, (re-)certification and(re-)validation process of a SHIELD framework instancewill be faster, easier and widely accepted.

Link: http://www.newshield.eu/

Please contact:Mariana Esposito, University of Naples, ItalyE-mail: [email protected] Fiaschetti, University of Rome, ItalyE-mail: [email protected] Flammini, Ansaldo-STS, ItalyE-mail: [email protected]

Figure 1:The nSHIELD framework



A Biofeedback System for

Self Empowerment and

Improved Quality of Life

by Johanna Mercurio

We have designed “Affective Health”, a biofeedback

system that supports personal empowerment rather

than setting tasks for our bodies. Rather than reducing

our bodies to objects that can be kept in perfect shape,

our approach looks at what it means to be human.

A wave of wellness and health systems focusing on biofeed-back are gaining popularity in the commercial world throughapps that mirror factors such as one’s pulse, sweat levels orenergy consumption. Commercial systems such as the NikeFuel-band, Polar’s pulse meter and Affectiva’s bracelet areallowing users to interact with their arousal reactions or

energy consumption in real-time, as part of their everydaylife. Most existing biofeedback systems aim at relievingpain, creating interesting experiences, or making ushealthier.

Our interest in bio-sensor-based interaction began a decadeago when we were exploring affective interaction design [1].In affective interaction design, meaning, dialogue, and emo-tions are created through the interaction itself – they are notgiven entities to be recognized, identified and treated by thesystem.

Affective Health is a system consisting of a bio-sensor wrist-band, developed by Philips Research, which measures move-ment (tri-axial accelerometer) and arousal level (GalvanicSkin Response). The bio-sensor data is transferred, viaBluetooth, to a mobile phone in real-time and it logs the dataon the phone. A mobile application visualizes the move-

ments’ trough shapes while arousal is fluently represented bya colour scheme. The measurements can indicate one’s cur-rent bodily state relative to the history of previous states. Itcan portray situations that are stressful and engaging as wellas peaceful moments.

The system is designed to encourage the users to understandand interpret the feedback it gives them. The system is builton the notion of interactional empowerment [2] with thedesign allowing users to form their own, personal interpreta-tion with the bio-sensor data and construct the meaning ofthe picture painted by the system over time. In this way thesystem empowers users to identify, reflect on, and find pat-terns in, their behaviour through a value-free design stance.It is through the interaction over time that the system startsmaking sense, mirroring bodily reactions or the users’ activi-ties back to them. An interactional perspective on design willnot aim to detect, and inform the user of, a singular accountof the “right” or “true” interpretation of the user’s behaviour,but rather make experiences available for reflection andinteraction. By carefully crafting the feedback from thesystem, timing its response, and making the design mirrorthe users’ visceral, emotional reactions, the system pulls theusers into what we term an “affective loop experience” [3].In this interaction, users actively take control over their ownsomatic reactions, learn how to master their own bodies andacquire a better understanding of themselves.

Our long-term trials of the Affective Health system show thatpeople need more social solutions. People want to engagesocially, exercise together, and chat. This raises the question:To what context would Affective Health be best suited? Wehave developed initial design concepts to understand whichdata users would like to share, for instance contextual datafrom social platforms or location or calendar information. Itis also important to consider how the data could be stored inthe cloud in a safe and secure way. The next step is to furtherexplore how bio-data could be shared socially in real-time orpast time, remotely or directly.

The Affective Health system is developed by SICS andfunded in collaboration with EIT ICT Labs. The start-upcompany for the system, Mirrormirror, is in the initial stagesof commercialization.

Link: http://www.sics.se/ah

References:[1] P. Fagerberg, A. Ståhl, and K. Höök: “Designing ges-tures for affective input: an analysis of shape, effort andvalence”, in proc. of MUM 2003, Norrköping, Sweden,2003 [2] K. Höök, A. Ståhl, P. Sundström, and J. Laaksolaahti:“Interactional empowerment, in proc. of CHI’08, ACMPress, 647-656, 2008 [3] A. Ståhl, P. Sundström, and K. Höök: “A foundation foremotional expressivity”, in proc. of Designing for UsereXperience 2005, AIGA: American Institute of GraphicArts, 33.

Please contact:Johanna Mercurio, SICS Swedish ICTE-mail: [email protected]

Researcher Johanna Mercurio with the Affective Health system


Software and Hardware-

Intensive Activities

for Supporting Creative

Learning

by Michail Giannakos and Letizia Jaccheri

Our Toys project is investigating how teenage girls

participate in the creative production of digital artifacts.

Furthermore, the project introduces them to digital and

computer literacy. We capture the students’ attention

with character development using physical artifacts, and

use computer programming to bring the character to life.

Digital artifacts that enable people to exchange, create, anddistribute information have, in the past couple of decades,profoundly reshaped the way we work and live. The creativeproduction of digital artifacts in learning activities has beenlinked to teaching new computer literacy skills [1]. Beyonddesktops, ubiquitous technologies not only allow a more

active and physical engagement, but also provide the oppor-tunity for novel and creative interactions. Physical and dig-ital enhancement also provides the facility to convey experi-ences that are not possible in the physical world, forexample, turning our thoughts into reality (ie, through a 3Dprinter). In turn, this physical and digital enhancement canprovide opportunities to encourage or even enhance furtherexploration, discovery, reflection, and collaboration.

In our approach, we designed and developed a workshopprogram called “Our Toys”. Our Toys is based on opensource software and hardware and consists of tutorials onopen source tools, artifact-enriched creative sessions and stu-dents’ demonstrations/presentations.

To date, 66 children have participated in our workshop pro-gram. The workshops took place at creative centres ofUniversity and ReMida centre. ReMida houses a variety ofmaterials for use in creative and educational projects. Thecentre is a cooperation between the municipality, the educa-tion project, Reggio Children, the municipal waste company(recycling) and the local business community. Studentsworked according to Reggio Emilia education principles.The main idea is that the child takes the initiative in creativeactions. ReMida centres are creative places with many

appealing objects where children start to work without beingprompted by adults.

The children attending the workshops were instructed andassisted by the programming artist, the leader of the ReMidacentre, a senior researcher, the project manager and two HCIresearchers. The children completed and published six inter-active works and eleven installations (see example, Figure 1)based on the software/hardware and the recycled materialswithin the centre. At the end of the project, students’ interac-tive projects were presented in an official exhibition in theuniversity (http://www.itovation.org).

Children engaged in programming languages (ie, Scratch)and programmable hardware platforms (ie, Arduino), whichenable them to engage in the world of creativity with digitalenriched artifacts, like robots and interactive installations.

The early results of our initiative [2] showed that introducingyoung people (in particular, girls) to creative software andhardware intensive activities is a good way to attract studentsto the field of Computer Science by increasing their interestand engagement. In particular, we found out that ourapproach [2]: raises awareness of technology, makes the

experience more intense, invites children to explore bound-aries, and increases collaboration and the exchange of viewsand ideas.

Our Toys project is partly supported by the ERCIM “AlainBensoussan” Fellowship Programmem co-funded by the ECMarie-Curie Actions and NTNU, in collaboration withReMida centre and Trondheim Electronic Art Centre.

References:[1] L. Buechley, M. Eisenberg, J. Catchen, A. Crockett:“The LilyPad Arduino: Using Computational Textiles toInvestigate Engagement, Aesthetics, and Diversity in Com-puter Science Education” in proc. of CHI ‘08, ACM Press,423-432

[2] M.N. Giannakos, L. Jaccheri and R. Proto: “TeachingComputer Science to Young Children through Creativity:Benefits and Perspectives”, in proc. of CSERC ‘13, ACMPress.

Please contact: Letizia Jaccheri, Michail Giannakos (ERCIM Fellow)NTNU, NorwayE-mail: [email protected], [email protected]

Figure 1: Example of interactive installations (left); giving life on the installations through programming (right).

PROMISE Winter

School 2013 on

Bridging between

Information Retrieval

and databases

by Nicola Ferro

The main mission of the PROMISE

EU FP7 network of excellence is to

advance the evaluation and

benchmarking of multimedia and

multilingual information access

systems. Together with the ELIAS

research network on information

access system evaluation, funded by

the European Science Foundation,

PROMISE organized a winter school

on “Bridging between Information

Retrieval and Databases” as a week

long event in Bressanone, Italy, from

4-8 February 2013.

The aim of the school was to give par-ticipants a grounding in the core topicsthat constitute the multidisciplinaryarea of information access to unstruc-tured, semi-structured, and structuredinformation. The idea of the schoolstemmed from the observation that,nowadays, databases are increasinglyusing techniques that have traditionallybeen typical of information retrievaland, viceversa, the use of database-ori-ented techniques is becoming morecommon in information retrieval.

17 experts from academia and industrywere invited to speak on a large varietyof topics ranging from introductorytalks on databases, informationretrieval, experimental evaluation, met-rics and statistics to advanced topicssuch as semantic search, database key-word search, semi-structured search,and performance evaluation. Focusedlectures were devoted to bridging gapsbetween information retrieval and data-bases and to the management andsharing of research data via evaluationinfrastructures. Hot topics concernedevaluation with respect to usefulness,crowdsourcing, evaluation on socialmedia, and moving from evaluation toapplications.

52 participants from 16 countriesattended the courses (17% MsC stu-dents, 63% PhD students, 10% post-

W3C at the Mobile

World Congress 2013

by Marie-Claire Forgue

The World Wide Web Consortium

(W3C) was present at the Mobile

World Congress 2013 25-28 February

2013 to demonstrate the impact of

the Open Web Platform and HTML5-

based Web apps, proposed as the

best development solutions

addressing reduction in costs and

complexity. Nowadays, HTML5-based

Web applications are enhanced with

new capabilities such as responsive

design, WebRTC (Web Real-Time

Communication), near field communi-

cation (NFC), Web payments, and full

integration of video.

In 2005, W3C launched its Mobile WebInitiative, and, since then, has beenpresent at almost every Mobile WorldCongress, one of the biggest events

focused on mobile devices and networks.Back in 2005, any claim that the Webhad a big role to play on mobile deviceswas met with incredulous smiles andfunny stares at best.

Fast forward to the 2013 edition of thecongress, and while W3C certainly hadthe largest HTML5 logo on display at itsbooth, the real message came from thespread of these logos on many, manyother stands. Not only has the Web gainedvisibility on mobiles, with several sys-tems now providing HTML5-baseddevelopment platforms for their systems(Windows 8, Blackberry, Tizen), but also,one of the highest profile announcementsat the congress was around FirefoxOS,Mozilla’s mobile operating systementirely based on Web technologies.

In 2013, W3C showcased demonstra-tions of HTML5 and other open W3C

Web technologies at its booth in the“App Planet” Hall of the new congressvenue, Fira Gran Via, in Barcelona.W3C presented two Web apps featuringthe creation of an online photo galleryfor a variety of devices. In addition,W3C Members such as Adesis,Ayumento de Zaragoza, Intel, Joshfire,Mozilla, Nokia, Tapquo, and partnerssuch as the EU project webinos, demon-strated their unique Open Web Platformapplications.

The W3C booth received much atten-tion from developers, industry leaders,the media, analysts and many others.With this large and broad audience,W3C staff and W3C members werepleased to discuss how the Web is trans-forming all industries, includingmobile, television, advertising, games,publishing, automotive, and health care.Specifically, a wide range of stake-holders are now looking at acceleratingthe adoption of Web technologies in theautomotive industry, leading to the

recent creation of the W3C Automotiveand Web Business Group. Also, theFebruary 2013 W3C electronic booksworkshop reported technical discus-sions focused on Open Web Platformtechnologies currently used in eBooksand the need for improvements of thesetechnologies for future digital publica-tions.

ERCIM is the European host of W3C.

Links: W3C@MWC'13:https://www.w3.org/2013/MWC/webinos project: http://webinos.org/W3C eBooks workshop report:http://www.w3.org/2012/08/electronic-books/rapportebook.htmlW3C Automotive and Open Web Plat-form Business Group:http://www.w3.org/community/autowebplatform

W3C booth at the Mobile World Congress

Events

56 ERCIM NEWS 93 April 2013

Events

http://www.w3.org/community/autowebplatform

Advertisement

docs, 10% academic) with a back-ground mostly on databases (32%),information retrieval (40%), both(15%), and natural language processing(9%). 15 scholarships (supported byELIAS) were made available to assistparticipation. The multidisciplinarity ofthe participants and lectures stimulatedlively exchanges of ideas and manyquestions. Most of the speakers stayedfor the entire week, enriching the dis-cussions. Interestingly enough, theschool turned out to be a brainstormingand discussion opportunity also for thelecturers, since they had the occasion ofmeeting colleagues from different fieldswith different perspectives, on acommon ground of shared topics andissues.

To favour discussion and knowledgesharing, participants were asked tobring a poster describing their ownresearch activities and plans. The threebest posters were awarded a small prizeand the winners were invited to con-tribute a short paper on their activitiesfor the volume on the school lectures,currently under preparation.

The school presentation slides can befound at the link below. Information onall PROMISE publications, includingthe Proceedings of the PROMISE 2012winter school on “Information Retrievalmeets Information Visualization” canbe accessed on the main PROMISE website.

Links:PROMISE: http://www.promise-noe.eu/events/winter-school-2013/ELIAS: http://www.elias-network.eu/

Please contact:Nicola FerroUniversity of Padua, ItalyE-mail: [email protected]

Call for Contributions

ERCIM/EWICS

Workshop on

dependable

Embedded and

Cyberphysical Systems

at SAfECOMP 2013

Toulouse, France, 24 September2013

The ERCIM Working Group onDependable Embedded software-inten-sive Systems will again organize aworkshop at the SAFECOMP confer-ence jointly with EWICS TC7, theEuropean Workshop on IndustrialComputer Systems, TC7, Safety,Reliability and Security.

The first day is dedicated to workshopsand tutorials. The organisers would liketo invite ERCIM members to contributeto and participate in our workshopwhich can be attended independent fromSAFECOMP. No extra costs will arisefor participants since the additional costis covered by the dissemination budgetsof the organizers.

Embedded systems are everywhere –comfort, health, services, safety andsecurity of people depend increasinglyon them. In combination and close inter-action with the real-world environmentand humans, they become so-called“Cyber-physical Systems”, acting inde-pendently, co-operative or as “systems-of-systems”. The impact on society as awhole is tremendous. Smart (embedded)systems are regarded as the most impor-tant business driver for Europeanindustry. They are a targeted researcharea for European ResearchProgrammes in Framework 7, in the

57ERCIM NEWS 93 April 2013

Participants of the PROMISE Winter School

http://www.atlantis-press.com/proceedings/

ARTEMIS Joint Undertaking, and inseveral dedicated Programmes andEuropean Technology Platforms(ARTEMIS, EPoSS) or the future JTIElectronics in Horizon 2020. Theirapplication is not only in the traditionalareas of aerospace, railways, automo-tive, or process industry and manufac-turing, but also in robotics and servicesof all kind, in home appliances (smartenvironments, smart homes, ambientassisted living) and health care.

Sessions are planned on the followingtopics: • dependable and resilient embedded

systems• autonomous systems and robotics• systems-of-systems.

These sessions will cover aspects fromdesign, development, verification andvalidation, certification, maintenance,standardization and education &training. In contrast to the SAFECOMPconference mainstream, the workshopwill include reports on on-going workaiming at discussions and experienceexchange. Reports on European ornational research projects as well asindustrial experience reports are wel-come.

Workshop proceedings will be publishedby LAAS-CNRS, and distributed to par-ticipants during the workshop. The work-shop papers will also be published onlineon the HAL/Arxiv open publication site.

Important dates:• 21 May 2013: Full paper submission

deadline • 15 June 2013: Notification of authors• 28 June 2013: Final camera-ready

paper due

The international programme committeeis composed of selected EWICS andERCIM members, led by the workshoporganizers (see contacts below).

More information: http://conf.laas.fr/SAFECOMP2013/?q=node/28

Please contact: Erwin Schoitsch, AIT Austrian Instituteof Technology, AARIT, AustriaE-mail: [email protected]

Amund SkavhaugNTNU, NorwayE-mail: [email protected]

Call for Abstracts

NETTAB 2013 -

Workshop on Semantic,

Social and Mobile

Applications for

Bioinformatics and

Biomedical Laboratories

Venice, Italy 16-18 October 2013

NETTAB Workshops are a series ofinternational meetings on “NetworkTools and Applications in Biology”introducing the most innovative ICTtools as they are being applied to the bio-medical domain. In the past, workshopshave been focused on themes like multi-agent systems, scientific workflows,Web Services, Semantic Web, collabora-tive research, and biological wikis.

ICTs have permeated society withnewer forms of social participation. Inbiology, we already rely on many socialtools and applications, eg for distrib-uted annotations, Wiki knowledgebases, documentation and productivity.Internet is increasingly accessed withmobile devices. Health and lifestylemobile applications are widely usedand the rapid adoption of mobile solu-tions in medicine and healthcare isalready a reality, but we cannot say thesame for life sciences. Semanticmethodologies and technologies areinstead well established in “-omic”projects and the bioinformatics com-munity was an early adopter ofSemantic Web technologies.

In the NETTAB 2013 workshop,mobile, social and semantic solutionsfor bioinformatics and laboratory infor-matics problems will be explored. Asavvy combination of these technologiescould greatly enhance the research out-come of life scientists and markedlysimplify the workflows in biomedicallaboratories.

Deadlines• 5 July: Oral communications submis-

sion• 31 July: Posters submission.

More information: http://www.nettab.org/2013/

Events

58 ERCIM NEWS 93 April 2013

Call for Papers

Eighteenth

International ERCIM

Workshop on formal

Methods for Industrial

Critical Systems

Madrid, 23-24 September 2013

The aim of the FMICS workshop series,is to provide a forum for researchers whoare interested in the development andapplication of formal methods in indus-try. In particular, FMICS brings togetherscientists and engineers that are active inthe area of formal methods and interest-ed in exchanging their experiences in theindustrial usage of these methods.

Topics of interest include:• Design, specification, code generation

and testing based on formal methods.Methods, techniques and tools to sup-port automated analysis, certification,debugging, learning, optimization andtransformation of complex, distrib-uted, real-time systems and embeddedsystems.

• Verification and validation methodsthat address shortcomings of existingmethods with respect to their industri-al applicability (eg, scalability andusability issues).

• Tools for the development of formaldesign descriptions.

• Case studies and experience reportson industrial applications of formalmethods, focusing on lessons learnedor identification of new researchdirections.

• Impact of the adoption of formalmethods on the development processand associated costs.

• Application of formal methods instandardization and industrial forums.

Important Dates• 3 May: Paper submission• 24 June: Notification• 12 July: Final version due• 23-24 September: Workshop

The workshop is organised by theERCIM Working Group “FormalMethods for Industrial Critical Systems

More information: http://lvl.info.ucl.ac.be/Fmics2013/Fmics2013

http://conf.laas.fr/SAFECOMP2013/?q=node/28


CWI researcher floor Sietsma

Youngest Phd in the Netherlands

Floor Sietsma (20) of CWI in Amsterdam is the youngestPhD in the recent academic history of the Netherlands.Recently elected “Nerd of the Year” by readers of the Dutchmagazine Quest, she was 20 years old when she defended herthesis “Logics of Communication and Knowledge” on 13December 2012 at the University of Amsterdam.

Sietsma’s research concerns knowledge transfer in commu-nication and makes use of epistemic logic, a logic of knowl-edge. She applies this to analyze shared knowledge in com-plex e-mail exchanges with visible (cc) and hidden (bcc)receivers. She also develops a complete logic of manipula-tive communication, for instance in a game of Liar’s Dice,where players mutually agree to lie and deceive.

Sietsma started her PhD research at CWI in 2010 at age 17.She held a Bachelor’s degree in computer science and aMaster’s degree in logic, and was at the age of 12 the

youngest student ever at aDutch university. For herresearch at CWI she wasawarded a special, personalgrant by the NetherlandsOrganisation for ScientificResearch (NWO), because ofher unique talents. 2.5 yearslater, she already had gatheredenough material to defend herthesis. Sietsma also studiesPedagogical Sciences to beable to help gifted children.

59

Advertisement

In Brief

Warsaw Center of Mathematics

and Computer Science Established

The Warsaw Center of Mathematics and Computer Science(WCMCS) is a joint project of two scientific institutions: theFaculty of Mathematics, Informatics and Mechanics of theUniversity of Warsaw (MIMUW), and the Institute ofMathematics of the Polish Academy of Sciences (IMPAN).The Center is built on the longstanding cooperation, in bothteaching and research, between the two. The Center was des-ignated as the “Leading National Research Center” (KrajowyNaukowy Ośrodek Wiodący, KNOW) by the Polish Ministryof Science and Higher Education in July 2012. The Centerorganizes scientific seminars, schools and conferences, andalso participates in several research grants.

WCMCS offers a wide range of cooperation projectsaddressed at students, PhD students, young and seniorresearchers also from outside Poland. The current projectsinclude student internships, PhD internships, PhD fellow-ships, post-doctoral positions, as well as research positionsfor the leaders of research projects. For more detailed infor-mation about the research areas, the current projects, and theapplication procedures, please visit the WCMCS homepage(http://wcmcs.edu.pl).

Photo: CWI

http://OpenConf.org/ERCIM

�� "�� !��"��

��

��"��

ERCIM is the European Host of the World Wide Web Consortium.

Institut National de Recherche en Informatique et en AutomatiqueB.P. 105, F-78153 Le Chesnay, Francehttp://www.inria.fr/

Technical Research Centre of FinlandPO Box 1000FIN-02044 VTT, Finlandhttp://www.vtt.fi/

Austrian Association for Research in ITc/o Österreichische Computer GesellschaftWollzeile 1-3, A-1010 Wien, Austriahttp://www.aarit.at/

Norwegian University of Science and Technology Faculty of Information Technology, Mathematics and Electri-cal Engineering, N 7491 Trondheim, Norwayhttp://www.ntnu.no/

Universty of WarsawFaculty of Mathematics, Informatics and MechanicsBanacha 2, 02-097 Warsaw, Polandhttp://www.mimuw.edu.pl/

Consiglio Nazionale delle Ricerche, ISTI-CNRArea della Ricerca CNR di Pisa, Via G. Moruzzi 1, 56124 Pisa, Italyhttp://www.isti.cnr.it/

Centrum Wiskunde & InformaticaScience Park 123, NL-1098 XG Amsterdam, The Netherlandshttp://www.cwi.nl/

Foundation for Research and Technology – HellasInstitute of Computer ScienceP.O. Box 1385, GR-71110 Heraklion, Crete, Greecehttp://www.ics.forth.gr/FORTH

Fonds National de la Recherche6, rue Antoine de Saint-Exupéry, B.P. 1777L-1017 Luxembourg-Kirchberghttp://www.fnr.lu/

FWOEgmontstraat 5B-1000 Brussels, Belgiumhttp://www.fwo.be/

F.R.S.-FNRSrue d’Egmont 5B-1000 Brussels, Belgiumhttp://www.fnrs.be/

Fraunhofer ICT GroupAnna-Louisa-Karsch-Str. 210178 Berlin, Germanyhttp://www.iuk.fraunhofer.de/

SICS Swedish ICTBox 1263, SE-164 29 Kista, Swedenhttp://www.sics.se/

Swiss Association for Research in Information Technologyc/o Professor Abraham Bernstein, Department of Informatics,University of Zurich, Binzmühlestrasse 14, CH-8050 Zürichhttp://www.sarit.ch

Magyar Tudományos AkadémiaSzámítástechnikai és Automatizálási Kutató IntézetP.O. Box 63, H-1518 Budapest, Hungaryhttp://www.sztaki.hu/

University of CyprusP.O. Box 205371678 Nicosia, Cyprushttp://www.cs.ucy.ac.cy/

Spanish Research Consortium for Informatics and MathematicsD3301, Facultad de Informática, Universidad Politécnica de Madrid28660 Boadilla del Monte, Madrid, Spain,http://www.sparcim.es/

Science and Technology Facilities CouncilRutherford Appleton LaboratoryChilton, Didcot, Oxfordshire OX11 0QX, United Kingdomhttp://www.scitech.ac.uk/

Czech Research Consortium for Informatics and MathematicsFI MU, Botanicka 68a, CZ-602 00 Brno, Czech Republichttp://www.utia.cas.cz/CRCIM/home.html

Subscribe to ERCIM News and order back copies at http://ercim-news.ercim.eu/

ERCIM – the European Research Consortium for Informatics and Mathematics is an organisa-

tion dedicated to the advancement of European research and development, in information

technology and applied mathematics. Its member institutions aim to foster collaborative work

within the European research community and to increase co-operation with European industry.

Portuguese ERCIM Groupingc/o INESC Porto, Campus da FEUP, Rua Dr. Roberto Frias, nº 378,4200-465 Porto, Portugal

I.S.I. - Industrial Systems InstitutePatras Science Park buildingPlatani, PATRAS, Greece, 265 04 http://www.isi.gr

Universty of WroclawInstitute of Computer ScienceJoliot-Curie 15, 50–383 Wroclaw, Polandhttp://www.ii.uni.wroc.pl/

University of SouthamptonUniversity RoadSouthampton SO17 1BJ, United Kingdomhttp://www.southampton.ac.uk/

http://fellowship.ercim.eu

http://ercim-news.ercim.eu

Documents

˚3(&,$/ 6+(0(: ˘2%,/( C20376,1* - ERCIM NewsA/52 ,1 6+,5 ,557(: Keynote Strategies of the European ICST Public Research Organisations towards Horizon2020 by Domenico Laforenza Joint