Prospects for the Internet of Things

Global Synchronization and the Challenges of

Building Network Awareness

The Internet

of Things

Crossroads The ACM Magazine for Students W IN T E R 2 01 5 VOL .22 • NO.2 XRDS.ACM.ORG

XRDS

Check out the new acmqueue app

FREE TO ACM MEMBERS

acmqueue is ACM’s magazine by and for practitioners, bridging the gap between academics and practitioners of computer science. After more than a decade of providing unique perspectives on how current and emerging technologies are being applied in the field, the new acmqueue has evolved into an interactive, socially networked, electronic magazine.

Broaden your knowledge with technical articles focusing on today’s problems affecting CS in practice, video interviews, roundtables, case studies, and lively columns.

Desktop digital edition also available at queue.acm.org. Bimonthly issues free to ACM Professional Members. Annual subscription $19.99 for nonmembers.

Keep up with this fast-paced world on the go. Download the mobile app.

acmqueue_cacm_fp_ads_AS.indd 4 10/6/15 12:34 PM

ACM’s Career & Job Center

Are you looking for your next IT job? Do you need Career Advice?

Visit ACM’s Career & Job Center at: http://jobs.acm.org

The ACM Career & Job Center is the perfect place to begin searching for your next employment opportunity!

Visit today at http://jobs.acm.org

The ACM Career & Job Center offers ACM members a host of career-enhancing benefits:• A highly targeted focus on job

opportunities in the computing industry

• Access to hundreds of industry job postings

• Resume posting keeping you connected to the employment market while letting you maintain full control over your confidential information

• Job Alert system that notifies you of new opportunities matching your criteria

• Career coaching and guidance available from trained experts dedicated to your success

• Free access to a content library of the best career articles compiled from hundreds of sources, and much more!

ACM_printadv1_final.indd 1 11/21/13 11:36 PM

2 X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

Crossroads The ACM Magazine for StudentsW IN T E R 2 0 1 5 V OL .22 • NO.2

2

begin5 LETTER FROM THE EDITORS

8 INBOX

9 INIT A New Revolution is Underway By Laurynas Riliskis

11 ADVICE Addressing Academic Publishing By Marius-Tudor Benea

12 UPDATES Fortifying Computing in Student Societies By Vassilis Kalantzis

13 MILESTONES The Timeline of Things By Jay Patel

14 ACCOLADES Women Who Paved the Way for IoT: A teacher finds her voice By Kayalvizhi Jayavel

16 CAREERS Energy and IOT: An engineer’s perspective By Madison Capps

18 BLOGS Why “Celebrate Women in Computing”? By Nur Al-huda Hamdan

Affordance++: The tale of animating IoT objects By Patrik Jonell and Pedro Lopes

Cover Art by Iwona Usakiewicz / Shutterstock.com

3X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 3

The Internet of Things

Left

and

Mid

dle

Imag

es b

y An

drij

Bor

ys A

ssoc

iate

s /

Shu

tter

stoc

k.co

m; R

ight

Imag

e by

Nic

o El

Nin

o /

Shu

tter

stoc

k.co

m

end70 LABZ

SWARM Joint Open Lab By Teodoro Montanaro

71 BACK IP Addresses By Finn Kuusisto

72 HELLO WORLD According to Sensor 22, Benny is Preparing Dinner By Lara Zupan and Marinka Zitnik

76 ACRONYMS

76 POINTERS

77 EVENTS

80 BEMUSEMENT

features22 FEATURE

The Internet of Things… of All Things By George Hurlburt

28 FEATURE Prospects for the Internet of Things By Vinton G. Cerf

32 FEATURE The Genie in the Machines By Jonathan Caras

36 FEATURE Global Synchronization and the Challenges of Building Network Awareness By Alyssa B. Apsel and Enkhbayasgalan Gantsog

40 FEATURE Trends in Internet of Things Platforms By Michael Andersen

44 FEATURE Querying Flying Robots and Other Things: Ontology-supported stream reasoning By Daniel de Leng

48 FEATURE Toward Computing over Encrypted Data in IoT Systems By Hossein Shafagh

54 FEATURE The Ambient Intelligence Course at Politecnico Di Torino By Luigi De Russis

58 FEATURE Panasonic and the OpenDOF Project: Open-source vision in a large company By Bryant Eastham

62 FEATURE Toward a Web of Systems By Florian Michahelles and Simon Mayer

68 PROFILE Matthew Pryor: Using tech to manage droughts, from Australia to California By Adrian Scoica

32 48 77

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 244

EDI T ORI A L BOA RD

Editor-in-ChiefSean FollmerStanford University, USA

Departments ChiefAdrian ScoicăUniversity of Cambridge, UK

Issue EditorLaurynas Riliskis Stanford University, USA

Issue Feature EditorTalia KohenBar Ilan University, Israel

Feature EditorsSergiy BogomolovUniversity of Freiburg, Germany

Judeth Oden ChoiCarnegie Mellon University, USA

Shudong HaoUniversity of Colorado Boulder, USA

Numair KhanNew York University, USA

Richard GomerUniversity of Southampton, UK

Nidhi RastogiRensselaer Polytechnic Institute (RPI), USA

Yang ShenUCLA, USA

Department EditorsAbhishek BhattacharyaAmity University, India

Marius-Tudor BeneaUPB, Romania and UPMC, France

Tejas KhotMumbai University, India

Bryan KnowlesWestern Kentucky University, USA

Finn KuusistoUniversity of Wisconsin-Madison, USA

Darshit PatelPimpri Chinchwad College of Engineering, India

Nur Al-huda HamdanRWTH Aachen University, Germany

Jay PatelUniversity of California Berkeley, USA

Marinka ZitnikUniversity of Ljubljana, Slovenia

Digital Content EditorPedro LopesHasso Plattner Institut, Germany

Web EditorAbhineet SaxenaGuru Gobind Singh Indraprastha University, India

News EditorsDoris LeeUniversity of California, Berkeley, USA

Marius-Tudor BeneaUPB, Romania and UPMC, France

BloggersDavid Byrd University of Baltimore, USA

David GuerraUniversitat de Lleida, Spain

Vasileios KalantzisUniversity of Minnesota, USA

Vassilios KarakoidasAthens University of Economics and Business, Greece

Maria KechagiaAthens University of Economics and Business, Greece

Norene KellyIowa State University, USA

Dimitris MitropoulosAthens University of Economics and Business, Greece

Wolfgang RichterCarnegie Mellon University, USA

Abhineet SaxenaGuru Gobind Singh Indraprastha University, India

Olivia SimpsonUniversity of California, San Diego, USA

Gustavo Fortes TondelloUniversity of Waterloo, Canada

Udayan UmapathiMIT Media Lab, USA

A DV ISORY BOA RD

Mark Allman, International Computer Science Institute

Bernard Chazelle, Princeton University

Laurie Faith Cranor, Carnegie Mellon

Alan Dix, Lancaster University

David Harel, Weizmann Institute of Science

Panagiotis Takis Metaxas, Wellesley College

Noam Nisan, Hebrew University Jerusalem

Bill Stevenson, Apple, Inc.

Andrew Tuson, City University London

Jeffrey D. Ullman, InfoLab, Stanford University

Moshe Y. Vardi, Rice University

EDI T ORI A L S TA FF

Director, Group Publishing Scott E. Delman

XRDS Managing Editor & Senior Editor at ACM HQ Denise Doig

Production Manager Lynn D’Addessio

Art Direction Andrij Borys Associates, Andrij Borys, Mia Balaquiot

Director of Media Sales Jennifer Ruzicka jen.ruzicka@acm.org

Copyright Permissions Deborah Cotton permissions@acm.org

Public Relations Coordinator Virginia Gold

ACM

Association for Computing Machinery 2 Penn Plaza, Suite 701 New York, NY 10121-0701 USA +1 212-869-7440

CON TAC T General feedback: xrds@acm.org

For submission guidelines, please see http://xrds.acm.org/authorguidelines.cfm

PUBLICATIONS BOARD

Co-Chairs Jack Davidson and Joseph A. Konstan

Board Members Ronald F. Boisvert, Nikil Dutt, Roch Guerrin, Carol Hutchins, Yannis Ioannidis, Catherine C. McGeoch, M. Tamer Ozsu, Mary Lou Soffa

SUBSCRIBE

Subscriptions ($19 per year includes XRDS electronic subscription) are available by becoming an ACM Student Member www.acm.org/ membership/student

Non-member subscriptions: $80 per year http://store.acm.org/acmstore

ACM Member Services To renew your ACM membership or XRDS subscription, please send a letter with your name, address, member number and payment to:

ACM General Post Office P.O. Box 30777 New York, NY 10087-0777 USA

Postal Information XRDS (ISSN# 1528-4981) is published quarterly in spring, winter, summer and fall by Association for Computing Machinery, 2 Penn Plaza, Suite 701, New York, NY 10121. Application to mail at Periodical Postage rates is paid at New York, NY and additional mailing offices.

POSTMASTER: Send addresses change to: XRDS: Crossroads, Association for Computing Machinery, 2 Penn Plaza, Suite 701, New York, NY 10121.

Offering# XRDS0171 ISSN# 1528-4972 (print) ISSN# 1528-4980 (electronic)

Copyright ©2015 by the Association for Computing Machinery, Inc. Permission to make digital or hard copies of part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page or initial screen of the document. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, republish, post on servers, or redistribute requires prior specific permission and a fee. Permissions requests: permissions@acm.org.

COMMUNICATIONSAPPS

Available for iPad, iPhone, and Android

Available for iOS, Android, and Windows

Access the

latest issue,

past issues,

BLOG@CACM,

News, and

more.

5X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

LETTER FROM THE EDITORS

UPCOMING ISSUES

Spring 2016 (March issue)

Digital Fabrication

Summer 2016 (June issue)

Cultures in Computing Deadline: March 4, 2016

Fall 2016 (September issue)

Quantum Computing Deadline: June 3, 2016

A s a child growing up in the ‘80s and ‘90s, my brothers and I endlessly watched “Back to the Future” on VHS tape. For us, 2015 was always the future; the future of flying cars, self-drying jackets, and self-lacing shoes. I dreamed of being a mad inventor like Doc Brown. It was always Doc Brown’s lab that drew me back in—

he had this amazing lab, with all of these different contraptions to make breakfast (to varying degrees of success) and tools to make different inventions come to life.

yond the curtain to see how the sau-sage is actually made; we even have a section in XRDS devoted to looking inside different labs. I recently toured CERN in Geneva and saw the LHC and the CMS lab space, the modern equivalent of Wright’s simple experi-mental setups. As I stood in the gi-ant CMS experiment hall, it felt like a cathedral to science. Inside the giant control room, which operates all of the accelerators, one feels part of this epic quest to understand the world in a deeper, richer way.

Given my love of lab spaces, I was so fortunate to have studied at the MIT Media Lab—a place where you could make almost everything. Fast-forward to the future, now in 2015 I

There is something magical about labs, about spaces for making, for discovering. The paintings of Joseph Wright depicted scenes of inquiry—beautiful, brutal, and a bit frighten-ing—during the 18th century. Anat-omy lessons and experiments with new chemical techniques took cen-ter stage, but always with a view of the laboratory as spectators looked on. These paintings show children and adults wrapped in disbelief and drawn closer into the crude science of the day, hand-made instruments and blown glass gleaming in the light. Wright highlighted science, but also our fascination with it and the spaces in which it is done. We love behind-the-scenes tours, getting to look be-

Back to the Future— Building a great grad research lab

6 X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

guess I am truly a mad inventor. And I am about to get a mad inventor lab, as I start my new role as an Assistant Professor at Stanford University. Now as my time as a graduate student is over, and I am starting my own lab, the question I keep returning to is: What makes a great graduate lab? Ob-viously, the space alone doesn’t make a great lab. Culture, community, ac-cess to tools, and the actual space all contribute to a great research envi-ronment. In looking back at my time at the Media Lab and visiting other labs around the world, here is what I think makes a great lab:

A Community of Diverse Exper-tise. At the Media Lab, we were for-tunate to have such a diverse set of grad students—from CS, EE, ME, and even design and architecture. There was always an expert in any area you could imagine less than three min-utes away, who would be happy to chat about electric field sensing or the stages of childhood educational development. Encouraging the shar-ing of skills and expertise benefits the group in many ways.

Supporting Collaboration. The days of the lone inventor are over. The myth of the great artist still rings through-out the halls of the ivory tower, but collaboration is key to breakthrough work. Encouraging grad students to work together and making everyone feel assured they will get credit for his or her work is key.

Flexible Spaces. At MIT there are stories of the Rad Lab, in Building 20, a temporary space quickly built dur-ing World War II. Because there were always plans to tear down the build-ing, researchers could drill holes in walls or take out the floor to double the height space. They ended up be-ing in that space for 55 years, and this allowed them to have control over the space and enabled rich collabora-tion and research. Having the ability to reconfigure easily allows for more flexible thinking and encourages re-searchers to take ownership of the space. In the d.School at Stanford, Scott Doorley and Scott Witthoft have created a space that begs to be recon-figured—everything is on wheels, mo-bile, and temporary. Their wonderful book Make Space outlines their beliefs

and strategies about how space can influence creativity.

Encourage Prototyping through Access to Tools. Many early research-ers get stuck thinking about research problems, often wondering “Will I find the right one?” or “What about this aspect of the problem?” My be-lief is we think best while actually building and doing the research, whether it is through writing, cod-ing, or machining. Make something and reflect on it. Access to tools plays a big part in that process, as any po-tential roadblock toward getting to the minimum viable prototype takes time away from research. Having the right tools and parts on hand is key to answering questions as fast and as cheaply as possible.

With these principles in hand, I’m off to start my Physical Interfaces Group, where my students and I will investigate the intersection of robot-ics, haptics, and human-computer in-teraction. This means I have to leave XRDS behind. But, as I leave XRDS, I do it knowing that it is in great hands. We have an amazing team of depart-ment and feature editors who work tirelessly to make XRDS shine every issue. In addition, there is great lead-ership of those teams with Adrian Scoica and Pedro Lopes. And last but not least, Denise Doig, our Senior Editor at ACM, wears so many hats and makes sure we stay afloat. I want to thank all of them for making the experience at XRDS so much fun and so inspiring. And, we have found two amazing Editors-in-chief to take over: Jennifer Jacobs of the MIT Media Lab and Okke Schrijvers at Stanford CS. We recently had an in-person meeting with the new EIC’s and other editors to think about the future of XRDS, with ex-citing ideas ranging from video content, to podcasts, to holographic issues of the magazine. Jennifer and Okke have a passion for increasing diver-sity in CS, which, I am sure, will con-tinue to resonate in these pages. It is so exciting for me to see where XRDS will go in the next few years. I will be reading the magazine and follow-ing its journey on the web to find out. I hope you will too.

—Sean Follmer

ACM’s Interactions magazine explores critical relationships between people and technology, showcasing emerging innovations and industry leaders from around the world across important applications of design thinking and the broadening � eld of interaction design.

Our readers represent a growing community of practice that is of increasing and vital global importance.

To learn more about us, visit our award-winning websitehttp://interactions.acm.org

Follow us on Facebook and Twitter

To subscribe:http://www.acm.org/subscribe

INTER AC TION S

Association for Computing Machinery

IX_XRDS_ThirdVertical_V01.indd 1 3/18/15 3:35 PM

INSPIRING MINDS FOR 200 YEARS

Ada’s Legacy illustrates the depth and diversity of writers, things, and makers who have been inspired by Ada Lovelace, the English mathematician and writer.

The volume commemorates the bicentennial of Ada’s birth in December 1815, celebrating her many achievements as well as the impact of her work which reverberated widely since the late 19th century. This is a unique contribution to a resurgence in Lovelace scholarship, thanks to the expanding influence of women in science, technology, engineering and mathematics.

ACM Books is a new series of high quality books for the computer science community, published by the Association for Computing Machinery with Morgan & Claypool Publishers.

8

begin

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

Imag

e by

Mik

el M

arti

nez

de O

saba

INBOX

Semi-pro filmmaker, editor, AE animator and VO artist; Twitter (@AndrewJHCI)

OTHER TWEETSBlack Cat newspaper is out! http://paper.li/m_mina/1317191278?edition_id=f4c9dc20-6e35-11e5-a927-0cc47a0d164b Stories via @XRDS_ACM @c_z @FromAtom—Minakuchi, M. Frivolous HCI researcher / nominal professor @ Kyoto Sangyo University; Twitter (@m_mina)

EXPLORING INNOVATIONRegistration is now open for ExploraVision, a K-12 science competition https://shar.es/17PR6a —ACM XRDS Magazine XRDS is ACM’s magazine for students by students; Twitter (@XRDS_ACM)

@XRDS_ACM Thanks for sharing! Do you know of any young scientists who are submitting projects?—Toshiba Innovation Innovative people sharing bright ideas! Toshiba Innovation is your hub for STEM education, ExploraVision and emerging technology news; Twitter (@ToshibaInnovate)

ACM CHIPLAY 2015That’s it. To quote @toupsz GG #chiplay15. Thanks @GustavoTondello for the @XRDS_ACM roundup http://xrds.acm.org/blog/2015/10/acm-chi-play-2015-xrds-insiders-view/ —Lennart Nacke User Experience Designer, Associate Professor @hcigamesgroup, Games User Researcher; Twitter (@acagamic)

Be sure to check my post! ACM CHI PLAY 2015: XRDS insider’s view! http://xrds.acm.org/blog/2015/10/acm-chi-play-2015-xrds-insiders-view/ #chiplay15—Gustavo Tondello Gamification Scientist and Consultant, Ph.D. student,

HCI Games Group. Logosophy researcher; Twitter (@GustavoTondello)

@XRDS_ACM @Gustavo-Tondello @RinaRenewe r all here at CHIPLAY 15—Dennis L. Kappen Gamification, Industrial Designer, User Interaction Designer, Brand Experience Designer, Exergaming Research; Twitter (@3D_Ideation)

EXTRA, EXTRA READ ALL ABOUT VR@XRDS_ACM just pub-lished our article on how Substitutional Real-ity can help bring #VR home! @ed_velloso http://

xrds.acm.org/article.cfm?aid=2810044 …—Adalberto Simeone Lecturer at the University of Portsmouth, UK; Twitter (@Adal_LS)

My first blog post at @XRDS_ACM on Virtual Reality: http://xrds.acm.org/blog/2015/10/explor-ing-virtual-reality-are-we-there-yet/ …—Andrew J Hunsucker Ph.D. student at Indiana University in Human Computer Interaction with a focus on design pedagogy.

How to contact XRDS: Send a letter to the editors or other feedback by email (xrds@acm.org), on Facebook by posting on our group page (http://tinyurl.com/XRDS-Facebook), via Twitter by using #XRDS_ACM in any message, or by post to ACM Attn: XRDS, 2 Penn Plaza, Suite 701, New York, New York 10121, U.S.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 9

INIT

computers. Cerf discusses considerations for the IoT when it comes to interoper-ability and standardization, as well as security and privacy. He addresses important ques-tions like: What are the trade-offs between interoperability and long-term profitability of a company, the balance be-tween giving a doctor access to medical information in an emergency while protecting patient data against nefarious uses, and who will, or should, take responsibility when things go wrong?

Things can go wrong not because of maliciousness, but because of miscommu-nication. With so many new devices we need new ways to interact with them, both for them to understand us and for us to understand them. Jonathan Caras addresses the important topic of inter-action with things, and how this can lead to magical ex-periences. Not only is the way we will interact with the sen-sors important, but also the capability to personalize the experience. With all the data produced from surround-ing devices, we will need to choose and prioritize—very much as we do today with our news feed—what is impor-tant to us and what is not. It is the possibility of person-alization and interaction ev-erywhere that will create the magic of the IoT.

To have a great user experi-ence we need to enable things

Every second of our lives, we rely on five basic senses: sight, touch, hearing, smell,

and taste. Using these senses we craft, improve, and prog-ress as a human race. With re-spect to sensing, Lord Kelvin famously made the follow-ing observation in the 19th century: “If you can not mea-sure it, you can not improve it.” Since then, nearly two centuries of groundbreak-ing research in mathemat-ics, physics, and chemistry created the scientific revolu-tion that has resulted in the technologies surrounding us today. This revolution made it possible to measure and quantify every aspect of the physical world with a much greater accuracy. Today, we have robots in factories that manufacture faster and more precisely than ever before, de-vices in buildings that mea-sure and adjust heating and air conditioning, and even cars that can drive mostly on their own. Tomorrow, the Internet of Things (IoT) will take this further by seam-lessly expanding our sensing capabilities across the globe with only our imagination as a limit.

The IoT refers to the idea of connecting “everything” to the Internet. This will change the landscape of the Internet as we know it to-day. No longer crowded with computers and data centers, the Internet will be huddled

with sensors and actuators—the “things.” Smartphones, watches, thermostats, ovens, washing machines, fitness trackers, glucose meters, and cars are already part of our lives in the IoT. The pro-liferation of more connected devices that can sense, act, and communicate opens the door to many new applica-tions. Imagine self-driving cars that communicate to drive closer together or fur-ther apart, shoes that warn you about obstacles, or facto-ries and farms that directly connect to shops in order to produce the exact amount of groceries needed for cus-tomers. Minimizing waste, optimizing efficiency, and improving the quality of our lives by reducing the num-ber of low-level decisions we need to take will give us back time to do greater things. The IoT is not a far-off dream; it is

happening all around us. Ac-cording to ABI Research, by 2020 more than 40 billion de-vices will be connected to the Internet, with the potential to impact our lives much like the invention of the printing press did in the 15th century.

The current issue of XRDS provides us an overview of IoT, covers groundbreaking research enabling the IoT, and addresses the safety and security issues that need to be solved going forward. We begin with George Hurlburt, who discusses the capabili-ties and potential for the IoT regarding sensing, thinking, and acting. Hurlburt pleas for a multidisciplinary approach for unlocking the IoT’s full potential. Former ACM Presi-dent and the “Father of the In-ternet,” Vint Cerf, asks what happens if 100,000 refrigera-tors attack Bank of America. All these new Internet-con-nected devices could poten-tially be reprogrammed to send viruses to the network that also hosts the bank’s

A New Revolution is Underway

We are in the first stages of a new revolution in which technology is becoming a more intricate part of our lives...

Gartner’s report placed IoT at the top of the hype cycle for emerging technologies in 2015 (for the second year in a row), and estimated plateau reach in 5-10 years.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 210

begin

10

the open-source community to accelerate the emergence of the IoT. While, Florian Mi-chahelles and Simon Mayer describe Siemens’ efforts to utilize the IoT in industrial automation, energy genera-tion and transmission, con-trol technology, mobility, and medical technology.

The things are here to stay; they will become smart-er, more accurate, better con-nected, and more power-effi-cient as time passes. Things integrated with reasoning and artificial intelligence have the potential to become our personal shopper and chef; to revolutionize health care (e.g., sensors detecting early symptoms of diseases); to enable smart cars leading to fewer accidents, less con-gestion, and mitigate pollu-tion; and for countless other applications. We are in the first stages of a new revolu-tion in which technology is becoming a more intricate part of our lives, with the po-tential to have a profound im-pact. To achieve this goal, we need to deal with challenges in engineering, human-com-puter interaction, reasoning, while safeguarding privacy and security.

Are you on board? —Laurynas Riliskis,

Issue Editor

Biography

Laurynas Riliskis is a postdoc at The Stanford Information Networks Group doing research in Secure IoT.

to communicate efficiently with each other. Low-power devices communicate in delay-tolerant networks with con-straints very different from ones in the traditional com-munication networks. The IoT requires extreme power effi-ciency and an ever-changing network topology, whereas the traditional Internet has many plugged-in machines and routers that connect with physical wires. Alyssa B. Apsel and Enkhbayasgalan Gantsog write about a new communi-cation method that uses little power and finds new devices efficiently. Inspired by the Southeast Asian male firefly, famous for synchronizing its flashes to attract mates, their research involves replicating the way fireflies synchronize flashes to find and commu-nicate with other radios. This method consumes as little as 100 µW while transmitting at 150Kbits/sec speed, which is four orders of magnitude more efficient than tradition-al Wi-Fi.

Besides low-power con-sumption, the devices have to be tiny and cheap—would you pay $1000 for a fitness tracker that weighs five pounds? At the same time, they have to be able to compute, communicate, store data, and actuate in the physical world. Creating such small devices requires tiny electronic components that are big troublemakers. Michael Andersen discusses

current trends in IoT devel-opment and describes what it takes to construct the physi-cal thing, the enabler of IoT: the ultra-low-power embed-ded device. Building these ultra-small devices brings new challenges: How do we test and rapidly develop such things? How can we measure and profile their power per-formance? And, finally, how can we keep prices low?

Imagine trillions of de-vices sampling every bit of our existence and continu-ously streaming data. This zettabyte information flow will only be useful if it can be interpreted in a meaningful way. In its essence, the sensor converts the observation of a physical phenomenon into an electrical output with a much greater resolution and speed than a human can. However, we need ways to avail and analyze this infor-mation for it to be useful to people. Daniel de Leng takes on this challenge and shares his research about using tem-poral logic and ontologies to reason about the world in a resource-constrained envi-ronment. As a working exam-ple, he discusses deploying drones for perimeter moni-toring.

Another approach to ana-lyzing all this data is to do it in the cloud. There is often val-ue in aggregating the data of many sensors to find trends and make inferences. A pre-requisite for enabling this

technology is the privacy of users should be guaranteed. However, a recent study from HP Labs found 80 percent of IoT devices have serious vul-nerabilities, 60 percent used unencrypted network traf-fic, and 70 percent did not require secure passwords. Security is always difficult, especially on resource-con-strained devices that are small and easy to lose. There-fore, it becomes a necessity to protect the data rather than the embedded device. Hos-sein Shafagh discusses the particular challenges that the IoT faces in computation on encrypted data.

The IoT extends far be-yond traditional computer science to chemistry, phys-ics, mechanical engineering, and psychology. This poses a challenge: How can we educate the designers and researchers in the multidisci-plinary future IoT? Luigi De Russis shares his experience teaching an undergraduate course that provides a holis-tic IoT development expe-rience. Such courses are a big welcome from industry, which is committing sig-nificant resources to IoT re-search and the development of tools and products.

Finally, we highlight IoT from an industry perspective. Bryant Eastham shares Pana-sonic’s vision of an open and secure IoT. The electronic gi-ant has released 10 years of research and development to

The number of objects experts estimate will make up the Internet of Things by 2020.

50B

11X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

ADVICE

Imag

e by

Yin

gko

/ sh

utte

rsto

ck.c

om

Addressing Academic Publishing

Whether you are working on a Ph.D. or you have already started research during your academic career,

there is one important topic you should learn how to deal with. Academic pub-lishing. Its main role is to offer you the tools to give visibility to your outstand-ing research project and its important results. And I mean it: Your research cannot be anything less than outstand-ing and the results less than important! The point is to make the world under-stand your opinion and consider your research in a favorable light. Provided you have the project, some results, and an absorbing story deeply rooted in the existing literature, how do you ensure the best outcomes?

The universally accepted units of measurement for outcomes of re-search are represented by scientific papers and by their citations, which underlie the metrics for evaluating both researchers and scientific jour-nals. By metrics we mean total count, the Hirsch index, the impact factor, or the immediacy index. While the metrics have their strengths and their flaws (being continuously improved or replaced by more advanced ones), the key to success lies, as in many other situations, in the art of finding a bal-ance between quantity (regarded as the number of papers published), qual-ity (generally interdependent with the potential for citations), and honesty.

Quantity is sometimes associated with your survival as a researcher. You are often constrained by the institu-tion you are affiliated with, or by your funder, to publish a required mini-mum number of papers. This can lead to some quality concessions but, at the same time, allows you to continue your research and, consequently, to gradu-

ally improve your work and your sub-sequent papers. Moreover, the need for quantity facilitates more interactions with other researchers, through peer-reviews or conference participations, which can also create the circum-stances for achieving improved qual-ity. Being a jointly supervised Ph.D. student, enrolled at two universities, I witnessed how one institution asked its students for a higher number of pa-pers, while one article in a good publi-cation was enough for the other. In the end, it is in your power to choose and this can be a first step in making the

world understand your opinion. Quality is desirable and you can

complement it by aligning with a high-ly ranked journal or conference, assur-ing your research receives better feed-back and increased visibility. Although the metrics used by rankings evolve, a good scientific journal or conference keeps up with the changes and it will continue to attract some of the best re-searchers. Therefore, tools like Thom-son Reuters’ Journal Citation Reports, Science Citation Index, Conference Proceedings Citation Index, or the CORE Conference Rankings will al-ways prove useful in making a choice. Your supervisor will also offer you considerable help in this direction, as he or she knows your work better that anyone else. Finally, as a CS student you are accustomed to ever changing technology, so don’t be afraid to exper-iment with the newest tools, like open access publishing or social networks for academics, such as ResearchGate.

Honesty, with yourself and with the other researchers, is and will un-doubtedly continue to be paramount for maintaining academic publish-ing standards. You should be honest weighing your work against the exist-ing literature. You should be honest deciding where to publish a paper. You should also be honest about the best moment to do it. In 2012 I attended a meeting of Ph.D. students with Dr. Les-lie Lamport. I recall him saying he was always guided by a principle of finish the paper first and then start to seek the right publication for it. Two years later I better understood the impor-tance of his words. There is no higher recognition for using such a strategy than producing Turing-award winning research.

—Marius-Tudor Benea

Hospital equipment and scrubs can report harmful bacteria and viruses before being used.

12

begin

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

Pho

to b

y S

arah

Sch

atts

chne

ider

UPDATES

Fortifying Computing in Student Societies A focus on outreach initiatives.

RPI’s ACM-W members gather at The Grace Hopper Celebration of Women in Computing Conference 2015.

W ith computing and com-puter technology being present in virtually every aspect of modern life, it

is certainly not surprising that cur-rent enrollment rates in computer science-related departments is steadi-ly increasing. For some students, the concept of studying computer science was an everlasting call since their first interaction with a computer. For oth-ers, computer science is an intriguing field that allows one to explore and ex-press creativity.

Having a genuine interest certain-ly is the first step, but feeling part of the community is equally important on the road to a successful academic career. ACM student chapters across

the globe have been very active in pro-moting computing and enhancing its spirit. In this issue, we focus on the actions of two chapters: the ACM stu-dent chapter at the University of Min-nesota (UMN), Twin-Cities; and the ACM-W student chapter at Rensselaer Polytechnic Institute (RPI).

At UMN, the ACM student chapter has long devoted efforts to further-ing the interest and knowledge of its members in the field of comput-ing. They offer a number of ways for members to get involved in the com-puter science program. The chapter also promotes great opportunities for leadership and networking within the field. Chief among these is Min-neHack, the annual hackathon event

put on by the group. Supported by Major League Hacking,1 this is one of the largest hackathon events in the Minnesota area. It gives students of all skill levels a chance to hone their skills and bond with their peers. ACM-UMN also puts on industry tech talks and an academic lecture series each semester to encompass both the in-dustrial and academic realms of com-puter science.

The single most important activity of the chapter is community build-ing. According to chapter officer Yev-geniya Polukeyeva, “The ACM-UMN chapter fosters a supportive commu-nity for all of its members. No mat-ter the skill level or student status, a welcoming hand is extended to all. Members learn and grow every day from interacting with their peers, who are knowledgeable not only in techni-cal endeavors, but in numerous other areas as well. The supportive environ-ment of the group keeps people com-ing back and promotes longevity with-in the computer science program, as well as cultivating a broad range of re-lationships with fellow members that last well past the school years.”

At RPI, the ACM-W chapter is also heavily involved with establishing opportunities for outreach and en-hancing the role of female students in computing. Since its founding, ACM-W @ RPI has already accomplished great milestones in alignment with

1 If you are interested in hackathons, Major League Hacking is organizing a series of events in different countries, check out the schedule: https://mlh.io.

The year when the underlying con-cept of IoT was first introduced in Jay B. Nash’s book Spectatoritis.

1932

13X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

its mission to support and strengthen diversity within computing at RPI. According to their chair, Sarabeth Jaffe, “We have been very active all this time. Last semester, we partnered with NYISO to put on a Toastmasters workshop to help our members mas-ter the art of public speaking. Some of our members attended the inaugural GHC event hosted by ABI in New York, a one-day immersive event modeled after the Grace Hopper Celebration of Women in Computing Conference. Chapter members also attended NYC-WiC hosted in Syracuse, NY, a confer-ence aimed at promoting the academ-ic, social, and professional growth of its participants by bringing together talented students, faculty, and in-dustry leaders.” Sarabeth adds, “Our members networked with some of the biggest names in tech and even got the chance to have dinner with ACM-W Chair and Union College Professor of Computer Science, Valerie Barr. This semester, we’ve held a women in CS meetup geared toward creating a supportive environment for freshmen CS majors.” This past October, the chapter sent more than 27 students and faculty members to the 2015 Grace Hopper Celebration of Women in Computing in Houston, TX.

Biography

Vassilis Kalantzis received his computer engineering diploma in 2011 and his master’s degree in computer science and technology in 2014, both from the Computer Engineering and Informatics Department, University of Patras, Greece. Since 2013 he has been a Ph.D. candidate with the Computer Science and Engineering Department at the University of Minnesota. His research interests span the areas of numerical linear algebra and parallel computing with applications in the fields of big data analytics and physics.

The Timeline of Things

The Internet of Things (IoT) encompasses many interacting components including item identification, networks, sensors, and communications protocols. Presented are a few key inventions that have helped drive the field forward.

1948 Norman Woodland and Bernard Silver begin exploring early bulls-eye shaped barcodes,

which will later allow retailers, such as grocery stores, to automatically track product information and inventory.

1968 Ted Paraskevakos develops a method on top of the telephone system to automatically

identify callers—the predecessor to our modern-day caller ID. Paraskevakos’s system is one of the earliest demonstrations of machine-to-machine communication.

1982 Carnegie Mellon students connect their Coke vending machine to the Internet. They

can view the quantity and temperature of the sodas from their computer terminals.

1999 Kevin Ashton, who coined the phrase “Internet of Things,” co-founds the Auto-

ID Labs at MIT. The primary goal of the lab is to further develop radio-frequency identification (RFID) as pervasive, inexpensive identifiers for individual items.

2011 Nest Labs releases their Learning Thermostat product. This thermostat is

sensor-driven and Wi-Fi connected, and is one of the most recognizable domestic IoT innovations today.

—Jay Patel

MILESTONES

Machines can tune in to the sound of their engines, learn from other machines what a failure sounds like, and if necessary, alert their owner.

14

begin

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

begin

ACM-W ambassador. I served as the organization’s faculty advisor for two years. Appropriately enough, we began operations on March 8th, International Women’s Day. Later, I helped establish the ACM-W chapter at Amrita Universi-ty in Coimbatore. In 2011, I became in-volved with JDuchess-Chennai, a Java user group for women located in and around Chennai, the city that is home to two of SRM’s campuses. I was hon-ored to represent JDuchess-Chennai at an Oracle Java conference in May 2012, which was held in Hyderabad. I was also interviewed about the role of women in technology. A month later, back at SRM, together with other wom-en of ACM, I organized an internation-al women in computing conference.

Three years ago, I began working on a Ph.D. in the Internet of Things (IoT), at a time when most people had never heard of such a thing. How, they asked, could I work on an advanced degree for something that did not exist? But, once again, I believed in myself, and I be-lieved in how important the IoT would become.

In 2014 my students and I started a club at SRM called the Internet of Things Alliance, or IOTA for short. The club is steadily growing, and has been incredibly active in a very short period of time, especially during the summer months. To highlight a few of IOTA’s accomplishments, in the last two years we conducted two summer intern-ships. One on embedded systems—as that was my post-graduate specializa-tion—and one on the IoT. Both were free of cost, and we trained nearly 50 students. We also held two exhibitions to show our work to the SRM commu-nity. Our tech fest, “Aaruush,” is an an-

Editor’s Note: In the summer of 2015, the newly-established UPES ACM-W student chapter reached out to XRDS with an idea to organize an article writing competition to promote the visibility of women in computing. Contestants were asked to write about the role played by female scientists in the emergence of the Internet of Things.

The competition ran for three months, and concluded on October 1st.After receiving submissions from all across India, the top three winners were selected by an editorial jury from XRDS.

It is our pleasure to share with you the winning submission; a personal story about a great teacher who grows into her role as a computer science educator.

—Adrian Scoică, XRDS Departments Chief

If my story connects with a single person, I will have succeeded.

I am Kayalvizhi Jayavel, an as-sistant professor in information

technology at Sri Ramaswamy Memo-rial (SRM) University in India. I love my job, but 20 years ago I never imagined teaching as my calling. This is my story.

As a young girl, as with so many others like me, my sole goal was to get good marks in school, an objective I met with regularity. By the time 12th grade rolled around, or 12th standard in my country, I aspired for admittance into a medical or engineering school and the status such a career would pro-vide. But I also wanted as high a return on as little effort as I could muster, so I turned to computer science. I wasn’t particularly enamored of the field or anything, but it came with a lot of hype back then. Life had other plans,

though, and in 1997 I was selected for the electronics and communication engineering (ECE) program. In a word, ECE was all Greek to me!

Even though I didn’t really under-stand what was going on, I still knew how to do one thing: continue to get the grades. It was easy for me. I saw the Indian educational system as one that did not necessarily check for skills or comprehension, just the ability to successfully complete tests. I longed to be able to apply the knowledge, but believed everyone simply got along the way I did. Until one day in 2001, the last year of my undergraduate education, I realized that was not the case.

After observing my peers, I was shocked to learn many of them not only understood but applied what they had learned. It dawned on me, it was not the educational system that needed to change, but myself. And slowly I did. I had always enjoyed teaching, and one fine day in 2003 SRM University called to interview me for a position. I was thrilled, and even more thrilled when they hired me. Another unexpected twist came my way when they hired me not for electronics and communication, but information technology. My friends and colleagues warned me to be care-ful; not only would the subjects be all new, but when it came time to move on from teaching no one would hire me in either field. I paid them no mind. I be-lieved in myself.

It’s been a busy, challenging, and ex-hilarating 15 years since, punctuated by a wide variety of significant events for both me and the university. In 2010, I created the ACM student chapter for women (SRM ACM-W)—an event inau-gurated by Gayatri Buragohain, India’s

Women Who Paved the Way for IoT: A teacher finds her voice

ACCOLADES

C2Sense sensors can detect high levels of ethylene, and notify you when certain fruits are ripening or spoiling.

15X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

of none,” I believe the IoT demands its practitioners be a Jack or Jill of all trades. Steve Jobs said “dots connect backwards.” One will not know when things happen, or necessarily why they happen, but when they connect, it can be beautiful. My students have complimented me by saying I am ex-traordinary at connecting seemingly unrelated topics. Working for women, collaborating with ACM, and being on the ground floor of IoT have all been the dots that have connected for me. I believe in the future of the IoT, and the future of women in this field at my university and beyond.

Biography

Kayalvizhi Jayavel works as an assistant professor in the Department of Information Technology at SRM University. She leads the Internet of Things alliance (IOTA) club at SRM. The club aims to nurture young talents, develop pilot projects, and publish papers on the same.

nual event celebrating the spirit of in-novation at SRM. During the event, we held an awareness program for the en-tire University under a program called “Wednesday Speaks.”

In February 2015, the IOTA Club conducted a workshop on “IoT and Raspberry Pi” in collaboration with EFY (Electronics for You) Techcentre, which attracted nearly 100 students. Syam Madanapalli, an IoT pioneer in India, helped lead the course. We had previously been involved with EFY; in November 2014, I attended an IoT workshop in Bangalore organized by the group. On campus we began con-ducting weekly meetings to motivate students to pursue careers in IoT. A month later, in March, IOTA was short-listed to participate in the Smart City Research Colloquium held at IIIT Alla-habad, which the Indian government has declared the lead university for its “smart city” initiative. I presented my paper there. Later that month, we gave a demonstration and speech about “Ar-duino and IoT” at Arduino Day. Adding more meaning to our work, my team and I recently participated in a work-shop on the IoT and its relevance to all, which was held at Gnanamani College of Technology, Tamil Nadu. Hearing about our work, Cognizant Technol-ogy Systems, an international IT com-pany, sent their employees to study the feasibility of an industry-academic col-laboration with our team. This would be 100 percent interdisciplinary, as we know there can be no IoT without working across disciplines.

Our future looks as bright as our past. We are implementing a smart campus within our University. And re-cently, two of my students attended an

IoT hackathon in Bangalore. The best part of all this is my female students—who always kept away from embedded systems, hardware, and IoT—have start-ed working on projects and papers. We are expecting a lot more activity in IoT from them and others on our campus.

There is an old proverb that says when a flower blooms at its fullest, nothing can stop its fragrance. With IOTA, our goal is to work quietly and keep a low profile. One day, we believe, our accomplishments will speak loud-ly for themselves.

I mention all of these things about myself, and my years at SRM, to make a broader point. It is important to have a varied set of skills and to understand how all of our work is related. While my friends and peers remarked “you will be a jack of all trades, but king (re-ally, shouldn’t they have said queen?)

Technologist Kevin Ashton coins the term “Internet of Things.”

1999

16

begin

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

begin

CAREERS

of which provides data every 15 min-utes. In addition to the information generated by the smart meters, data is also received from a similar number of asset monitoring equipment, weather sensors, and power flow sensors. Alto-gether, these sensors generate more than 1TB of data per day that our soft-ware platform must ingest, standard-ize, and process in the cloud. Going forward, these numbers are expected to grow dramatically. The number of smart meters in Europe alone, for in-stance, is projected to exceed 240 mil-lion by 2020.

As the networks we work with ex-pand in size, we have to evolve our ca-pabilities in both scale and complexity, which creates fascinating engineering challenges. Along with the questions of how to store, process, and access such quantities of data, we must address how to identify and highlight mean-ingful portions of this information so our customers can leverage it to their fullest advantage. The accumulation of data is only as useful as what can be extracted from the data. Companies that can master these hidden insights can gain a significant competitive ad-vantage. Big data analysis of online search behavior, for example, led to un-precedented changes in understand-ing consumer retail and commercial opportunities, such as targeted adver-tising, long-tail purchasing behavior, and consumer segmentation. Now the analysis of new sensor data is unlock-ing previously impossible business problems for our customers, such as predicting asset failures, real-time cus-tomer segmentation, detecting theft, and understanding the overall health of the sensor network.

T here’s something uniquely re-warding about applying tech-nology to an industry that delivers real-world, tangible

results. Beyond the inherent satisfac-tion of creation that can arise from software engineering, there’s the added bonus of knowing your code is touching the world in a very physical way. After I graduated with my under-graduate degree from the Massachu-setts Institute of Technology in 2009, I spent a few years in cognitive science research labs working to understand human vision and its applications to computer vision. However, I found my-self longing to work on something with a more immediate and broader reach, and a few years later I found my perfect match: working with cyber-physical systems and the “Internet of Energy” at C3 Energy. Here, I lead a team that is creating visual application develop-ment tools for our internal engineers, external customers, and partners.

INTERNET OF ENERGYC3 Energy was founded in 2009 by a group of Silicon Valley software vet-erans, including our CEO Tom Siebel,

with the goal of building a next-gen-eration software analytics platform that would help change and improve how the energy grid operates. We pro-vide energy companies with software for customer energy analytics, sensor and smart meter analytics, and predic-tive maintenance. With more than 100 employees, over half of whom work as engineers, each individual has room for real project ownership and impact as we tackle the technical challenges posed by processing and leveraging data from these massive networks. The ability to truly own my work and have a strong voice in the development process was a major influencing factor in my choice to work at C3 Energy. As a new engineer at the company, I was thrilled my first feature made it to pro-duction in less than three weeks.

During the past five years, many in-dustries have witnessed an explosion of investments in sensors and measure-ment technologies as well as the emer-gence of the Internet of Things (IoT). Energy companies—an industry in a market of more than 2 trillion USD—are no exception to these developments. As our customers deploy millions of sensors, their decades-old operating processes must change dramatically to handle these new data streams. Hav-ing a front-row seat to this transforma-tion has been extraordinary. With each new feature I work on, I can see work-flows become more efficient, savings increase, and communication improve between what were formerly isolated “data silos” of information.

QUICKLY GROWING COMPLEXITYC3 Energy’s current customers operate 55 million smart meters globally, each

Energy and IOT: An engineer’s perspective

The relative isolation of academic programming is replaced with a culture of building on each other’s strengths.

Smart cement, equipped with sensors, can monitor stresses and cracks, and send alerts of problems to avoid catastrophes.

17X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

Imag

e C

ourt

esy

of C

3 En

ergy

gether and error-prone workarounds. For those of us at C3 Energy, working to perfect these skills will enable us to continue providing our customers new insights into their networks and to surmount all challenges, which the ever-increasing number of sensors in the Internet of Energy present.

Biography

Madison Capps received her undergraduate degree from MIT in 2009 and was subsequently a research assistant at MIT and Brown University. Moving to Silicon Valley, she worked for a shipping solutions company and an online education startup before finding C3 Energy, where she works as a software engineer.

Given the budding state of the Internet of Energy, we also face the challenge of making our applications flexible enough to quickly and easily adapt to new directions of inquiry and interest. We are even working toward empowering our customers to build applications independently. My team specifically focuses on developing and maintaining a framework and related visual tools for creating applications. “Modular, configurable, and gener-alizable” has become our unofficial mantra, and an average day finds me equally likely to be sketching mock-ups, discussing APIs, or creating reus-able JavaScript components.

ADAPTABLE, FUTURE-PROOF ENGINEERINGOne thing that excites me about work-ing in this industry of smart and con-nected devices is how the engineering processes are as evolved and connect-ed as the physical technologies. Soft-ware engineering is a collaborative and communicative development environ-ment. Working code is a priority, but the relative isolation of academic pro-gramming is replaced with a culture of building on each other’s strengths. Our success is as driven by our ability to work together through all stages of development from the very onset of a project, as by our individual talents.

Regardless of what sector you may find yourself working in, learning to communicate technical concepts and requirements, as well as thinking about how code fits into the greater structure of a codebase, are the great-est skills you can develop. Design, both visual and technical, is key to creating a clean product, and the best way to

develop this skill is to practice. Keep lists of potential edge cases, sketch mockups, explain code to or work with other programmers, and be mindful of how code will need to change as new requirements are created. While being intimately acquainted with the full stack is not always necessary, un-derstanding the basic difficulties and standards at each layer enables you to make more informed design choices. Knowing which solutions are a best fit for each part of the code base also promotes more scalability and can eliminate “hacking” of features to-

Smart city grids will communicate with smart cars to optimize traffic flow. A stoplight will only turn red if necessary.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 218

begin

BLOGS

Why “Celebrate Women in Computing”? By Nur Al-huda Hamdan

The ACM-W society is one of the biggest advocates of wom-en in computing. They dedicate several events and awards to celebrate prominent women in computer science and related fields. In September of this year, the ACM-W Europe chapter held their second womENcourage (http://women-courage.acm.org/) event at Uppsala University in Sweden. womENcourage creates an environment for women with similar scientific backgrounds to interact, network, and explore career opportunities. At this event, 200 people at-tended from 28 countries including the Middle East, India, China and the U.S.A.

But why is there even a need for events dedicated to women in computer science?

Gender stereotypes threaten women in male-dominated work environments with discrimination from three sources: men, other women, and self-discrimination. Commonly, the ratio of women to men decreases rapidly in more advanced academic or professional positions. In her keynote, Prof. Åsa Cajander mentioned as a consequence of this phenomenon women are perceived less competent within a group and are assigned to the group’s social tasks.

This leaves a woman feeling isolated within her team, and could eventually affect her performance.

A higher risk women face comes from within. Prof. Cajander called this risk the “imposter syndrome,” where a woman feels she does not deserve her success and assigns it to chance or to other people. Some women also believe similar success could have been achieved by a male counterpart with less time and effort. Positive discrimination, such as scholarships offered for women or women quota systems, also threaten women. In many situations, this type of discrimination leads women to be more criticized for their actions compared to males, especially by other women.

Events such as womENcourage help individuals become aware of these risks and provide opportunities to mediate them. At the event’s core, it provides the opportunity to network. Women are able to get acquainted with each other in a friendly social-scientific atmosphere and share their experiences. Discussion groups further aim to build mutual understanding and empathy, and to evaluate different techniques to neutralize gender stereotypes. Aside from the social aspects of the conference, many scientific sessions are offered. Technical talks are presented by leading researchers and industry representatives. Due to the nature and purpose of the conference, however, the scientific content covers a variety of topics without going deep into them. One advantage of Im

age

by S

erge

y N

iven

s

The XRDS blog highlights a range of topics from conference coverage, to security and privacy, to CS theory. Selected blog posts, edited for print, are featured in every issue. Please visit xrds.acm.org/blog to read each post in its entirety. If you are interested in joining as a student blogger, please contact us.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 19

this model is participants can choose to attend any session and learn about new topics with little-to-no prerequisites.

Several hands-on workshops also take place and provide basic training for new technical skills. The value of the workshops stems from the variety, recency, and type of exposure they offer participants. A career fair is also held as part of the conference. The fair provides women, and female students in particular, with a rare opportunity to interact with industry representatives without the potentially overpowering male presence typical of other conferences. During this event, ideas are discussed, business cards are exchanged, and new collaborations are planned.

In addition, a Codess hackathon was co-located within the conference this year and took place one day before the main event. Five teams competed to design and develop hardware prototypes using new Intel technology for a humanitarian purpose. Codess by Microsoft is another type of initiative that encourages gender diversity in computer engineering and development. Such technical challenges help women gain confidence and accept positive recognition.

One of the main conference topics was teaching computer science. In the near future, computer skills will not be restricted to a number of university programs or job titles, but will become essential tools in the hands of the skilled. ACM-W could take advantage of events such as womENcourage to expose women from other fields to computer science and provide them with the basic skills to empower them in their work. Currently, womENcourage reaches out to women who are already involved in computer science disciplines. Inviting female students in middle school and high school to participate can bear many

advantages. On one hand, this would provide women a chance to act as mentors for these girls. Mentorship can help people reflect and understand the essential role they are playing in evolving computer science into a more gender-diverse field. Another advantage would be encouraging young women to join computing by presenting them with successful female figures, and diluting the computer scientist’s stereotypes. To apprehend the influence of these serotypes, you should read Dr. Cheryan’s research “Ambient Belonging: How Stereotypical Cues Impact Gender Participation in Computer Science,” published in the Journal of Personality and Social Psychology.1

Events celebrating women in computing can push even further by encouraging more male participation. Creating a social-scientific environment with “an almost” equal gender participation (equality doesn’t necessarily mean diversity) can allow both genders to experience true diversity. Men would experience how it feels not to be the dominant gender, which could create empathy toward female colleagues. Women would get a chance to speak their minds, sensing the support of other women in the group. The resulting discussions could eventually lead both genders to understand how diversity liberates women to take active roles in their teams, and how that in turn encourages new perspectives and contributes to the success of the final product.

1 http://depts.washington.edu/sibl/Publications/Cheryan, Plaut, Handron, Hudson, 2013.pdf

Biography

Nur Al-huda Hamdan is a Ph.D. candidate and research assistant at RWTH Aachen University, Germany. She does HCI research on wearables, interactive textile, and imperfect design.

L-to-R: Keynote speakers from academia and industry shared their work experiences and interacted with the audience; conference participants experienced cutting-edge technology during the career fair; and international teams from various computing disciplines collaborated in the Codess Microsoft and Intel hackathon event.

The number of Internet connected wireless light bulbs and lamps worldwide by 2020.

100M+

Imag

es b

y K

rist

ina

Lida

yova

and

Tom

as O

rave

c

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 220

begin

BLOGS

cars [3], smart homes, sensor networks, and wearables. This list would endlessly go on as more and more objects are embedded with required IoT components: a sensor, a battery, a microprocessor, and a (usually wireless) transmission modem. The next time you are bored, play this game: Find the closest IoT device next to you and spot the components. Your Wi-Fi controlled coffee machine has them all, and yes, it is a real IoT device [4]. Many people have written lovely tales around IoT enabled houses and IoT daily interactions [5]. These stories usually feature ghost-like devices that act upon your environment, such as “smart” thermostats that talk to your smartphone, “smart” doors that open only to you (or to your digital-self, whose identity is an encrypted RFID pocket card), and a “smart” refrigerator that texts you when you are out of soy milk. There is so much “smart” around us, yet most of us are not that excited about Wi-Fi coffee machines or smartphone-controlled thermostats. Where’s this missing “smartness” that would really change our experience? What do these devices do for us, really?

There is a split in IoT: Some devices are hidden as they sample and report the world to us, while others are exposed to our interaction possibilities. Let’s take the example of those 23 million IoT enabled cars [3]. You can find these devices inside a contemporary car in different fashions: voice-activated systems, navigation systems (interfaces to Google Maps, etc.), and the hidden workers, such as mechanical health sensors that communicate and report to your mechanic—you name it.

It seems IoT vendors, tech makers, researchers, and so forth focus solely on sensing and communications. Why is that? Well, that happens because most everyday objects cannot be easily augmented with actuators. While having a sensor that counts how many sips of coffee you had can probably work with a tiny-tiny ATMEGA processor and a low-powered last generation Bluetooth modem running on a coin cell battery, it is unlikely the coffee cup can run away when you have too many sips as a reminder that you drank too much coffee. So why can’t the coffee cup be animated? This vision does not happen because adding a motor to it would not work with such a tiny battery. However, we do think that would be an interesting means of expression, both for the coffee mug and for your own experience. Some researchers have explored the notion of animating real objects at the expense of one-off art pieces like a water faucet that curls up and plays with you [6] or a sofa that changes shape as you sit on it [7], etc. But to bring that to all the devices in your household would mean a lot of batteries, a lot of motors, and a lot of actuators. This

Affordance++: The tale of animating IoT objects By Patrik Jonell and Pedro Lopes

Upon hearing the phrase the “Internet of Things” our mind imagines meshes of entangled devices working around the clock, carefully sampling the environment with their tiny sensors and reporting to us at a distance, in order to satisfy mankind’s voracious and inexplicable appetite for efficien-cy and more data. Many know the Internet of Things (IoT) has become both a buzzword and a trillion dollar mar-ket—1.9 trillion USD to be more precise [1]. Forbes further cites an astonishing 16 billion interconnected devices as of last year’s evaluations [2].

So two questions come to mind: (1) Where are all those “smart” devices?; and (2) why are those devices not enhancing our (human) experience?

This first question is easy. IoT devices are hidden inside objects around us and attached to us: connected

Sensors embedded in sport gadgets can help fans keep close track of their favorite teams.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 21

is why we are stuck with a web of interconnected objects that do nothing else besides sensing and reporting. We’d like to think there is another type of experience to be had, one that is much more personal and physical. One that is beyond Fitbit bracelets, coffee machines that know how many espressos you drank, and other monitoring devices.

PHYSICAL INTERACTION WITH THE IOTOur contribution to the IoT is affordance++, a concept that allows everyday objects to become animated as we interact with them. For example, a drinking mug that doesn’t want to be grasped by its body when it’s too hot, will suggest to be held by the ear; a spray can that always shakes itself before you use it; a door handle that helps you correctly open it, despite the fact you have no clue in which direction the handle rotates in that particular part of the world; or a meeting room that reminds you to knock before entering if a meeting is being held inside.

As a matter of fact, none of these examples are IoT devices with a big battery. In fact, none of them have a single motor. They are as simple as the IoT devices we know. So what makes the spray can shake? Your muscles. Instead of actuating motors we actuate the user’s muscles with electrical muscle stimulation. Electrical muscle stimulation works by sending electrical impulses through electrodes attached to the user’s skin and into the motoric nerves. The impulses act as a control signal that triggers the muscle fiber contraction. This is the same technique used by physicians in muscle rehabilitation. In fact, we even use the same device you find in clinics, except we use a computer to control it and retrofit it in a wearable bracelet. So in short, once you reach out for an object, this object helps you to better manipulate it by telling your muscles how to do so.

In a lab study participants understood the poses that the objects suggested as the best approach on how to use them. This seems to work for the harder case of unfamiliar objects too, like a patented avocado slicer or a magnetic sweeper, where the visual affordance is not sufficient to suggest how to properly operate the object.

CONCLUSIONAffordance++ is a research prototype built to explore far-out concepts, and for that it uses non-practical technolo-gies such as optical motion tracking, which requires lots of calibration. But we crafted it to explore the futuristic notion that the affordance of an object could be extended beyond the current attributes of our relationship with these objects. Furthermore, the notion is not limited to actuat-ing the user with electrical stimulation. This is simply the most direct and mobile way to achieve actuation of hand poses; other methods, such as hand-worn exoskeletons like Dexmo,1 are possible.

Lastly, we would like to re-emphasize the idea that affordance++ moves the actuation to the user instead of to the objects, which is what allows us to minimize the technological effort of animating every single object that we interact with (e.g., hundreds to thousands), and provides a more embodied experience with the IoT.

References

[1] Gartner. Gartner Says the Internet of Things will transform the data center. Press Release. 2014. http://www.gartner.com/newsroom/id/2684616

[2] Press, G. Internet of Things By The Numbers: Market estimates and forecasts. Forbes. Aug. 2, 2014. http://www.forbes.com/sites/gilpress/2014/08/22/internet-of-things-by-the-numbers-market-estimates-and-forecasts/

[3] Naughton, K. Autos morph into iPhones as buyers want Wi-Fi with wheels. Bloomberg Business. Jan. 10, 2014. http://www.bloomberg.com/news/articles/2014-01-10/autos-morph-into-iphones-as-buyers-want-wi-fi-with-wheels

[4] Davies, C. Smarter’s WiFi Coffee Maker adds caffeine to IoT. SlashGear. The Access Point. Blog. Jan. 5, 2015. http://www.slashgear.com/smarters-wifi-coffee-maker-adds-caffeine-to-iot-05361984/

[5] Huang, D. How will your connected coffee machine impact your network? June 23, 2015. http://blog.merunetworks.com/blog/2015/06/how-will-your-connected-coffee-machine-impact-your-network/

[6] Togler, J., Hemmert, F., and Wettach, R. Living interfaces: The thrifty faucet. In Proceedings of the Third International Conference on Tangible and Embedded Interaction. (Cambridge, UK, Feb. 16–18). ACM, New York, 2009.

[7] Grönvall, E., Kinch, S., Petersen, M. G., Rasmussen, M. K. Causing commotion with a shape-changing bench: experiencing shape-changing interfaces in use. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (Toronto, April 26–May 1). ACM, New York, 2014.

Biographies

Patrik Jonell is a software engineer at TriOptima and a co-founder of Jukeboss. Jonell was as a master student in the Technical University of Berlin and at the Hasso Plattner Institut in Potsdam, Germany. He received the ACM CHI 2015 Best Paper award for his work on affordance++. He is especially interested in physical interaction and interacting with technology beyond the screen.

Pedro Lopes is a Ph.D. student of Prof. Patrick Baudisch’s Human Computer Interaction Lab at the Hasso Plattner Institut in Germany. Lopes creates wearable interfaces that read and write directly to the user’s body through our muscles (a.k.a. proprioceptive interaction). He augments humans and their realities by using electrical muscle stimulation to actuate human muscles as interfaces to new virtual worlds. He also enjoys playing improvised music and is the digital content editor of XRDS.

1 http://www.dextarobotics.com/products/dexmo

To watch affordance++ on video scan the QR code, which redirects you to https://www.youtube.com/watch?v=Gz4dphzBb6I}

Connected navigation in air travel has the po-tential to save 2 to 5 percent annually in fuel and CO2 emissions in 2025, for a potential economic impact of $4.2 billion to $5.2 billion per year.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 222

feature

Imag

e by

Che

sky

/ S

hutt

erst

ock.

com

The Internet of Things… of All Things A spendthrift refrigerator, a garrulous cellphone, and a loafing automobile, there’s a new technology in town everyone’s talking about.

By George HurlburtDOI: 10.1145/2845143

to yield around $11.1 trillion worth of world-wide business annually by 2025, representing a major technology fu-eled shift in the global economy.

The IoT model involves sensing, thinking, and acting, usually occur-ring iteratively in that order. The IoT al-ready contains a myriad of sensors, and more are being added every day. Sensor data requires some form of processing, which constitutes the thinking phase of the model. The processed data, then, initiates some type of action.

Humans possess five primary sens-es: vision, sound, taste, smell, and touch. In addition to these, and inde-pendent of them, humans can also

M ark Weiser envisioned the technology in the early 1990s. As it evolved, people gave it different names: ubiquitous computing, pervasive computing, and ambient computing. Today, we call it the Internet of Things (IoT), a term RFID pioneer Kevin Asthon claims to have coined in 1999. Today, the IoT sits atop the 2015

Gartner Hype Curve for emerging technology. And yet, there is no universally accepted definition of what the “Internet of Things,” or a “thing” actually is.

Wikipedia defines the IoT as “the network of physical objects or ‘things’ embedded with electronics, software, sensors and connectivity to enable it to achieve greater value and service by exchanging data with the manufacturer, operator and/or other connected

devices. Each thing is uniquely identifiable through its embedded computing system but is able to in-teroperate within the existing In-ternet infrastructure.” According to Gartner, the IoT is the “network of physical objects that contain embed-ded technology to communicate and sense or interact with their internal states or the external environment.” The European Research Cluster on IoT (IERC) gives a slightly different definition, describing it as “a dynam-ic global network infrastructure with self-configuring capabilities based on standard and interoperable com-munication protocols where physical

and virtual ‘things’ have identities, physical attributes, and virtual per-sonalities, use intelligent interfaces, and are seamlessly integrated into the information network.”

THE MODELDespite the lack of a clear definition, the technology itself, according to its Gartner status, is very well hyped. The few facts that are known about the IoT, perhaps, justify the hype. The “things” exceeded the number of people on the planet in 2008. By 2025, their number is expected to reach 50 billion (see Figure 1). Its economic impact will be equally massive: The IoT is expected

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 23

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 224

feature

ized, detect all manner of physical phenomena: motion, direction, accel-eration, velocity, rotation, tilt, yaw, roll, pitch, pressure, force, load, torque, po-sition, compression, tension, stretch, strain, vibration, presence, proxim-ity, temperature, humidity, moisture, chemical and gas composition, elec-trical and magnetic fields, radiation, frequency, flow, leaks, levels, duration, and time—you name it.

In brief, if a quantity can be mea-sured, there is likely a suitable sensor for it in the IoT. This means the IoT will rapidly become an all-sensing in-strument greatly augmenting, yet far exceeding, the range of human nervous sensation. The sensory reach of the IoT is a key aspect of its potential value.

Thinking. Hunger triggers a signal indicating a need to eat. The sensation is transmitted to the brain, which, in turn, interprets the relative urgency of the situation. As urgency increases, the brain transmits interrupt signals that tend to override any other activi-ties. This form of selective processing involves organizing and managing the myriad simultaneous sensory signals hitting the brain at any point in time.

The human brain is a scale-free net-work of some 100 billion nodes (neu-rons) with some 100 trillion (1014) con-nections. It is roughly 73 cubic inches in size. Currently, some 7.3 billion in-stances of the brain exist, with untold trillions of human-to-human connec-tions [1]. IBM’s Watson has 32 core-pro-cessors with billions of possible wired connections. It has multiple instances and currently occupies a space equal to three large pizza boxes [2]. In 2011, the Internet was a scale-free network of an estimated 13.5 billion indexed nodes and trillions of connections. The Inter-net has only one instance, and it spans the entire globe.

Computation has a long way to go before it rivals the human brain. The secret is in the compactness of the brain, which leads to superior wet-ware computational efficiency borne of connection proximity. This suggests the IoT will be superior to the brain only when its nodes can connect as fast as neural communication Despite Moore’s Law, the human brain will clearly hold the advantage for some time to come. This reality refutes some

detect balance, motion, temperature, humidity, pressure, itch, pain, relative location of body parts, muscle tension, stretching, thirst, hunger, chemicals in the bloodstream, magnetism, and perhaps even time. The IoT can sense everything a human can sense, but also far more.

Sensing. Sensors in the IoT (see Figure 2) can read the entire electro-magnetic spectrum ranging from the slowest, low-energy subsonic waves to the fastest energetic cosmic rays.

Thus, it can sense frequencies that humans cannot: subsonic and ul-trasonic waves, radio waves, micro-waves, X-rays, infrared, ultraviolet, and all other forms of ionizing radia-tion. It can capture audible sounds (hearing) and visible light (sight). Specialized sensors can yield “snap-shots,” or dynamic streams of infor-mation—both audio and video—suit-able for human consumption.

Other sensors in the IoT (see Figure 2), many of which are microminiatur-

Figure 1. The Internet of Things: An explosion of connected possibilities.

EXPLOSIVE GROWTH OF THINGS ON THE INTERNET OVER TIME

50.1B

ITEM

S

42.1B34.8B28.4B22.9B

14.4B11.2B8.7B

500M

1M

18.2B

2020

YEAR

2019201820172016

201420132012

2003

1992

2015

Figure 2. Sensors galore.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 25

large networked system. Intercon-nected smart homes, buildings, cit-ies and infrastructure grids, autono-mous land vehicles and drones, all manner of connected wearables, and integrated health informatics sys-tems are clearly within the realm of networked systems. One issue will obviously be how to place bounds on such immense networks—to pre-serve security and integrity if not for other reasons—while also allowing necessary information to be shared among them as they begin to overlap into interconnected networks of net-works. The IoT, most likely, will not become the singular all-encompass-ing entity many seem to expect, but rather an amalgam of loosely coupled unique instances.

Acting. This brings us to acting. Following our earlier discussion, the meaning of this stage of the model should be fairly evident. If the term “acting,” however, becomes “interact-ing,” many additional factors come into play.

While some firmly believe the IoT is all about autonomy, in fact, that is not the case. Consider the example of a smart home. Ideally, it should main-tain a constant humidity and tempera-

of the high-profile AI doomsday predic-tors. That is not to say, however, the IoT cannot run amok and create disasters on its own. As we shall examine, that possibility does exist.

Perhaps the IoT is not the sophis-ticated computational system many suppose it to be. Indeed, specialized processing is required to handle the output from various devices. Con-sider the devices that take advantage of various frequencies in the electro-magnetic spectrum. These include: Doppler, planar array, monopulse, continuous wave and ballistic radars, light detecting and ranging (LIDAR/LADR) systems, IR and UV sensors, frequency modulated, amplitude modulated and short-wave radio, television, ground proximity system (GPS) satellites, Wi-Fi, mobile devic-es, and X-ray systems. These devices help humans orient themselves in space, communicate with one anoth-er, diagnose diseases, assess various situations, and isolate objects out-side direct sensory perception. While these devices may help humans un-derstand and assess situations, in themselves, they do not possess suffi-ciently profound knowledge to trigger all but the simplest of actions.

This is not to say software is sim-plistic. In fact, when federated, soft-ware forms complex interdependent networks where cause and effect are only indirectly related. This poses a real computational challenge for the IoT. The design and testing of highly interactive sensor-fed software in sys-tems, systems of systems, and, even-tually, networks of networks requires deep knowledge of network dynamics. In such networks, the ever-present po-tential of rigid lockup or a descent into chaos exists, characteristic of complex adaptive systems. Moreover, in the presence of mammoth corporation-halting hacks and ubiquitous mal-ware, there is an immediate urgency for applied systems dynamics and the mathematics they entail.

The level of software interaction, influenced by constant sensor-bom-bardment, also represents a persis-tent big data issue. When the inputs are straightforward sensor readings, the volume and velocity aspects of big data apply. Unfortunately, inputs come

from many sources in addition to sen-sor, such as allied software packages in the federation, alien sensors, and hu-man operators. This brings variety into the mix, and the associated semantics calls for veracity. Thus, it is not sur-prising some forms of dynamic ontol-ogy is becoming necessary to deal with abstractions not only within systems, but to disambiguate the software that drives them.

If the IoT is to eventually fulfill the dream of a fail-safe computational powerhouse, network dynamics will have to come into play. In fact, the IoT, by its very nature, represents a

Even the most autonomous of systems need to be told where to deliver their human or systemic payload.

Figure 3. A comparison of the area, nodes, connections, and instances of the human brain, the Internet, and IBM Watson.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 226

feature

augment human activity and will in-evitably require some level of human intervention. Even the most autono-mous of systems need to be told where to deliver their human or systemic pay-load. When confronted with danger-ous or unforeseen situations, they will require some form of human decision-making, especially in the design and testing phases, but likely throughout the life cycle. Rapidly changing envi-ronmental variables are too plentiful for any other alternative.

CONCLUSIONThe IoT human-machine interface re-quires a multi-disciplinary approach. Not only are computer scientists and software developers needed, but subject matter experts from many fields must also be fully engaged. Nor is it purely up to the neurolo-gists, cognitive psychologists, and anthropomorphic experts to shape the interfaces. Engineers, chem-ists, and physicists must engage in designs best suited to the intended purpose. Most importantly, func-tional subject matter experts such as doctors, architects, city planners, fo-rensics people, and many others—in-cluding skilled crafts people—must contribute their knowledge to assure a given IoT system accomplishes its purpose. Despite its name, the IoT is not the sole prudence of the com-puter scientist. Rather, each instance will require a dedicated cross-disci-pline community working collabora-tively under common visions to make the IoT truly effective.

References

[1] Zimmer, C. 100 trillion connections: New efforts probe and map the brain’s detailed architecture. Scientific American. January, 2011.

[2] Marks, P. Watson in your pocket: Supercomputer gets its own apps. New Scientist 222, 2967 (2014), 17–18

[3] Linert, P. Google driverless test cars to have steering wheels, brakes. Insurance Journal. May, 2015.

Biography

George Hurlburt is chief scientist at STEMCorp, a nonprofit that works to further economic development via adoption of network science and to advance autonomous technologies as useful tools for human use. He is engaged in health informatics and course development. He sits on the editorial board of IT Professional and is a member of the Board of Governors of the Southern Maryland Higher Education Center.

© 2015 ACM 1528-4972/15/12 $15.00.

ture via its servomechanisms. Lights and appliances should turn on and off according to a timetable of events. The lawn would be watered when sensors indicate the need. In reality, however, extreme weather can exceed a system’s thermal limits, servomechanisms can fail, schedules vary widely, and water restrictions might draw large fines. Thus, the homeowner is required to spend time and energy regularly inter-acting with the smart home. Variables outside the realm of the home are con-stantly at work, requiring some level of ongoing system tweaking.

An autonomous drone in human occupied airspace provides an ex-ample on a larger scale. The Next Generation (NEXT GEN) Air Traffic Control system will hardly be autono-mous. While many control actions of the human controllers will be trans-mitted as digital text messages that are readable by onboard systems, hu-mans, nevertheless, remain firmly in control. In the case of autonomous aircraft, the air traffic controller will be in direct contact with the drone’s operator, even if the drone is largely flying semi- autonomously.

The famous Google autonomous vehicle, now declared street-legal for testing in California, must still have

emergency breaking and steering systems to allow human engineers to override the Bayesian Network con-trolling the vehicle in case of a pend-ing emergency during road tests [3]. It is highly likely, as road and weather sensors emerge in grid fashion to better inform vehicles of conditions ahead, human over-rides will also be required in the larger infrastructure.

While perhaps simplistic, these ex-amples illustrate an important point. Sensors may well fuse with program-mable, perhaps even self-organizing software, in the IoT. At the end of the day, however, these devices are built to

The IoT will be superior to the brain only when its nodes can connect as fast as neural communication .

Figure 4. Where the smart IoA stuff will be.

As one of the founding organizations of the Heidelberg Laureate Forum http://www.heidelberg-laureate-forum.org/, ACM invites young computer science and mathematics researchers to meet some of the preeminent scientists in their field. These may be the very pioneering researchers who sparked your passion for research in computer science and/or mathematics.

These laureates include recipients of the ACM A.M. Turing Award, the Abel Prize, the Fields Medal, and the Nevanlinna Prize.

The Heidelberg Laureate Forum is September 18–23, 2016 in Heidelberg, Germany.

This week-long event features presentations, workshops, panel discussions, and social events focusing on scientific inspiration and exchange among laureates and young scientists.

Who can participate?New and recent Ph.Ds, doctoral candidates, other graduate students pursuing research, and undergraduate students with solid research experience and a commitment to computing research

How to apply:Online: https://application.heidelberg-laureate-forum.org/Materials to complete applications are listed on the site.

What is the schedule?Application deadline—February 3, 2016.We reserve the right to close the application website early depending on the volumeSuccessful applicants will be notified by end of March/early April 2016.

More information available on Heidelberg social media

PHOTOS: ©HLFF / B. Kreutzer (top); ©HLFF / C. Flemming (bottom)

Inviting Young Scientists

Meet Great Minds in Computer Science and Mathematics

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 228

feature

Prospects for the Internet of ThingsThe future of the Internet of Things may rely on our ability to tackle issues of safety, security, and privacy, while creating standardized systems that are easy to use and configure.

By Vinton G. Cerf DOI: 10.1145/2845145

To the extent that we rely on soft-ware to operate various devices (such as door locks), one wants assurances that all failure modes are safe. In the event of power off conditions, should all the doors stay locked or open? What if there is a fire? What if the doors have deadbolts that do not open manually (especially from the inside)? What if the garage door opener fails in such a way that the doors are opened instead of closed? What if there is no manual way to operate the garage door if the power goes out or the con-troller battery fails? I am sure you can invent many other similar scenarios that should trigger thoughts about the proper boundaries of program opera-tion and attention to fail-safe design.

One might reasonably ask how the software animating these devices was composed and what consideration

T he phrase “Internet of Things” (IoT) is used to refer to devices that house computing and communication capacity, are capable of being linked to the Internet, and are controlled, or at least monitored, remotely. Another term associated with the concept is cyber-physical systems. In principle, such devices have been emerging for

over a decade, and perhaps as many as two decades, concurrent with the rapid growth of the World Wide Web. Fifteen years ago, I encountered the first digital picture frame from CEIVA. Yet, the first toaster connected to the Internet was implemented in 1990 and shown at the INTEROP exhibition. An idea that had been the subject of humor in the Internet Engineering Task Force community for years: “Someday, toasters will be on the Internet!”

The range of “things” that might be capable of Internet-based interac-tion is very broad. It could include ro-bots; drones; heating, ventilation and air conditioning (HVAC) systems; fire sensors; security alarm systems; au-tomobiles; picture frames; household appliances including lawn mowers, televisions, music players, kitchen equipment and washing machines; among many others. Office equip-ment—such as printers, scanners, and projectors–are increasingly net-worked. Power generation systems and battery-backup systems may also fall into this category. As long as the move-ment away from electro-mechanical control toward computer-based con-trol continues, almost anything can become a part of this growing mass of networked devices. Of course, among the most prominent of devices that

have become part of this universe one finds mobile phones, tablets, laptops, watches, and other wearables, such as Google Glass.

SAFETY IN THE INTERNET OF THINGSBecause many appliances play im-portant roles in daily life, user safety should be a paramount concern for the makers of these devices and for their users. Broadly speaking, one wants to be assured that use of such Internet-enabled devices will not risk harm to person or property. The HVAC system should not allow unlimited heating, lest the system move into unsafe oper-ational territory. I once returned home to find my remotely controlled fire-place operating at full blast, evidently for several days, owing to a battery fail-ure—and this was a purely locally con-trolled system, not Internet controlled.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 29

Imag

e by

Chr

isto

s G

eorg

hiou

nal intent on breaking in. Vulner-ability to remote disabling of security systems or alteration of operational parameters could pose extreme risks.

At one time, a bit tongue-in-cheek, I expressed concern for the headline: “100,000 REFRIGERATORS ATTACK BANK OF AMERICA!” While intended to be slightly absurd, this is not a com-pletely facetious concern. Even if the program governing the operation of a refrigeration unit is simple in its de-sign, reprogramming the computer and communications to allow the de-vice to perform denial of service at-tacks, send spam, or propagate worms, viruses, and Trojan horses could have disastrous consequences. The com-puters used to control appliances may be quite powerful, even if their power is not needed for the specific applica-tion. Physical access to the devices’

was given to assure that errors have been minimized? It has even been suggested that a “cyber-underwriter’s laboratory” be created for precisely the purpose of evaluating the safety of devices that include software and/or communications as part of their op-erational profile.

SECURITYTo the extent that these devices are reachable through the Internet or lo-cally by way of Wi-Fi, Bluetooth, or pro-prietary radio links, attention is also drawn to security. Security is a very gen-eral term. In this brief article, security is taken to mean the device is protected against unauthorized access, control, and export of operational information, including sensor data. This includes protection against unauthorized modi-fication of the software or firmware.

The question is made somewhat com-plex if an attacker has physical access to the device, requiring some form of tamper resistance or tamper detection, and the ability to trigger warning mes-sages or to exhibit indelible indications of improper access.

The risk factors are varied depend-ing on the devices, their function, their computing and communication capacity, and their physical and com-munications accessibility. Audible alarms for security systems that might be heard by neighbors (the car, the house, etc.) may be appropriate for some cases. Physical tamper resis-tance and detection (locks, seals, or enclosures) may be very appropriate in some cases. Access to information such as video feeds, or even simply the temperature of every room in a house, could be extremely useful to a crimi-

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 230

feature

One of the challenges with the IoT comes with ephemeral users—house-guests, visitors to a business, and at-tendees at casual gatherings. It has become common for home routers to offer different access to the Internet for family members, houseguests, and casual visitors. One does not wish for a visit that creates long-term and poten-tially unauthorized access to the Inter-net-enabled things found in homes, hotels, offices, and other institutions. At the same time, this process of pro-viding ephemeral access needs to be reasonably simple. It used to be the case that an IT expert was needed to register a visitor on an office network, but this seems less needed now that captive port and distinct guest net-work Wi-Fi access can be provisioned.

PRIVACYPrivacy and security are linked, but the focus here is on confidentiality. The information accumulated by an Inter-net-enabled thing may have significant privacy aspects. Video cameras should only be accessible to authorized par-ties. The same can be said for informa-tion, such as temperature sensor data. The data, if accessible to unauthorized parties, could be used to determine home occupancy patterns.

However, this is an area of some tension. Suppose a robbery is in prog-ress, one might want the police to have access to the cameras. If a fire is in progress, access to information about room temperatures and maybe even occupancy could make the dif-

Security is taken to mean the device is protected against unauthorized access, control, and export of operational information, including sensor data.

ference between life and death. A similar argument could be made con-cerning medical instrumentation. You may not want the world to know about your diabetic status or heart-beat, but doctors may need to moni-tor or be alerted in case something medically risky is transpiring. Medi-cal information might be vital to sav-ing a life if the victim is in an unfamil-iar town in a new country where his or her medical records are inaccessible. But the patient would not want this incident to give hospital personnel permanent access to these records. One needs the ability to grant tempo-rary access that cannot be extended without proper vetting.

Imagine voice monitors in every room in the house in case fire respond-ers need to determine whether anyone is in a room and conscious. But one would not want such monitoring to be accessible except under specific condi-tions. The ability to control access to device management and information seems vital to preserve privacy.

INTEROPERABILITYIn the United States, the Smart Grid Interoperability Panel (SGIP) was cre-ated in 2009 with the aim of develop-ing a catalog of standards so the evo-lution of IoT devices might achieve some degree of interoperability. This ranges from methods of control, au-thentication, software update, con-figuration, and access to information generated by the devices. In principle, interoperation across smart devices would benefit their users; allow for competition; potentially harmonize safety, security, and privacy require-ments and responses; and allow for comparisons across products with re-gard to their features.

Interoperability is not necessarily a natural goal of many manufacturers. Indeed, the opposite may be true. For example:

˲ “You must buy OUR ink cartridges to use with OUR printers or you will risk voiding our warranty.”

˲ “We must be first to market with our new product—interoperability can go hang.”

˲ “If everything is standardized, the criminals will have an easier time taking over everything.”

programming interfaces might permit such an intrusion.

Just as in the safety question, con-cerns about the software of an Inter-net-enabled appliance arise when we consider security: Is the software vul-nerable to attack via the network? Can it be hacked using locally accessible, physical paths? Does the device know how to authenticate local or remote us-ers so as to resist unauthorized control or release of content? These concerns arise in the case of entertainment equipment that may be displaying con-tent still under copyright, for example. One wants to be able to update these network-enabled devices, but needs as-surance that the new software comes from a legitimate source. There are tensions between automatic updates and updates requiring manual inter-vention to provide some kind of hu-man barrier to the installation of mal-ware. Most users would not likely be in a position to judge whether an update comes from a legitimate source.

A solution to this problem likely lies along the path of digitally signed software and verifiable certificates showing the software comes from a recognized source. One hopes the source of the software will not, itself, have been compromised into sending malware to the device(s) in question. A similar line of reasoning goes along with control and access to the content or data held in the device. One wants the device to resist releasing any data or accepting any control from any source that cannot be strongly au-thenticated using public-key crypto-graphic methods as well as conven-tional, symmetric-key cryptography for confidentiality.

It is also apparent local authenti-cation may be vital to allow users to control the equipment locally. Double-factor authentication might be helpful for local and remote cases. Users might be required to show conventional user names and password(s), as well as sup-ply a one-time password generated by a physical device, such as a mobile or one-time password generator with physical (e.g., USB) interfaces or near field connection capacity. A mobile application might be suitable for this purpose, but raises questions about the mobile’s security features as well.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 31

curity, safety, privacy, and the like. It is not entirely clear how these incen-tives may be achieved. There might be obvious marketing advantages in being able to credibly claim to have achieved these objectives. The idea of certifying authorities might have an important role. With regard to safe-ty assurances, it is entirely possible legislated penalties for negligence might provide additional incentives.

Discussions of “cyber-insurance” have been cropping up of late, espe-cially with regard to privacy violations resulting from system penetrations. But insurance does not, in and of itself, produce safety, security, privacy, or ease of use. There is little doubt, how-ever, that incentives will be important to achieve the desired results.

This short offering has only touched lightly upon the breadth and depth of the challenges put before users, prod-uct designers, standards makers, legislators, operators, and software programmers by the IoT—however it is defined. We are likely going to have to live through turbulent times before this dynamic new Internet space can be tamed, if ever. We cannot neglect the challenge, lest, by accident, we cre-ate a fragile, brittle, and unsafe future.

Biography

Vint Cerf served as chairman of the board of the Internet Corporation for Assigned Names and Numbers (ICANN) from 2000-2007 and has been a visiting scientist at the Jet Propulsion Laboratory since 1998. Cerf served as founding president of the Internet Society (ISOC) from 1992-1995. Cerf is a Fellow of the IEEE, ACM, and American Association for the Advancement of Science, the American Academy of Arts and Sciences, the International Engineering Consortium, the Computer History Museum, the British Computer Society, the Worshipful Company of Information Technologists, and a member of the National Academy of Engineering. He currently serves as past president of the Association for Computing Machinery, chairman of the American Registry for Internet Numbers (ARIN), and completed a term as chairman of the Visiting Committee on Advanced Technology for the U.S. National Institute of Standards and Technology. President Obama appointed him to the National Science Board in 2012.

Cerf is a recipient of numerous awards and commendations in connection with his work on the Internet, including the U.S. Presidential Medal of Freedom, U.S. National Medal of Technology, the Queen Elizabeth Prize for Engineering, the Prince of Asturias Award, the Tunisian National Medal of Science, the Japan Prize, the Charles Stark Draper award, the ACM Turing Award, Officer of the Legion d’Honneur, and 25 honorary degrees. In December 1994, People magazine identified Cerf as one of that year’s “25 Most Intriguing People.”

His personal interests include fine wine, gourmet cooking and science fiction. Cerf and his wife, Sigrid, were married in 1966 and have two sons, David and Bennett.

© 2015 Copyright held by Owner(s)/Author(s). 1528-4972/15/12 $15.00.

˲ “Privacy by obscurity is a good idea—let’s not share any of our opera-tional interfaces.”

I am sure readers can invent their own slogans. From the user’s perspec-tive, familiarity with user interfaces, commonality of data formats, and the potential for third parties to offer ag-gregation services—such as control or data analysis applications—suggest interoperation is attractive. More-over, if every device maker requires a branded controller, customers may end up with a closet full of specialized control devices for each class of object to be managed.

STANDARDIZATIONStandardization is the handmaiden of interoperability. One does not have the latter without the former. It is some-times argued standardization inter-feres with innovation, and to some degree this is a supportable argument. But once certain ways of doing things are found to be suitable for an array of appliances, it is advantageous to identify standards that will promote interoperability and the processing of raw data obtained from Internet-en-abled devices. Thus, standardization can promote competition and innova-tion and should not be discarded as an important tool in the 21st century.

Standards will also assist in the evaluation and comparison of device performance, safety, security, and privacy. Creating new products using standards will also promote their in-troduction into the ecosystems since the standards support interoperability and possible control and monitoring from standard platforms.

Standards should not necessarily be thought to be wholly static. Although there is some utility in backwards com-patibility, except where such backward compatibility creates a cost, perfor-mance, or security barrier.

It seems likely IPv6 will be a popu-lar choice for low-level addressing of devices in a local network and pos-sibly through the Internet. One does not, however, want attackers to shot-gun packets across the Internet in an attempt to disrupt operation or to invade the security of the user’s con-stellation of equipment. This is where strong authentication and standard-

ization of access control may prove valuable. While different methods of authentication may seem attractive as a way to avoid the “eggs in one basket” scenario, it may also be the case that standardizing on methods and imple-mentation will allow them to be more strongly evaluated for security.

CONFIGURATIONThe nightmare scenario for the IoT might be moving from a home with hundreds of such devices to a new home that has many hundreds of pre-existing devices. The task of pro-gramming them all to be operated on a common network (or not!) could be daunting. Tools are needed that would help users to configure large numbers of devices in a brief time. However, it is vital the user not ac-cidently (or deliberately) program devices that are not in the user’s net-work or allow his or her devices to be brought under the control of unau-thorized users. One still wants a way to download new software, but with the restriction that it comes from rec-ognized sources. One does not want to spend either an afternoon or days typing in IPv6 addresses, installing cryptographic certificates, and other-wise manually configuring hundreds of devices. Tools will be needed to op-erate at scale. Ordinary users should be able to configure newly acquired devices into existing systems without having to call in experts or put their entire network at risk while adding or removing devices. Ephemeral us-ers should be easy to configure and to deactivate without compromise of security. Various levels of access should be configurable depending on need. Third parties might need to be authorized to see data for analytical purposes, to control for security or authorized management purposes. None of this is trivial to achieve and will required thoughtful design, stan-dardization, attention to security, and ease of use.

INCENTIVES AND CHALLENGESFinally, it seems important to draw attention to the fact that achieving all of these objectives will require a set of incentives that drive product developers and operators towards se-

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 232

feature

The Genie in the MachinesThe ultimate goal of the Internet of Things and wearable revolution is to gift every person with their own magic genie, who will understand all of their needs and desires and thereby enrich the world around them.

By Jonathan CarasDOI: 10.1145/2845149

it. Over time, enhancements like speed dial, voice mail, and caller ID made the experience of calling more individual-ized. Products, devices, and services became more automated and custom-izable until phones made the leap into the category we now call “smart” de-vices, which offer an unprecedented level of both interconnection and cus-tomization.

WHAT IS SMART?Smart TV, smartphone, smartwatch, the list goes on, but what makes some-

A t last he clasped his hands in prayer, and in so doing rubbed the ring, which the magician had forgotten to take from him. Immediately an enormous and frightful genie rose out of the earth, saying: ‘What wouldst thou with me? I am the Slave of the Ring, and will obey thee in all things.’ Aladdin fearlessly replied, ‘Deliver me

from this place!’ whereupon the earth opened, and he found himself outside.”—Aladdin and the Magic Lamp1

“

As technology advances, we see more and more ideas that once existed only in the realm of fantasy migrate to sci-ence fiction, and then, ultimately, into our own lives. In Speaker for the Dead, sci-fi author Orson Scott Card describes a wearable device named Jane, which “lives” in the communication network that connects the planets. Using this small earpiece, Ender, the book’s pro-

tagonist, stays one step ahead of the people around him. As more and more devices, products, and services become interconnected, this type of smart as-sistant is becoming a reality.

If we look back at the history of con-nected devices, we see a clear trend toward two concepts: interconnection and customization. The first electronic two-way communication device most of us interacted with was a phone. The phone existed in a fixed location and provided an identical user experience for every person who interacted with

1 Excerpted from The Book of the Thousand Nights and a Night (1885) translated by Sir Richard Francis Burton.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 33

Imag

e by

And

rij B

orys

Ass

ocia

tes

/ S

hutt

erst

ock.

com

similar to an interaction with another human being. By leveraging a combi-nation of intuition and a human-like series of input methods, a high-quality smart device can deliver a custom-tai-lored user experience that feels intelli-gent (and thus “smart”) to the end user.

EVERYTHING IS CONNECTEDModern computer human interaction is becoming less focused on a specific device or product, and more focused on service and integrated products. I expect my tablet, web browser, and

thing “smart”? How clever does my device need to be in order to be called a smart device? Although there is no decisive definition, the title “smart” is generally applied when:

1. A product can operate autono-mously to perform customized opera-tions;

2. A human can connect to it and interact with it; and

3. The device can connect to a net-work and communicate with other technologies and devices.

What separates high-quality smart

devices from often gimmicky, low-quality smart devices? A high-quality smart device “knows you.” It knows what it needs to know about you with-out you telling it, and it changes itself to cater to your specific needs. Google Now is a great example of a smart de-vice that “knows me.” When I pack up my workstation at my office, I can glance over at my phone and the cur-rent traffic report for my commute is already displayed. A high-quality smart device will allow you to interact with it in a native and comfortable manner,

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 234

feature

customer service…”—will detect the use of profanity and route a caller to a live customer support if they conclude the caller is frustrated. It is exciting to imagine this type of intuitive comput-er comprehension and custom inter-action being incorporated into many of the human-computer interaction points throughout our day.

Facial expressions are another pow-erful and insightful tool, which can be utilized by software developers, to en-hance the customization of interaction to a level that has not yet been incorpo-rated in today’s products. Machines will use this powerful communication channel to help them optimize when and how they interact with humans. Your watch may detect you are in a bad mood based on the input collected by your Nest home surveillance camera or webcam, and determine it is better to wait another hour, until after you eat, before reminding you the trash needs to be taken out.

EXTREME PERSONALIZATION WILL BE EVERYWHEREImagine a drive-through menu that knows what you ate for lunch every day last week, knows if you went to the gym today like you promised yourself you would, and knows the feedback of your IoT implanted upper gastroin-testinal endoscopy sensor. With all of this information, the drive-through restaurant will provide meal options custom tailored to your exact desires and needs.

The greatest challenge at this level of integration will be cooperation from both the end user and the different de-vices. Cooperation from the end user will be needed to not only adapt new behaviors, but also, more importantly, to give up some amount of privacy in exchange for added comfort. Different devices will need to share the infor-mation they collect with one another. Over time, standards will develop to encourage interplay between devices, products, and services, but the chal-lenge of human comfort regarding pri-vacy relies on how this level of “smart” integration is portrayed by society.

If we view the services as Big Broth-er, collecting data on us for large cor-porations to exploit and profit from, it is unlikely a completely connected IoT

smart TV to learn from my behavior, adapt, and display a custom experience for me. When this connected approach is the norm, the extent to which our de-vices can learn about us, and thus adapt to our needs, will enable us to achieve a more natural and human user experi-ence than what is currently possible.

The idea of enhancing user efficiency and satisfaction via a cus-tom-tailored user experience is not exclusive to the communication in-dustry. Google does an excellent job at optimizing search results and YouTube video recommendations. If I play a video from a link I click on Twit-ter using my smartphone while sitting in my office, this action will influence the recommended videos on my smart TV in my living room at home. (Our home Mac mini even has different user accounts for my wife and me.) Person-alized layouts, backgrounds, folders, and “recents” all increase user efficien-cy and allow for increased personal investment in the product. By looking at the current products and services on the market, we can create innovative improvements and find better ways to customize the user experience to the individual.

HUMAN-COMPUTER COMMUNICATIONHow do we build products that have improved personalization and cus-tomization for the specific end user? Conceptually, humans interact with computers the same way computers interact with each other, through dif-ferent input and output methods. But, when we create products or services designed for interaction with people, we are limited by the input and output methods of a human being. Aligning human and computer input and out-put is critical in the advancement of technology.

A human can receive input through sight, sound, touch, temperature, tex-ture, pressure, smell, and taste. Smell and taste have yet to be used to any serious extent as an input method in human-computer interaction, but one could imagine some fun, or at least interesting, possibilities in these ar-eas. The vast majority of computer-to-human communication in today’s devices is through visual and audio

means. In third place, we have haptic feedback, which is delivered primarily through vibrations in portable devices.

Some of the most exciting potential advancements in our time lie in the possible output methods for humans. A human communicating to a com-puter can create output in the form of visual movement, touch, or speech. Touch, such as pushing a button, is currently the most prevalent method, with speech slowly catching up. Speech is a powerful and alluring output meth-od because it is the primary means humans use to communicate with each other. The more similar human-to-human and human-to-computer interaction can be, the more natural the interaction will feel for the end user, and the more powerful the poten-tial for innovation thus becomes.

Another potentially powerful, but currently underutilized, method of human-computer interaction, is visual communication. Humans, by nature, are visual communicators. Even be-fore we can speak, humans have the ability to convey a great amount of in-formation through facial expressions and body movement, which are then interpreted by the eyes. If you place two people in a room together who do not speak a word of the same language, they will quickly begin to communi-cate via visual cues: hand motions, fa-cial expressions, pointing, etc. Various research studies have found anywhere between 60 and 90 percent of human communications is non-verbal.

MACHINES THAT UNDERSTAND OUR EMOTIONSI have been told modern PBX sys-tems—those automated robots you get when you call a company “press 3 for

A high-quality smart device can deliver a custom-tailored user experience that feels intelligent (and thus “smart”) to the end user.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 35

me or force me to perform a behavior strikingly different than the behaviors I am accustomed to. A case study that drives home this point is Google Glass versus Android Wear. People have been wearing wristwatches for more than 200 years. The spring gear watch was essentially the first “wearable technol-ogy” that ever existed. We are familiar with the concept of a highly advanced piece of machinery that tells critical, time-sensitive information attached to our wrists. Google Glass, on the other hand, presented a new model for wear-able technology. It felt foreign to both the wearer and those around them. The idea of a constant camera point-ed at the rest of the room made the subjects in the ever-potential picture uncomfortable, thereby making the wearer feel uncomfortable and alien-ated. The idea of an invisible computer that projects a user interface is a lofty and ambitious vision, but the imple-mentation was too far removed from the wearer’s current habits to keep the Glass on their face day after day.

Where are we heading with the current advancements in IoT and wearables? Most attempts to predict the technological and cultural revolu-tions of the next 10–20 years have left the reader underwhelmed and under- delivered on its promises. The jetpacks and capsule meal supplement prom-ises were met with faster trains and whole food delivery startups. We have a track record of solving common prob-lems, but not in the fantastic ways that futurists predict. What we can predict with confidence in the near future is a relaxation on personal privacy, an increased investment in voice and emotional interaction with machines, and a growing aspect of the genie that caters to our wishes and needs.

Reference

[1] Biever, C. Home, sweet robotic home. New Scientist 2961 (March 22, 2014). https://www.newscientist.com/article/mg22129615-300-roomba-creator-robot-doubles-need-more-charisma/

Biography

Jonathan Caras is chief technology officer for Glide Talk LTD. His knowledge of technology, user experience and management allows him to successfully lead the development team, incorporating the needs of product, marketing, and business development.

© 2015 Copyright held by Owner(s)/Author(s). Publication rights licensed to ACM. 1528-

4972/15/12 $15.00.

experience can be achieved easily in the next few years. Alternatively, if one views this integration and the result-ing—more customized and better opti-mized—experience as a personal genie in the machines who exists and knows you, the user will certainly embrace the benefits of this heightened level of con-nected world.

BALANCING PRIVACY AND PERSONIFICATIONHow do we change the way people view the collection of personal data? How do we transform the watchful eye from being an invasive menace to a helpful and eventually essential tool? Through personification. As computers, prod-ucts, and services become more hu-man—interacting with us in a more human form—users will connect on a more personal level with the software or device.

Roomba, a robot vacuum, has suc-ceeded on the personalization front. According to the company’s founder, more than 80 percent of Roomba own-ers name their robot [1]. When owners speak to customer support and are given the option to send in their vacu-um for a replacement, owners become uncomfortable with the idea that their robot will be replaced. (“No, you can’t take Rosie away!”)

MAKING SOFTWARE “LIKEABLE”The idea a computer should evoke feel-ings of compassion, humor, personal-ity, and other emotional elements has helped in the success of many prod-ucts on the market today. Siri has an ever-growing number of quick-witted responses to questions—“Siri, what is your favorite animal?” “A tauntaun, but only when I’m cold.” The virtual assis-tant also has customization points—“How do I look?” “Jonathan, you look great today.” By giving Siri a personal-ity and having Siri use my name, I con-nect with Siri on an emotional level and feel more comfortable using my phone via the natural communication form of speech. In turn, this makes me more comfortable with Siri know-ing personal things about me, similar to how a personal assistant would be expected to know how I like my coffee or what flowers and perfume my wife prefers.

My Roomba lets me know when it is stuck by playing sad, mournful noises and announcing, “Please move Roomba to a new location.” When it successfully completes vacuuming the carpet, my Roomba sings a happy little jingle to let me know it has com-pleted its mission and is satisfied with its work. This personality helps me as a human connect emotionally to the device and project empathy for the ro-bot’s sorrows or accomplishments.

The best type of personal assistant is one who can be zealous with tasks. Anyone can ask an assistant to order flowers on your anniversary. A good assistant comes to you and shows you the options for the flowers they already picked out based on your previously noted preferences.

The first case was one of proactiv-ity from the employer. The latter, and more valuable, case required only passive interaction on the part of the employer. Passive interaction is an es-sential element in high-quality smart devices. As computers are integrated into more aspects of our life, there will be more opportunities to integrate passive interaction.

Let us return to the idea of having our own magic genie device and ser-vice. How does this passive listening device help us improve the mundane nuances of our lives? Imagine a fake tooth implanted to measure the pH and chemical content of the food we eat and perform general health scans. This tooth would send data to my “magic genie,” which would remind me I have not been keeping to my diet recently. When I am driving home from work, it will recommend stop-ping at the Italian restaurant where I enjoyed the 500 calorie grilled salmon salad with balsamic vinegar. It will also recommend I purchase the mushroom alfredo fettuccine that my wife men-tioned she loved in a conversation ear-lier in the week. The more information the genie has about our preferences and us, the more it can help us experi-ence a unique, customized reality. This takes YouTube’s recommendation en-gine and brings it to another level.

In addition to all of the genie’s intu-itiveness, passive interaction, human feel, and integration with IoTs around my life, the genie must not embarrass

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 236

feature

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 37

Global Synchronization and the Challenges of Building Network AwarenessThe Internet of Things places new demands on wireless networks that cannot be met with conventional infrastructure, services, and protocols. But there is hope, specifically a new paradigm to enable wireless awareness through global synchronization.

By Alyssa B. Apsel and Enkhbayasgalan GantsogDOI: 10.1145/2845151

vacy. Central nodes must be capable of mapping the entire network at all times and regulating all traffic through a central location, thus allow-ing easy access to this information.

As the number of interconnected devices increases from a thousand-fold to a million-fold, direct device-to-device or peer-to-peer (P2P) communi-cation is desirable for several reasons. P2P communication only relies on localized bandwidth, which reduces burden on the central node. It also respects privacy, as there need not be

O ver the past decade, a mythology has been built around the idea of ubiquitous computing that would allow humans to move through environments, simultaneously interacting with a myriad of devices. This mythology ignores all the complexities of background sensing and computation required to perform this role.

Such technologies “disappear” quickly from the user’s point of view. It has been pointed out seamless networks already exist in the form of wireless networks that enable users to communicate without knowledge of the complex underlying infrastructure. However, wireless networks are, to a large extent, unaware of their environment and do not interact with it [1]. In fact, there are good reasons why, despite academic work and commercial

interest in the Internet of Things (IoT), significant challenges to ubiquitous computing persist on a grand scale.

The most obvious challenge is the limited capacity of existing mobile networks. A massive increase in the number of interconnected devices concurrent to the IoT would strain the existing infrastructure. Even low-bandwidth applications, like the ones currently used in social networking, impose overhead in the form of con-nection requests that can overload networks and pose challenges at the

MAC layer [1]. Such communication through centralized networks for low-bandwidth data seems inherently in-efficient. This is because in addition to overhead required to initiate com-munications, the route for communi-cations itself is circuitous rather than being direct (see Figure 1).

Another significant challenge is privacy. While centralized networks need hubs that are capable of sup-porting the total bandwidth and com-puting requirement of the networks, they also pose a risk to personal pri-Im

age

by A

ndri

j Bor

ys A

ssoc

iate

s /

Shu

tter

stoc

k.co

mfeature

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 238

feature

time must be used wisely in achieving and maintaining this synchrony.

Of course, if synchronization is re-quired for discovery, one cannot just rely on synchronization techniques that already assume neighbors have been discovered. Yet, this is exactly what most wireless P2P networks (like Bluetooth and Wi-Fi Direct) do to achieve synchronization and com-municate. They rely on an asynchro-nous backbone to achieve discovery and then synchrony. This means they must operate with their receivers “on” for long periods of time, consuming ex-cessive power and fundamentally lim-iting scalability. This is also why Wi-Fi Direct networks have been shown to consume up to 25 times more energy to achieve discovery in comparison to alternate synchronized discovery pro-tocols [2]. This also results in shorter

a central map of every device in the network. However, this requires know-ing who is local. This is a challenging problem for a number of reasons. It’s a bit like the chicken and egg situation. How do you communicate if start-ing up communication requires prior knowledge of who is nearby for which some form of communication should already exist? Protocols exist to regu-late communication once discovery is achieved; yet the challenge of realizing power-efficient, scalable device discov-ery remains [2].

Low power P2P wireless networks do exist in practice. However, real scal-ability has not been demonstrated on the scale required for IoT. True P2P net-works must be constructed out of sym-metric radios, and the channel must be symmetric. This means one receiver or transceiver cannot be made low power

with the assumption that another will be operating on a higher power bud-get. As a result, the radios must duty cycle to save power and extend both the reach and lifetime of these nodes. To achieve true low power required for background awareness of thousands of devices over very long lifetimes [2], this duty cycling must be very aggres-sive and should also be at the bit level [3]. Wireless receivers that are always on listening mode are incapable of these low power levels at very low data rates. This is due to circuit overheads and the noise floor of the receiver [3]. Frequent low-duty cycle communica-tion is needed for proximity awareness and discovery of neighbors in an active and changing environment. This re-quires synchronization so each device can share a common time base and communicate. However, power and

Imag

es b

y En

khba

yasg

alan

Gan

tsog

Figure 1. As the number of connected devices increases, routing low bandwidth IoT data through cell towers or the cloud is inefficient for local communication needs.

SALE

SALE

(a) (b)

Figure 2. Nonlinear, monotonically increasing, concave-down state function of the local oscillator of each node allows the synchronization of the nodes.

Time

Stat

e

Coupling Direction

Phase

Stat

e

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 39

discovery ranges and shorter standby times. Clearly, synchronized proto-cols seem to offer a good solution in response to the challenges in power efficient P2P. However, such systems depend heavily on global synchroniza-tion of all the nodes, which enable all devices to follow a common protocol with symbol-level time precision. Us-ing such a system, discovery of proxi-mate nodes can occur very quickly.

The final challenge is how to achieve the global synchrony that underpins these entire systems. GPS provides global synchronization for outdoor networks, but fails for many locations. Some type of P2P ambient synchroni-zation is required to augment this net-work and provide an expanded range of synchrony. We have already dis-cussed the limitations of systems such as Bluetooth and Wi-Fi Direct. Our group proposed an alternative based on biological models of synchrony [4]. Synchrony is observable in a wide va-riety of biological systems. Examples include cardiac neurons pulsing to-gether, birds flocking and forming v-shaped groups in migration, and peo-ple synchronizing to perform a wave in a sports stadium. None of these sys-tems achieve synchrony by using a per-fect clock or observing a single master node. Instead, they achieve synchrony by observing their neighbors and us-ing local information in a distributed network to give rise to a global emer-gent behavior. Perhaps one of the best-known and most studied examples of this is the Southeast Asian male firefly. These fireflies synchronize their flash-es to attract mates, a phenomenon that can be observed both throughout Thailand and in a number of YouTube videos. Mirrollo and Strogatz modeled this behavior as the impulsive cou-pling of relaxation oscillators [5].

The basic function of the system is relatively easy to understand. Each fire-fly oscillator has a state function with certain characteristics (see Figure 2). When the oscillator reaches the end of its cycle it resets, and couples itself to its neighbors. This pulse coupling advances the phase of the neighbor-ing oscillators, bringing them closer to the end of their cycle as well. It has been rigorously shown that in an ideal condition with perfect coupling and no

noise or delay, large systems of these oscillators will always synchronize. Our group demonstrated such a system of relaxation oscillators can even be shown to synchronize under less ideal conditions of realistic wireless systems including noise, delay, and only local coupling [6]. Based on this work, we have built a system of UWB impulse ra-dios that achieves robust synchroniza-tion in a multi-hop network, even when each radio can only communicate with its two nearest neighbors. This system demonstrated ad-hoc P2P communica-tion at low data rates of 150Kbits/sec, but consumed only 100 W per node—orders of magnitude lower than other P2P networks. This was possible by using the emergent nature of the syn-chronization algorithm to achieve syn-chrony well enough to enable bit-level duty cycling in the data transmission. Likewise, a similar radio has been demonstrated for wireless monitor-ing of neurotransmitters via mobile fast-scan cyclic voltammetry (FSCV) measurements in live and untethered rats. Synchronized and aggressively duty cycled radios enabled reduction of system power again to 30mW, dra-matically reducing the size and weight of the system battery and extending the lifetime of the nodes [7]. We are continuing work on these synchronous wireless networks to extend the range of communications to hundreds of me-ters by moving to a duty-cycled narrow-band radio system. Such a system rep-resents a fundamental change in the predominant paradigm of the central-ized wireless network. This system can coexist with GPS or other synchroni-

As the number of interconnected devices increases from a thousand-fold to a million-fold, direct device-to-device or peer-to-peer communication is desirable.

zation techniques, allowing the range of synchronous nodes to be expanded beyond the reach of GPS.

While ubiquitous computing offers an attractive vision of seamless tech-nology, it is fundamentally limited by the infrastructure of today’s wireless systems. In order to achieve seamless IoT, a wireless network capable of prox-imity awareness must emerge. Proxim-ity awareness with hundreds, thou-sands, or millions of devices requires careful use of both power and band-width, which can only be achieved with P2P, synchronized, and duty-cy-cled networks. To this end, rethinking wireless networks in this direction, whether through the biological inspi-ration of oscillator networks or some other means, is likely to revolutionize the way we perform computing.

References

[1] Corson, M. S., Laroia, R., Li, J., Park, V., Richardson, T., and Tsirtsis, G. Toward proximity-aware internetworking. IEEE Wireless Communications 17, 6 (2010), 26-33.

[2] Baccelli, F., Khude, N., Laroia, R., Li, J., Richardson, T., Shakkottai, S., et al. On the design of device-to-device autonomous discovery. 2012 Fourth International Conference on Communication Systems and Networks (COMSNETS) (Bangalore, Jan. 3–7). IEEE, Washington D.C., 2012, 1–9.

[3] Dokania, R. K., Wang, X. Y., Tallur, S. G., and Apsel, A. B. A low power impulse radio design for body-area-networks. IEEE Transactions on Circuits and Systems I: Regular Paper 58, 7 (2011), 1458–1469.

[4] Wang, X. Y., Dokania, R. K., and Apsel, A. B. A crystal-less self-synchronized bit-level duty-cycled IR-UWB transceiver system. IEEE Transactions on Circuits and Systems I: Regular Papers 60, 9 (2013), 2488–2501.

[5] Mirollo, R. E., Strogatz, S. H. Synchronization of pulse-coupled biological oscillators. SIAM Journal on Applied Mathematics 50, 6 (1990), 1645–1662.

[6] Wang, X. Y., Dokania, R. K., and Apsel, A. PCO-based synchronization for cognitive duty-cycled impulse radio sensor networks. IEEE Sensors Journal 11, 3 (2011), 555–564.

[7] Dorta-Quinones, C., Wang, X. Y., Dokania, R. K., Gailey, A., Lindau, M., and Apsel, A. B. A wireless FSCV monitoring IC with analog background subtraction and UWB telemetry. IEEE Transactions Biomedical Circuits Systems (June 2015).

Biographies

Alyssa Apsel is currently a professor of electrical and computer engineering at Cornell University, which she joined in 2002. The focus of her research is on power-aware mixed signal circuits and design for highly scaled CMOS and modern electronic systems. She received her Ph.D. from Johns Hopkins University in 2002 and her B.S. from Swarthmore College in 1995.

Enkhbayasgalan Gantsog is currently a Ph.D. degree candidate at Cornell University. His research interests include analog and mixed-signal circuit design in the context of system level challenges. He graduated from Lehigh University in 2011 with a B.S. in electrical engineering.

© 2015 ACM 1528-4972/15/12 $15.00

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 240

feature

Trends in Internet of Things PlatformsWith billions of IOT devices predicted to appear over the next few years, some things have to change.

By Michael AndersenDOI: 10.1145/2845153

˲ Power efficiency. With devices outnumbering people, we would go mad if we had to run around charging or replacing batteries in all our devic-es. Battery life needs to be measured in months or years, not in hours or days. If we imagine a user is willing to replace one battery a week at maximum, then in order to stay within that limit with 20 devices, each needs to last roughly five months of battery life. With the number of deployed devices predicted to be even larger in commercial build-ings (reaching into the thousands), deployed sensors may need lifetimes of 5-10 years in order to be viable from a maintenance point of view.

˲ Effective communication. IoT devices will likely provide very little functionality by themselves. The real benefits are in autonomous behavior once they are connected in an ensem-ble. They will need a communication

T he Internet of Things (IoT) space is flourishing on several fronts. Companies are rushing to “get their foot in the door” so to speak. Large, well-established companies like Intel, Oracle, and IBM are all eager to find a position for themselves in the evolving market. Every month a new IoT startup, flush with venture capital,

joins the game. We also see increasing interest from students and hobbyists, who form a large part of the development platform market. In this hubbub it can be difficult to track the progress of the field. What problems do we need to solve? How close are we to solving them? While these are complex questions to answer, we can pick out the critical ones fairly easily with some deductive reasoning.

The key feature of the IoT space is a focus on “things”—tangible objects, as opposed to services in the cloud or apps on your mobile phone. Further-more, according to the big players in the field, we are talking about a lot of things: Cisco puts this number at 50 billion by 2020 (more than seven times the number of humans on earth). By necessity, these devices are cheap, simple, and not doing much individu-ally. When connected in an ensemble, however, we start to see the emergent functionality associated with the IoT. Depending on which company’s marketing material you look at, you will see different examples of what this emergent behavior might be. Car manufacturers predict components that monitor themselves and warn you when they need replacement. Home entertainment companies promise seamless interaction between mobile

devices and sound systems. The mak-erspace movement is producing many home automation devices with house-hold appliances coordinating to make our lives easier.

While visions of the IoT are vast and differ from expert to expert, it’s clear the vision depends on some core necessities:

˲ Rapid development. The goal is for IoT devices to be cheap and plen-tiful. If devices are going to be cheap (think $2 each), they need to not only be made of cheap components, but must be easy to develop. The longer it takes a company to design and manu-facture a device, the higher the price needs to be to recoup costs. Further-more, in the same way that the web flourished because the barrier to entry was so low, many IoT devices will need to come from hobbyists and students, not just well-funded companies.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 41

breadboard and Arduino and com-plete a prototype in an afternoon. Li-braries exist for all the peripherals, and there is a vibrant community full of helpful individuals if one were to run into any difficulties. In the past, platforms like Arduino Uno were sim-ple to use because they used simple 8-bit microcontrollers that were easy to configure and use. This was great, but those microcontrollers were sim-ply not capable of much. Modern platforms are now using 32-bit ARM Cortex MCU’s that are pretty diffi-cult to setup and configure, but have far greater capabilities. To maintain a simple development environment, more effort is put into developing a software abstraction layer and collec-tions of libraries providing extensive functionality. The NRF51822 is a great example (see Figure 1). A small self-contained chip with an ARM Cortex

method that scales well when devices significantly outnumber humans. Have you ever had a phone or laptop stay con-nected to the Internet for months or years without a single human interac-tion? Well we are going to need that level of reliability for the same reasons we need long battery life.

I would argue, to really progress in solving these problems we need to address them in the everyday rapid prototyping platforms that you can buy off the shelf. Believe it or not, plat-forms like Arduino, Beaglebone, Mbed, Raspberry Pi, and the myriad of devel-opment kits released by vendors such as Atmel and Microchip are the first stop for most development—whether it be in an academic setting, a startup, or even well established R&D depart-ments. There are several reasons: sepa-ration of concerns (the software team iterates on the firmware with a dev kit,

while the hardware team designs and fabricates the eventual platform); rapid development (you need a prototype to show the investors, and you need it yes-terday); cost (spinning a single printed circuit board costs more than buying a dev kit); and ease of debugging (dev kits are known for good hardware, and generally have debug features). Regardless of the reason, however, it is clear the characteristics of your run-of-the mill development platform have a large impact on the architecture of the firmware that is developed on them. Firmware is changed as little as possi-ble when migrating from the develop-ment platform to production, as every change could introduce new bugs.

So how are the common develop-ment platforms like Arduino and Mbed doing? Well, they are clearly good at rapid development. A sea-soned programmer can whip out their

Figure 1. A Nordic semiconductor NRF51822 offering an ARM Cortex M0 processor and Bluetooth Low Energy radio, along with a soft-device abstracting the complexity of the hardware.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 242

feature

to observe the power characteristics of a device under test. This is a common, but very limiting, method as code must be extensively modified so the state transitions of interest can be observed repeatedly. Furthermore, even expen-sive oscilloscopes can only capture—for a couple of seconds in enough resolution—the short-lived spikes of a microcontroller. This makes it a far cry from a holistic power consumption de-bugging method.

In the rare case that an engineer has a special-purpose current mea-suring device on hand, they are still thwarted by the dreaded dynamic range problem. When designing an energy-efficient program, we care about measuring currents as low as 10uA (especially if you’re running off a coin cell), but when Wi-Fi fires up cur-rents can spike to 200mA. That is four orders of magnitude. Furthermore, if done correctly, that spike ought to last only a handful of milliseconds. Source measurement devices capable of giving you 1uA resolution, while providing up to 200mA, do so by sam-pling in the zero crossings of the 60Hz mains power (to reduce noise), so they can’t sample faster than 120Hz with-out losing accuracy. That means they will frequently miss a 2ms long spike. The bottom line is debugging power is simply much, much harder than debugging code functionality. If it is difficult even for engineers in reason-ably well-equipped labs, it is even more so for hobbyists and students who have less money to spend on equipment.

The only solution then is for the software environments and libraries, which came with off-the-shelf devel-opment kits, to take power efficiency seriously, and abstract it along with all the other abstracted peripherals. In the same way we don’t need to worry about tricky things like initializing flash caches, watchdogs, or phase locked loops (PLLs) when writing an Arduino sketch, the environment could be writ-ten in such a way that the chip is auto-matically in the lowest power configu-ration possible.

It is clear something needs to change on the power efficiency front if the development kits currently on the market are going to actually be suitable for developing IoT products.

M0 and a Bluetooth Low Energy (LE) radio in a single package. By packag-ing a “soft device” along with their software development kit (SDK), the complexities of Bluetooth are mostly abstracted away, leaving the developer with a simple, well-defined API.

If development kits are doing well on rapid development front, how are they doing on power efficiency? Well, that’s an interesting question. Put simply: not so well. The problem is not just the hardware; it’s the software. You see, modern microprocessors like the ARM Cortex found in Mbed and Arduino Zero (along with many other development kits) have the capability to be power efficient, but developers must be quite knowledgeable about the inner workings of the chip in or-der to achieve that in practice. Take for example the Cortex M0+ found on the Arduino Zero (an Atmel SAMD21). If configured carefully, it will sleep at roughly 2.5 uA, which is a 100-year battery life on two AA batteries. When running at full speed, however, it uses approximately 6 mA, which is only 16 days. That is a pretty big difference. For software with low average power consumption, developers need to make sure all the modules in the chip are configured correctly.

The problem is without any librar-ies or software framework for assis-tance, tweaking software to be energy efficient is not straightforward. It can be as difficult as getting it working in the first place, if not harder. How

well do you think you can debug your code if you had no debugger, no printf statements, and no toggling of IO pins? You see, on almost all develop-ment kits today, monitoring power consumption is not a priority. While Arduino and Mbed go to great lengths to provide a friendly USB serial connection for stdio, there is no indi-cation of how much power a board is using. Optimizing power consump-tion is like debugging blind. Even on those development kits that feature a jumper for measuring current, devel-opers need expensive equipment to get a current trace over a long period of time.

Even in relatively well-funded labs, you would be surprised how often an engineer will resort to a current shunt and an oscilloscope (see Figure 2) to try

In the same way that the web flourished because the barrier to entry was so low, many IoT devices will need to come from hobbyists and students, not just well-funded companies.

Figure 2. The go-to power measurement setup in labs across the world: A simple resistive current shunt connected to an oscilloscope.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 43

system power to roughly 10 uA. What we are exploring now is how to design the operating system on the device that allows the software to be well con-nected and energy efficient without significant complexity. New language features, for example, those found in C++14, allow for more natural expres-sion of event based programming con-structs and could help reduce the dif-ficulty of efficient firmware design.

In summary, there are a few key challenges that need to be solved in order to realize the IoT vision. Chief among them are development agility, power efficiency and communication. In order for the IoT to flourish, the barrier to entry needs to be low, and this implies that we will see changes to hardware and software appearing in the bread-and-butter development platforms used by the masses. Hope-fully, we will see better connectivity, better network stacks, and more pow-er efficient operating systems in the near future.

Biography

Michael Andersen is a Ph.D. student at University of California, Berkeley, and part of the Software Defined Buildings group. He works on ultra-low-power embedded systems for the Internet of Things (IoT), wireless sensor networks, and high-performance telemetry databases for handling the volumes of data that IoT applications are expected to produce.

© 2015 ACM 1528-4972/15/12 $15.00

What about communication? Well, that is not doing so well either. Con-sidering most development kits do not come with a radio, if you find one that does, or you find a radio module that can plug into your development kit of choice, you’ll find the support-ing code to be rather lacking.

Remember the IoT is about con-necting billions of devices together. This is not a trivial task, and requires a decent interoperable network stack. How often have you encountered an off-the-shelf development kit with a decent TCP/IPv6 implementation? I would wager never. Even in the re-search community, essential prob-lems like how to connect a sensor mesh network to the IPv6 Internet are not solved. Every group has their own border router solution and most of them require a great deal of baby-sitting to maintain in working order. Even at the link layer, there are unre-solved problems.

Bluetooth LE, or Bluetooth Smart, seems to be the most popular inter-connect type for devices on the mar-ket. It offers low power consumption, cheap components, and easy con-nections to mobile phones. Unfortu-nately, it is also lacking in many ways. Bluetooth is primarily a peripheral in-terconnect, meaning you need an app on your phone to talk to your smart devices. How do you expect to create a large autonomous system of intercon-nected devices if they all require a mo-bile phone with a specific application to be nearby? Can you even imagine installing hundreds of apps in order to interact with smart buildings and the myriad of heterogeneous sensors in them?

Perhaps we can shift to connecting devices directly to the Internet and interacting via services in the cloud. The old machine-to-machine (M2M) setups used in industry have every de-vice equipped with a small cell modem (sometimes even a satellite modem). This is great in that they function rela-tively autonomously… as long as some-one pays the bills. I bet it costs you $10 or more per month to get a data plan on your mobile phone. Are you willing to pay $120 a year to keep a $2 device con-nected to the Internet? Thought not.

What about Wi-Fi? If smart devices

are in a house or building, can they just use existing wireless infrastruc-ture? Well, remember the power effi-ciency problem? Even the latest gen-eration Wi-Fi modules use upwards of 200mA to transmit and more than 50mA to receive. For perspective, that means a 2xAA battery-powered device has a life of two days even if it is doing nothing except listening for commands. Technology such as IEEE 802.15.4 is a lot more promising in terms of energy efficiency, but has yet to see widespread adoption past aca-demia. It also has the disadvantage of not being able to connect to a phone. This may be changing though with products like Nest and LiFX using 802.15.4 for part (but not all) of their communication needs.

To investigate these challenges, a team at UC Berkeley is working on a platform (see Figure 3) that retains all of the advantages of those already on the market (rapid prototyping via hardware expansions and easy-to-use software environments), but also pays a great deal of attention to power con-sumption and connectivity (featuring both Bluetooth LE and IEEE 802.15.4). Features like the USB serial debugging or onboard sensors can be turned off with fine-grained power domain con-trol, allowing a reduction of whole-

Figure 3. The UC Berkeley Firestorm, an Arduino compatible IoT platform featuring a Cortex M4, Cortex M0 and BLE + IEEE 802.15.4.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 244

feature

Querying Flying Robots and Other Things: Ontology-supported stream reasoningA discussion on the role of ontologies and stream reasoning in Internet of Things applications.

By Daniel de LengDOI: 10.1145/2845155

Some of this data originates with hu-mans: In 2013, Twitter reported seeing a spike of 143,199 tweets in a single sec-ond. Other data comes from sensors or programs at potentially high intervals. In both cases, the data becomes in-crementally available in what we call “streams.” Stream reasoning is the reasoning over these streams, bearing in mind that any received data sample is likely to be forgotten shortly after arrival. Stream reasoning is the an-swering of questions or drawing con-clusions over streams, in part through aggregation of raw data into more abstract information using various

Imagine a world in which the Internet has become so ubiquitous that it extends itself even to everyday things, such as our cars, fridges, and milk cartons. How far are we then removed from being able to query things for the location of a delivery van, or as some companies foresee it, flying delivery robots? After all, sensor platforms could themselves

be regarded as complex things. Could I perhaps ask them more complex questions that involve many (complex) things?

Unmanned aerial vehicles (UAVs) have increasingly become a topic of discussion in today’s society, in part because of affordability and ease of use. Many are equipped with cameras for taking pictures or recording video, and commercial UAVs marketed for

the general public can sometimes be controlled through simple means such as tablets. The widespread deployment of UAVs has not come without conse-quence. UAVs have been blamed for privacy infringement, airspace intru-sions, and near misses with commer-cial airliners and military applications. But they also serve as useful tools for artists and farmers, help to monitor the health of levees, and have the im-mense potential to save lives in hard-to-reach disaster areas when things do go wrong. UAVs can assist rescue workers in acquiring information they themselves cannot readily obtain.

They too play a role in the Internet of Things (IoT) as complex entities that are part of a diverse ecology with the potential to provide a wider array of services compared to the more tradi-tional things. In this text we focus on a type of “stream reasoning” involving a wide variety of things at the applica-tion layer.

So what is stream reasoning, and what does it mean in this context? Many have pointed out the amount of raw unfiltered data that is produced per day is so enormous that any at-tempts at storing this data for process-ing is very difficult, if not impossible.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 45

[0,1440](Person(x) ∧ Inside(x, restricted) → ◊[0,15]¬Inside(x,restricted))

Here I stands for “it must always be the case within interval I,” and ◊I stands for “eventually it will be the case within interval I.” It is assumed the intervals are to be measured in minutes. Concretely, this formula re-turns False if there is an instance x when a person is inside of the restrict-ed area, and does not leave the restrict-ed area within 15 minutes; otherwise it returns True.

Such a formula would of course be easy to evaluate if only we had data for the truth values of the individual pred-icates at all time-points. It is more like-ly that we do not have this data in its completeness, nor does this data come from a single source. Thus, this entails some form of stream reasoning using

means of processing. In a way, stream reasoning attempts to make sense of a large volume of incrementally avail-able data as it arrives.

A PERIMETER MONITORING EXAMPLELet us consider a situation where we want to know whether someone is trespassing during a certain time pe-riod. Maybe the area is a construction site deemed unsafe. We are therefore not interested in “catching” a pos-sible trespasser. Rather, we first wish to warn the trespasser so they leave at their own accord. Sending someone over to deal with the situation is a sec-ondary course of action in case this is unsuccessful. Surrounding the re-stricted area is a security fence, which has two gates with sensors detecting whether they are closed or not. We also have a few cameras on-site, and we

have a private security guard in case human intervention is needed. We also have several small UAVs with cameras to monitor the area, and a third-party virtual server provider for computa-tional power at a price. Finally, all of these things are able to interact over the Internet and are aware of each oth-er’s identities as part of a local security network.

Suppose we wish to query this secu-rity network in order to determine for the next 24 hours if and when a tres-passer stays within the restricted area for more than 15 minutes. There are many possible languages to pose such a query in. We focus on temporal logics to leverage their expressivity with re-gards to temporal concepts. In particu-lar, using metric temporal logic (MTL), this query can be posed by evaluating the following formula:

Figure 1: A Yamaha RMAX used by Linköping University.

Imag

e by

Kar

ol K

orw

el

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 246

feature

from outside of the system. Transfor-mation specifications describe how to instantiate a transformation and what parameters to use, and so mul-tiple specifications can describe the same implementation with different parameters. CUs represent instantia-tions of transformations. They have unique names and can have subscrip-tions to existing streams. CUs can be created and destroyed as desired. DyKnow represents an instance of such flexible middleware architec-tures [1]. We can use systems such as DyKnow to evaluate temporal logic formulas in MTL. To do so, a configu-ration specification describing which transformations and CUs to use, and how to connect them, needs to be ex-ecuted. However, writing such a con-figuration specification is tedious and error-prone. It significantly compli-cates the writing of queries, and it is not very scalable. After all, we want to use information resources provided by the things we have access to. An-other complication is that a particular information resource may not always be available.

MANAGING SEMANTICSMisunderstandings happen. As a Dutch person, I am convinced my of-fice is on the ground floor, whereas my American partner claims it is on the first floor, and the university con-siders it to be on the second floor due to the presence of a basement. This works for humans, but can be disas-trous for machines. Do we use metric or Imperial units of measurement? What frame of reference is used for co-ordinates? Misunderstandings with regard to these examples can crash your UAV, literally. As a real-world ex-ample, the Mars Climate Orbiter “mis-hap” was the result of unintentionally using different units of measurement, resulting in the unfortunate loss of an interplanetary probe.

There is a need for things to have some degree of semantic understand-ing. Granted, specifications can help harmonize semantics, but how well does this scale when we add further systems into the mix, some of which were written by other people? Ideally we would like to simply write a query describing the information we are in-

various heterogeneous data resources. Could we reasonably require the UAVs to evaluate this formula? What is rea-sonable depends on whether the UAV has access to the required data re-sources, whether it has the necessary processing capacity, and whether the data resources are reliable enough. This is a multifaceted and interesting problem for which we do not yet have a general solution.

SETTING UP STREAM REASONINGGiven a complex query such as a tem-poral logic formula, we want to have the information that is referenced by the symbols used in such a query. From a streaming perspective, this informa-tion is represented as streams, and so the challenge becomes to generate a stream with the desired information. We call this an “applicable stream,” to indicate it is applicable in relation to the query.

There exist a number of approaches to stream processing. Complex event processing (CEP) considers event streams where every stream sample corresponds to an event. We can then consider complex events, i.e., tempo-ral combinations of (non-)events that are represented as events themselves. An example of CEP would go as fol-lows: First, there is an event that car A is closely behind car B; this is followed by the event that car A is beside car B; fi-nally we detect the event that car A is in front of car B. This could then be seen as the “overtake” complex event.

Data stream management systems (DSMS) take a different approach. Here streams are sequences of values on which we apply various variants of win-dow operations, such as sliding or tum-bling windows. This results in man-ageable tables unto which the usual database aggregation methods can be applied. Take for example a stream of speeds produced by a smartphone, as is not uncommon for runners. These values may individually deviate a lot due to hardware quality, so we could then take the average over a sliding window. Clearly there are many simi-larities between DSMS and CEP. Both CEP and DSMS are applicable for IoT applications.

The focus areas for stream process-ing are very diverse. Some research

focuses on query expressivity and the ability to make use of the absence rather than exclusively the presence of data. Another focus area is perfor-mance; how can we maximize the throughput of these systems? Yet an-other area considers streams of RDF triples in the context of the Semantic Web. Most of these areas consider a single-stream processing engine. We take a different approach that focuses on the problem of integrating many (specialized) stream processing en-gines within the robotics domain. This of course comes at a cost; we concern ourselves less with throughput optimi-zation, and we focus primarily on fast and quantitative sensor data.

These specialized stream process-ing engines effectively offer stream-based services, taking streams of data as input and producing output streams in accordance with their specifica-tions. Middleware architecture can be used to manage this multitude of services. Suppose such a stream rea-soning framework distinguishes be-tween streams, transformations, and “computational units” (CUs). Streams are named and can carry a vector of values in every sample. Every sample also contains the timestamp at which the sample became available, and the timestamp for which the data is valid. These two timestamps can be differ-ent, for example when a stream con-tains predictions about a future time. Transformations are stream-generat-ing functions that take other streams as input. They can, for example, be used to generate a stream of speeds from a stream of coordinates, or serve as data sources by relaying sensor information or importing streams

A lot of information comes in the form of streams. Being able to use this information has the potential to greatly enhance IoT applications.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 47

can be used to evaluate MTL formulas. If the formula evaluates to False, we can regard this as a violation event it-self, which can be sent to the security guard. This way multiple things offer services that are combined based on their semantic annotations.

But even here things can go wrong: CUs may crash or stall, streams may be of low quality or stop altogether, per-haps the wireless signal may go bad, or the server providing CCTV streams may go down. Some processing has to change to accommodate failures. This would not be possible if the subscrip-tions were syntactic, but since they are semantic we can attempt to repair the broken pipeline and clean up any CUs that are now inactive due to the broken inputs. None of this requires human input (but does not exclude it), which makes for a very adaptive system.

CONCLUSIONThe IoT opens up many exciting op-portunities for acquiring information resources and for the sharing of infor-mation between things. A lot of infor-mation comes in the form of streams. Being able to use this information has the potential to greatly enhance IoT applications. This is not limited to static household objects, but can be ex-tended to autonomous robots. Howev-er, one precondition is the information available is understood; the semantics of the streams, transformations, and CUs need to be made clear. We need to understand who provides what infor-mation with which semantics under which constraints at what price with-in which time period and with what quality. Ontologies and Semantic Web technologies can help with this.

When “things” can agree on the se-mantics of streaming data, many inter-esting applications will be possible.

References

[1] de Leng, D. and Heintz, F. Ontology-based introspection in support of stream reasoning. In Proceedings of the 13th Scandinavian Conference on Artificial Intelligence (SCAI). 2015.

Biography

Daniel de Leng is a Ph.D. student at the Department of Computer and Information Science at Linköping University in Sweden. His work focuses on on-demand knowledge acquisition for grounded spatio-temporal stream reasoning through collaboration and the Semantic Web.

© 2015 ACM 1528-4972/15/12 $15.00.

terested in without having to worry about how the information is acquired, unless we explicitly put constraints on that—a concept similar to declarative languages like Prolog or SQL.

DyKnow does this by representing its state in terms of streams, trans-formations, and CUs with the help of an ontology. Ontologies formally de-scribe concepts and the relations be-tween those concepts, and are based on description logics. They can be used as a common vocabulary or as a data model, among other things. We use an ontology to query whenever we need information about the state of a DyKnow instance. Facts are then rep-resented using RDF triples consisting of a subject, a predicate, and an object in accordance with the ontology. We can then also describe properties of individuals (objects), such as specific streams, CUs, or transformations. This allows us to assign properties to transformations, for example to de-scribe the semantics of their inputs and outputs. We call these properties “semantic annotations.” Since trans-formation instances are CUs, and CUs produce streams, these semantics in-directly describe the information con-tained within streams. This approach is similar to OWL-S and its service profiles, or the semantic sensor net-work (SSN) ontology.

One obvious weakness to this ap-proach is someone or something needs to provide these semantic annotations, although the same is true of transfor-mation specifications themselves. An-other is it presumes concepts already exist for annotating transformations with. Usually the desired semantic an-notations of transformations are appli-cation or domain specific. Ultimately it is the expressivity of the annotation language that determines whether queries using that language have the intended effects. This is by no means a trivial problem. But what benefits do we gain, assuming we have a suitable, developer-provided domain-specific annotation language?

Recall in order to set up stream reasoning, we require a configura-tion. Assume we have a description of some kind of desired information in the form of a query using the vocabu-lary provided by the ontology. The se-

mantic annotations make generating a configuration something that can be done automatically by recursively searching for appropriate transfor-mations. This is similar to configura-tion planning, and is based on intro-spective capabilities. The matching of desired semantics with semantic annotations is a procedure we call “semantic matching.” It returns a set of applicable transformation trees, where the leaves take no inputs and the root produces the desired informa-tion. Some of these transformations may use other things. If only a single solution is found, it can immediately be instantiated. If no solutions are found, the query cannot be answered, but some approximations may be pos-sible by relaxing constraints. If mul-tiple solutions are found, there are many ways of choosing between them. You can associate costs with instanti-ating transformations, favoring pre-existing CUs. Alternatively you may desire smaller trees as a heuristic for CPU load. You could also take into ac-count some preferential ordering over the possible information providers, such as preferring on-board sensors or external image processing.

Going back to the example, note both the cameras and the UAVs are able to provide camera data of vari-ous areas. If we have a way to detect and track people, we could determine whether the predicate “inside” holds. If the UAVs do not have the process-ing capacity, one might support us-ing external computation facilities to offload the heavy work. A progressor

The amount of raw unfiltered data that is produced per day is so enormous that any attempts at storing this data for processing is very difficult…

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 248

feature

Toward Computing over Encrypted Data in IoT SystemsThe multitude of IoT devices contributes to the enormous amount of data stored on corporate clouds. Yet the level of computing power has outpaced advances in privacy protection. Could encrypted search preserve the privacy of data, while utilizing the computing power of the cloud?

By Hossein ShafaghDOI: 10.1145/2845157

porate service provider. In other words, we should avoid unauthorized sharing of data, as well as disclosure of sensi-tive information, which can be learned from our data. Incidents from the past (e.g., online disclosure of sexual activity of health tracking users) show that user awareness and corporate inclination to address these issues are still missing.

An intuitive and simple approach to preserve the privacy of stored data is to only send and store encrypted data to the cloud, as already practiced by zero knowledge cloud storage vendors (e.g., Tresorit). This technique relies on efficient symmetric key encryption schemes, such as Advanced Encryption

Internet of Things (IoT) application scenarios are finding their ways into various aspects of our lives: health- and activity monitoring, home-automation, elderly care, and connected cars, to name a few. Ultimately, IoT applications aim to create unique forms of interaction with our environment and provide novel insights based on rich data collected

from the physical world. A major concern hereby is how to preserve the privacy of user data and still be able to provide insightful information. Methods based on encrypted data computing appear to be promising approaches. However, they bring new challenges with respect to functionality and computing resources, which must first be solved before they can be integrated into IoT systems. In this article, we discuss encrypted data computing

approaches, their use-cases, and the challenges yet to be addressed, especial-ly with regard to the IoT.

INTRODUCTIONIoT applications contribute to our increasingly cloud-centric and data-driven world by digitizing elements of the physical world that were previously non-digital. The potential of such in-creasingly larger amounts of data is yet to be fully explored. One important con-sequence of this increase in data collec-tion is the growing intrusion into sen-sitive user data. These intrusions stem not only from security attacks, but also from normal, more innocent data col-

lection mechanisms. Sophisticated ma-chine learning tools and inter-correla-tion techniques allow for the inference of private information even from seem-ingly innocuous data. For instance, con-sider common health monitoring ap-plications such as FitBit. The data FitBit collects—such as heart rate, location, step-counts, and quality of sleep—can be used to infer private information such as chronic diseases, social inter-actions, and mental health conditions, to name only a few examples. From the user’s point of view, inferring such information might be of value. How-ever, the user might not be interested in sharing such information with a cor- Im

age

by A

ndri

j Bor

ys A

ssoc

iate

s /

Shu

tter

stoc

k.co

m

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 49

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 250

feature

In the following, we first elaborate on challenges specific to IoT systems, con-tinue with an overview of encrypted que-ry processing techniques, and conclude with open research questions.

CHALLENGES OF IOT SYSTEMSIoT devices are inherently resource limited with regard to energy. Energy constraints mandate the use of low-power components, thus limiting the computational capabilities, transmis-sion range, and communication band-width of IoT devices. Furthermore, new emerging application scenarios differ from the currently dominant web or mobile application traffic in the follow-ing aspects.

Connectivity. Wireless, embedded devices can be connected to the Internet by means of a gateway. This connection, however, can either be long-term (as is the case in home access points) or oppor-tunistic (as is the case with smartphones for wearables). The connectivity charac-teristics impact the protocol design in order to guarantee tolerable communi-cation delays.

Machine-to-machine. Even if the user is interested in the data, he or she might not necessarily be involved in the com-munication loop of IoT devices. Devices collect data, which they store in the cloud; other devices can then be triggered by pre-defined events. This means there is no human being involved who could be intercepted while entering credentials, such as username or passwords.

Content. The communicated infor-mation in IoT scenarios is composed primarily of sensor readings and meta-data. By design, IoT devices can only store data for a limited amount of time. Hence, the need for cloud storage in the IoT is unavoidable, not only to store data,

Standard (AES) block ciphers. This way the cloud is unable to learn anything about the stored data. However, the downside of this approach is twofold: (i) the computing power of the cloud can-not be utilized to process the data; and (ii) in order to analyze the data, the user needs to first download, decrypt, and finally process the data. This requires high bandwidth and computing power on the user-side, and it results in unde-sired application delays. Moreover, us-ers lose the power to search their data, which is in fact the most important functionality considering the ever-in-creasing amount of data.

Encrypted search schemes, how-ever, allow an untrusted party to per-form search operations over encrypted data without the need for decryption. Consider the case of structured data stored in database systems, the under-lying search operations are performed through queries. These queries are structured based on simple mathemati-cal operations, such as addition, com-parison, and equality checks. Hence, an encrypted query processing, in its es-sence, should be capable of performing such mathematical operations in the ci-phertext domain.

Previous XRDS articles have shed light on encrypted search and fully ho-momorphic cryptosystems [1, 2]. These approaches lay the foundation for en-crypted query processing. Here we focus on practical solutions for interacting with encrypted data stored in databases. Practical solutions have to strike a good balance between security, efficiency, and functionality. Moreover, the engineering efforts to build such systems provide invaluable insights into how to further optimize cryptographic schemes, which may remain undiscovered in the theory.

Figure 1. The encrypted search pipeline.

IoT Devices

Cloud Database

User

Gateway

ACM Conference Proceedings

Now Available viaPrint-on-Demand!

Did you know that you can now order many popular

ACM conference proceedings via print-on-demand?

Institutions, libraries and individuals can choose from more than 100 titles on a continually updated list through Amazon, Barnes & Noble, Baker & Taylor, Ingram and NACSCORP: CHI, KDD, Multimedia, SIGIR, SIGCOMM, SIGCSE, SIGMOD/PODS, and many more.

For available titles and ordering info, visit: librarians.acm.org/pod

ACM Conference Proceedings

Now Available via Print-on-Demand!

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 51

but also to make this data accessible to other devices and services.

Resource asymmetry. The compo-nents involved in an IoT system—such as the IoT mote, the gateway, and the back end—exhibit a strong asymmetry in available resources. For instance the clock frequency reflecting computing power spans from a few MHz for the IoT device (e.g., 32 MHz ARM Cortex M3 mi-crocontrollers), to a few GHz at the gate-way (e.g., 1.4 GHz dual-core iPhones), and to potentially several GHz at the back end. The same asymmetry applies for both memory and bandwidth.

Energy for IoT devices is provided by batteries or is harvested, whereas a gate-way might be connected to the mains (access point) or equipped with a signifi-cantly stronger battery (smartphone). The back end, on the other hand, is con-sidered to have continuous access to sufficient energy. Consequently, the en-ergy issue is one of the most important design factors for IoT systems. When designing secure systems for the IoT, one has to bear in mind these particular properties and limitations.

ENCRYPTED QUERY PROCESSINGOne common way to store IoT data is in structured databases, such as SQL data-bases. In an encrypted query processing system, a plaintext SQL query is trans-formed to an encrypted query, such that the cloud cannot learn about the values in the query. The query is then executed over encrypted data and the encrypted result is sent back to the user. For exam-ple, consider a health data system. A sim-ple query to get all entries with heart rate larger than 100, for instance, is trans-lated into an encrypted query, where 100 is replaced with its order-preserving en-cryption (described later in this section) and the heart rate with the correspond-ing encrypted column name:

SELECT * FROM tab WHERE heart rate > 100 SELECT * FROM tab WHERE column-043 > 0x19

The fact that database queries are based on mathematical operations has inspired researchers to build such en-crypted database (EDB) systems, which can operate over encrypted data. Crypt-DB is one of the early systems, which

employed property-preserving encryp-tion schemes and partial homomor-phic encryption to build an efficient EDB [3]. To this end, such EDBs rely on the insight that to support common SQL-like queries, it is necessary to be capable of performing equality checks and have knowledge about the order of encrypted values. However, enabling computation over encrypted data also means leaking information. That is, any order-preserving encryption scheme will, by definition, reveal order relations. The encryption schemes are selected per column and account for the intended query type (e.g., min, order by, etc.). Hence, data items that are not involved in the processing of queries should be encrypted with the strongest cryptographic scheme (i.e., probabilis-tic encryption). For the encrypted query processing, the following encryption schemes can be utilized.

Random (RND). Probabilistic or random encryption is the strongest se-curity scheme, allowing no operation over encrypted data. This scheme is the conventional scheme, widely used in se-cure communication and storage. It has the property that the encryption of the same plaintext m results in two different ciphers c1 and c2 such that c1 and c2 are by no means related (i.e., semantically secure under chosen plaintext attack or CPA). AES in cipher block chaining (CBC) mode has such properties, and efficient hardware implementations of it are already integrated in most IoT devices. This allows the computation of AES-CBC on a typical IoT device in a few microseconds.

Homomorphic encryption (HOM).

IoT applications contribute to our increasingly cloud-centric and data-driven world by digitizing elements of the physical world that were previously non-digital.

Research on fully homomorphic crypto-systems has made significant advance-ments in the recent years and been able to show that arbitrary computations on encrypted values can be implemented [4]. However, the computations involved are presently prohibitively expensive even for full-fledged devices and highly infeasible for resource-limited devices. In order to support sum and average op-erations over encrypted data, it is how-ever sufficient to utilize additive homo-morphic encryption schemes, such that:

decrypt(c1 c2) = decrypt(c1) + decrypt(c2)

The Paillier cryptosystem is one of the most well-known, additive homo-morphic schemes [5]. One challenge of Paillier, with regard to the limited band-width in the IoT domain, is its ciphertext expansion. In the Paillier cryptosystem the plaintext size is between 1 to n bytes, where n is the key length. Regardless of the plaintext size, the ciphertext has a size of 2n. This would mean with a 1024 bit key, the encryption of a 32-bit integer value requires 256-byte space. Moreover, the encryption appears to be a costly operation for low-power devices. For in-stance, the encryption of a 32-bit integer requires 1.6 seconds on the ARM Cortex-M3 microcontrollers, which is several times higher than the 9 milliseconds re-quired on a desktop machine.

Deterministic (DET). Deterministic encryption allows for equality checks. The encryption of the plaintext m re-sults always into the same cipher c. AES in electronic codebook (ECB) mode is a block-cipher encryption with such a property. Due to this deterministic property it is in general advised not to use ECB for encryption of large packets, as an attacker can: change the order of the blocks or replace a block in an indis-tinguishable manner (i.e., substitution attack), or learn information about the plaintext with a histogram of repeated blocks. Therefore, for maximum secu-rity in DET, AES-ECB should be only used for plaintexts smaller than or equal to 16 bytes. For large plaintexts, AES in CMC mode can be utilized [6]. AES-CMC is a tweaked combination of AES-CBC with a zero initialization vector, where AES-CBC is applied twice on the input. The second CBC round is applied in the reverse order, i.e., from the last block to

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 252

feature

CONCLUDING REMARKSApplied cryptography is gaining popu-larity among cryptographers. System and networking researchers can ben-efit from this, in that they can capture more insights about novel cryptograph-ic schemes and explore their feasibil-ity and benefits in practical systems. The challenges due to the ecosystem of IoT applications require a more careful system design with regard to resource constraints. The effort of realizing theo-retical cryptographic approaches under real-world conditions sheds light on un-discovered weaknesses and creates op-portunities for enhancements in crypto-systems. It remains an important open research problem to design and prove secure and practical encrypted data pro-cessing schemes for the IoT domain.

References

[1] Kamara, S. Encrypted search. XRDS 21, 3 (2015).

[2] Wu, D. Fully homomorphic encryption: Cryptography’s holy grail. XRDS 21, 3 (2015).

[3] Popa, R. A., Redfield, C. M. S., Zeldovich, N., and Balakrishnan, H. CryptDB: Protecting confidentiality with encrypted query processing. In Proceedings of the 23rd ACM Symposium on Operating Systems Principles (SOSP ‘11). ACM, New York, 2011, 85-100.

[4] Gentry, C. A fully homomorphic encryption scheme. Ph.D. thesis. Stanford University: AAI3382729, Advisor: Dan Boneh. 2009.

[5] Paillier, P. Public-key cryptosystems based on composite degree residuosity classes. In Proceedings of the Annual International Conference on the Theory and Applications of Cryptographic Techniques (EUROCRYPT ‘99). ACM, New York, 1999, 223-238.

[6] Halevi, S. and Rogaway, P. A tweakable enciphering mode. In Advances in Cryptology (CRYPTO ‘03). ACM, New York, 2003.

[7] Boldyreva, A., Chenette, N., Lee, Y., and O’Neill, A. Order-preserving symmetric encryption. In Proceedings of the Annual International Conference on the Theory and Applications of Cryptographic Techniques (EUROCRYPT ‘09). ACM, New York, 2009, 224-231.

[8] Popa, R. A., Li, F. H., and Zeldovich, N. An ideal-security protocol for order-preserving encoding. In Proceedings of the IEEE Symposium on Security and Privacy. IEEE, Washington D.C.,2013, 463-477.

[9] Gentry, C., Halevi, S., and Smart, N. P. Homomorphic evaluation of the AES circuit. In Proceedings of Advances in Cryptology (CRYPTO ‘12). ACM, New York, 2012.

[10] Naveed, M., Kamara, S., and Wright, C. V. Inference attacks on property-preserving encrypted databases. In Proceedings of the 22nd ACM SIGSAC Conference on Computer and Communications Security (CCS ‘15). ACM, New York, 2015.

Biography

Hossein Shafagh is currently a second-year Ph.D. student at the computer science Department of ETH Zurich, Switzerland. His primary research interests are in the design of secure communication systems for low-power wireless devices.

© 2015 Copyright held by Owner(s)/Author(s). Publication rights licensed to ACM.

1528-4972/15/12 $15.00

the first block. This way, the first blocks become deterministically random and do not leak equality within a data item. The performance of DET is comparable to RAND.

Order-preserving encryption (OPE). The order relationship between the plaintext inputs m1, m2, and m3 is pre-served after encryption, i.e.,

if m1 ≤ m2 ≤ m3, then c1 ≤ c2 ≤ c3

This way, the order information among the encrypted data items ci is re-vealed, but the data itself is not. Order comparison is a common operation in SQL-like databases, such as for sorting, range checks, ranking, etc. One of the first provably secure OPE schemes is the approach introduced by Boldyreva et al. [7]. This OPE scheme is, however, as computationally intensive as Paillier en-cryption. The interactive OPE approach by Popa et al. solely relies on symmetric cryptography and trades computation overhead for latency (i.e., it involves more communication) [8]. This lightweight OPE scheme is referred to as mutable order-preserving encoding (mOPE), as the order encodings are mutable. Popa et al. prove that mOPE fulfills the ideal security (IND-OCPA), i.e., no additional information than the order is revealed. mOPE is more secure than any other OPE approach and, yet, one to two or-ders of magnitude less computationally intensive than traditional OPE schemes. Although mOPE appears to be more suit-able for the IoT, the status of opportunis-tic connectivity might have an impact on the performance of this protocol.

PRACTICAL SECURE SYSTEMSWhile designing a secure encrypted query processing system for the IoT do-main, the challenge becomes centered upon which entity should take the role of performing the encryption or decryp-tion operations and thus has access to encryption keys. Pushing this role to the origin of the data would allow protection of the data at the source, but poses the challenge of minimizing the impact of computational overheads on the battery-life of IoT devices. In the ecosystem of the IoT, the gateway could be utilized for heavy computations. Here the challenge is how to establish trust with the gate-way and execute private functions over

encrypted data. Once the data is stored in the correct form on the back end, the challenge becomes about how to allow other low-power IoT devices to securely retrieve information from the EDB.

It is important to stress that en-crypted query processing is possible due to utilization of weaker encryption schemes, such as DET and OPE, which leak information. Hence, the decision about which data types to encrypt with the property-preserving encryption schemes becomes very important. Re-cently, Naveed et al. showed how simple attack techniques can be used to disclose encrypted medical data in the OPE and DET schemes [10]. The attack schemes require access to auxiliary information, such as range of data and distribution of it, which is seemingly simple to retrieve for certain application scenarios. Given the protected plaintext has low entropy, techniques such as frequency attack can be used to learn about encrypted data. With respect to IoT data, the question is: How applicable are such attacks on sen-sor readings?

The resource asymmetry in IoT sys-tems makes research efforts on homo-morphic evaluation of the AES circuit appear promising [9]. Efficient and low-power AES encryption is readily available in most radio front ends of constrained devices. The possibility of transforming AES ciphertexts into homomorphic ones would allow for the minimization of the encryption overhead on IoT devices, while allowing systems to exploit the full potential of fully homomorphic encryp-tion in the powerful cloud. In more prac-tical terms, this means the current com-putational overhead of four minutes for a single AES-128 encryption operation must be significantly reduced.

The need for cloud storage in the IoT is unavoidable, not only to store data, but also to make this data accessible to other devices and services.

The newest ACM forum. Contributions that cover the vast information-rich world where computing is embedded everywhere.

ACM’s Ubiquity is the online magazine oriented toward the future of computing and the people who are creating it.

We invite you to participate: leave comments, vote for your favorites, or submit your own contributions.

Captivating topics. Net Neutrality and the Regulated Internet

The End of Life As We Know It

A Shortage of Technicians

The Fractal Software Hypothesis

Your Grandfather’s Oldsmobile–NOT!

Superscalar Smart Cities

Visit us at http://ubiquity.acm.org/blog/

UbiquityBlog_FP_V04.indd 1 3/13/15 10:03 AM

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 254

feature

The Ambient Intelligence Course at Politecnico Di TorinoAt Italy’s oldest technical university, students learn about IoT concepts and technologies by building end-to-end prototypical systems.

By Luigi De RussisDOI: 10.1145/2845159

function of an AmI system proceeds in four cyclical steps: sensing from the environment, from the people, and/or from the Internet; reasoning about the sensed data; acting in the envi-ronment; and interacting with users so as to keep them in the loop. Well-designed AmI systems should have six features: sensitivity, responsiveness, adaptiveness, transparency, ubiquity, and intelligence [2]. Knowledge of the function and requirements of AmI systems contributes to define the over-all educational goal of the course: to outline the multidisciplinary skills re-quired for designing IoT systems from a technological, procedural, and com-municative standpoint.

The course spans 14 weeks in the second semester—spring to sum-

T he last five years have seen the Internet of Things (IoT) become mainstream, and it is expected to continue growing in the coming years [1]. We are already seeing an increasing use of Internet-enabled devices that promise to improve—if not completely change—various aspects of our daily lives. IoT devices and technology,

moreover, are not merely a consumer phenomenon. Challenges in the field are attracting the attention of the research community and the industry, with big players eager for new market opportunities. The opportunities and challenges brought on by the IoT require a new generation of engineers and professionals capable of dealing with design problems in a versatile manner. The multidisciplinary nature of IoT systems precludes confining their

study to any single discipline. The edu-cational framework in which learning occurs, and the set of skills future IoT engineers must possess, are much larg-er than those required by traditional disciplines. Hence, current teaching programs need to be complemented with new courses that build upon, and integrate, the skills and knowledge ac-quired in different technical fields.

In 2014, the Politecnico di Torino—one of the major technical universities in Italy—took a step toward tackling this challenge by offering a new course: “Am-bient Intelligence: Technology and de-sign.” The course applies a team-based design-driven paradigm of teaching how to realize end-to-end IoT systems, starting from an idea that may solve an existing problem and ending with a pro-

totypical, but fully working, system. Pro-posed by the Department of Control and Computer Engineering, it is an elective course taught in English and available to all students enrolled in their third, and final, year of one of the 22 bachelor’s degrees offered at Politecnico.

COURSE OVERVIEWThe course approaches IoT system de-sign by applying principles and meth-odologies from the field of ambient intelligence (AmI) research, which is an area at the intersection of artificial intelligence, human-computer inter-action, ubiquitous computing, and sensor networks. An AmI system can be defined as “a digital environment that proactively, but sensibly, supports people in their daily lives” [2]. The Im

age

by G

abri

delc

a/co

mm

onsw

iki

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 55

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 256

feature

tem by applying AmI principles. The IoT system developed, together with various documents produced during the learning period, are the objects of the course grade. In addition, teams are expected to present their work and have an oral discussion about their projects. Every year a common theme for projects is proposed to bet-ter enable students to focus on ideas in line with the educational goals of the course. But each team is responsible for finding an idea that is suitable for the course and aligned with the typi-cal features of an AmI system. As an example, for the 2014–2015 academic year the theme was the “smart univer-sity campus,” students were required to identify, design, and prototype sys-tems bringing AmI features inside the Politecnico di Torino campus. Target environments could be classrooms, li-braries, offices, hallways, open spaces, cafeterias, and laboratories. The main requirement for developed ideas was to bring tangible benefits to students, teachers, staff, and visitors.

After approval from teachers, the teams start developing their ideas ac-cording to the proposed design meth-odology. This follows four main steps: 1. visions and goals definition, 2. func-tional and non-functional require-ments elicitation, 3. system architec-ture design and component selection, and 4. practical realization of the pro-totypical system.

Students are asked to produce a document, or an artifact, for each of the above-mentioned steps. The document is made publicly available on the Internet, and is evaluated for the final grade. The teaching team provides feedback and guidance at each step. Teams are encouraged to develop open and reusable solutions that involve sensing, actuation and interaction, and intelligence. The last requirement implies solutions should not be simply deterministic. Projects cannot be mobile-only, software-only, or hardware-only solutions, but must exploit different platforms and mix hardware with software and user-interaction. All projects must imple-ment the four main functional stages of an AmI system and include most AmI features.

The teams have access to various

mer—with four-and-a-half hours of classes and labs per week. Classes are video recorded and enrolled students can view the videos online. A freely available version of the video lectures is also available as a YouTube playlist (see http://bit.ly/polito-ami for further details).

The course applies the active learn-ing methodology with a bias toward practical design activities and super-vised work groups. In fact, traditional lectures comprise less than one-third of the course. These give the students the requisite background in:

˲ The Internet of things and ambi-ent intelligence definitions, and avail-able approaches for smart homes, smart buildings, and smart cities.

˲ Overview of application areas such as homes, buildings, or traffic, and types of applications such as mon-itoring, comfort, and ambient assisted living.

˲ Requirements and design meth-odologies for AmI systems

˲ Design, analysis, and specifica-tion of requirements and functional-ities related to user interaction with AmI systems

˲ Practical programming of AmI systems. This includes the teaching of the Python language, the Raspberry Pi

system, web protocols and languages, and web-based API and collaboration tools like Git.

Enrolled students come from dif-ferent academic programs—mainly computer engineering, electrical en-gineering, mechanical engineering, and industrial design. International-ity is considered vital for fostering new ideas and, as a consequence, 20–25 percent of students come from foreign universities.

Throughout the course, students are exposed to real or realistic design problems, and they are required to provide intelligent solutions in the context of environments, where hu-mans and commercial and custom-built devices interact and cooperate. Knowledge about programming tech-nologies is typically provided in the form of hands-on tutorials, with stu-dents actively engaged in formulating solutions to proposed problems, and teachers showing how to exploit par-ticular technologies and languages to address the issues under consider-ation.

Students are required to split into teams of three to four at the start of the course, and each team is sub-sequently guided to define require-ments, design, and realize an IoT sys-

Figure 1. The 2014 Ambient Intelligence Student Showcase held at Politecnico Di Torino, Italy.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 57

commercial materials and devices, provided by the Politecnico and local businesses, to realize their projects. In particular, they can work with two BTi-cino hardware developer toolkits con-sisting of various home automation de-vices; various ZigBee and Z-Wave home automation devices; one Philips Hue kit with six bulbs and an LED strip, two Pebble smart-watches; several Rasp-berry Pis and Arduinos spare hardware components such as breadboards, LEDs, and resistors; a few Android 4.x tablets; various computer peripher-als including webcams, microphones, and speaker; some RFID/NFC tags and readers; and a dozen Bluetooth bea-cons. For each technology, the relevant APIs and libraries are explained dur-ing the hands-on sessions.

Activities are not limited to “tradi-tional” study and homework: Students are encouraged to be proactive. Teams are required to give presentations, publish and update a public website to present their work, and produce a video pitch sharing the idea underly-ing their project. Moreover, each team is required to maintain all source code related to its project in an assigned Git repository. The private reposito-ries are generously hosted on GitHub. Project repositories are part of a dedi-cated GitHub organization and teams are required to have a corresponding website, created using GitHub Pages, for hosting all the realized deliverables and artifacts.

The course, although demanding, is also very satisfying from the teach-ers’ and the students’ standpoint. In their final evaluation questionnaires, students acknowledged the course “is very hard,” but they learned a lot about IoT systems and handling a complex project. Some of the stu-dents, moreover, said the course “was the best [they had] ever taken.” Teach-ers, on the other hand, were positively impressed by the students’ maturity, their commitment, and the variety of generated ideas.

STUDENT SHOWCASEAfter the end of the course, students who successfully passed the exam are encouraged to participate in the Ambient Intelligence Student Show-case. Students present their work to

peers, university staff, and to industry members, who have all been invited to attend the showcase. In addition to developing the students’ communica-tion skills, the showcase is an oppor-tunity for students to receive feedback on their project and to network with potential employers. Each team is sup-plied with a table, a poster holder, pow-er supply and network connection, and any other materials needed to perform a demo of their ideas.

The 2014 edition of the showcase took place on September 30th at the innovative enterprises incubator (I3P) of the Politecnico di Torino with more than 100 people attending. The guests were almost equally distributed be-tween students, Politecnico staff, fac-ulty members, and employees from lo-cal startups and businesses.

The 2015 edition of the showcase was also held at I3P with 11 teams showing their projects, targeting dif-ferent application areas in the smart campus domain. The names of the 11 projects, and a brief description of their functionality, are as follows:

˲ ItsYourTurn, a system that will help students and professors to get in touch more easily and efficiently.

˲ MarcoPoli, for finding out which places inside the campus have critical settings, in terms of humidity, tem-perature, people congestion, light, and noise level.

˲ NeverLate, a smart-watch for stu-dents who would like to never miss out a lecture and enjoy breaks during long lessons.

˲ NoNoise, a system that helps you find a place with the desired noise level for every student activity.

˲ SmartClassSchedule, a system that aims at improving the classroom schedule by putting contextual infor-mation in front of each classroom.

˲ Wc Info, for warning students about out-of-order bathrooms and waiting times, with real-time updates.

˲ TrackDown, a system that uses indoor tracking to lower the possibili-ties of losing valuable items inside the campus.

˲ EasyPark, to direct university staff to the closest parking lot with available space, and check license plates upon entrance.

˲ WellCleaned, for enhancing bath-

room maintenance by constantly mon-itoring the levels of soap, toilet paper, trash cans, and informing students and maintenance staff.

˲ SmartMakeYourBag, a system that uses active tags to remind stu-dents which items to add or remove from their bags, depending on daily class schedule.

˲ MyBikePlace, a system to know where you can leave your bike quickly and safely on campus, near your posi-tion, or around your classroom.

All the projects realized in the 2015 edition of the course are described at http://ami-2015.github.io. As a dem-onstration of support and interest in these topics, the 2015 edition of the showcase is sponsored by four IT cor-porations, and by one startup that op-erates in the smart home domain.

CONCLUSIONThe IoT has already proliferated our daily lives and it is expected to fur-ther evolve in the coming years, thus, requiring proper training for the next generation of technical profession-als. The course aims to face the chal-lenges and opportunities brought on by the IoT. The course also teaches students how to handle the complex-ity of the involved systems, and, at the same time, how to design solutions centered on people needs in multidis-ciplinary teams.

ACKNOWLEDGEMENTI would like to thank all of the 112 stu-dents enrolled in the course in 2014 and 2015, as well as my colleagues Fulvio Corno and Dario Bonino whom shared with me the course design—and workload—during these two years.

References

[1] Evans, D. The Internet of Things: How the next evolution of the Internet Is changing everything. Cisco Internet Business Solutions Group. White paper. San Jose, CA. 2011.

[2] Cook, D. J., Augusto, J. C. and Jakkula, V. R. Ambient intelligence: Technologies, applications, and opportunities. Pervasive and Mobile Computing 5 (2009), 277-298.

Biography

Luigi De Russis is a postdoctoral research assistant at the Department of Control and Computer Engineering at Politecnico di Torino. He received his Ph.D. in computer engineering from Politecnico di Torino in 2014. He is a teaching assistant in the ambient intelligence course.

© 2015 ACM 1528-4972/15/12 $15.00.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 258

feature

Panasonic and the OpenDOF Project: Open-source vision in a large company Is the decision to go open-source always purely altruistic? Not for many large companies, and that is not a bad thing.

By Bryant Eastham DOI: 10.1145/2845161

many of the leading software stacks are open-source projects. The AllSeen Alliance, an open-source project with more than 170 member companies, is an example of the importance of an open-source IoT initiative. In March 2015, Panasonic created the non-profit OpenDOF Project (opendof.org) as a means to make its IoT software stack open source. Looking back on Matsu-shita’s action 83 years earlier, the ques-tion remains, why did Panasonic make this decision? The answer requires us to start from the beginning.

In October of 1932, Konosuke Matsushita purchased the rights to two patents critical to the manufacture of radios and then donated them to the community. He took this unprecedented action to expand the nascent radio industry in Japan. Matsushita was the founder and president of what would become Panasonic, global manufacturer of household appliances,

consumer electronics, and thousands of other commercial products. That highly successful global company was still a distant vision in the 1930s, when the company had only recently started to sell radios. What would guide Matsushita to make such a bold decision? What principles guide companies both large and small to invest time and money in something that they turn around and “give away”? Is altruism the only explanation or is there more to it?

The success and longevity of the open-source movement certainly proves there are compelling reasons for companies to give away the results of their software development. One of the largest open-source projects, the Linux kernel, reports annually on its contribution statistics.1 So far they have concluded each release has con-tained contributions from more than 1,400 developers representing more

1 http://www.linuxfoundation.org/publications/linux-foundation/who-writes-linux-2015

than 200 companies. More than 75 percent of large enterprises use Linux as their primary cloud platform [1], an increase of 14 percent over the last three years. Another large open-source group, the Apache Foundation, reports they have several thousand contribu-tors. Clearly, more and more compa-nies are relying on open-source proj-ects. In fact, a recent report indicates 78 percent of companies rely on open-source projects, and 64 percent partici-pate in them [2].

In the Internet of Things (IoT) space,

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 59

acting with each other or the collec-tion and analysis of data from these devices, a means of communicating information is required. This requires networking protocols, similar to TCP/IP, which forms the basis for the In-ternet, but at a higher “application” level. There is no shortage of network-ing protocols that have been, and are used, but there is also no universally accepted standard. In the absence of a standard, we can work to identify com-mon requirements that any IoT system should satisfy. These requirements

First, it helps to understand a little of the IoT area itself. The term “Inter-net of Things” is far too generic and therefore can be used by almost any-one to mean just about anything. For the purposes of this article, let us boil the term down to its basics and assume that it means: information exchange, analysis, and presentation on a mas-sive scale, typically between devices that operate more or less independent of human interaction. The term “In-ternet” indicates a variety of different network transports, like ethernet, in

reference to the global Internet that we are all familiar with; the term “things” focuses attention on the device rather than people. We can think of the IoT like email for devices, combined with data collection and analysis on the scale of the Internet itself. If one be-lieves the market projections, the scale of the IoT will greatly eclipse every cur-rent people-centric messaging system. This makes sense since most people have multiple things that will become connected.

Whether we consider devices inter-

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 260

feature

from this kind of development, there are specific benefits for IoT solutions. Most important is the obvious benefit that open-source projects allow users to inspect the code. We live in an in-creasingly pessimistic time, where re-cent revelations on government and industry snooping and lack of security have made people distrustful of large, closed solutions. In many cases, this distrust spills over to companies that store and analyze our data. The ability to inspect code helps to mitigate this distrust. Even if we do not look at every line of code, the knowledge that many eyes are looking (or at least could be looking) is advantageous.

Also important, and related, is the benefit of increased security in open-source projects. Few developers are security experts. The problem is even if developers build on the efforts of those who are security experts, they must still integrate those secure solutions into their software—a task that itself can inadvertently introduce security flaws. This means the best IoT solutions will both leverage the security solutions of experts, as well as allow the review of their final product by the community at large (including those who are secu-rity experts). Ensuring the community gains the security benefit described is a goal of the Linux Foundation’s Core Infrastructure Initiative (www.corein-frastructure.org), which helps to ad-dress one shortcoming of open-source projects: The inevitable decrease in focus over time for even the most suc-cessful and leveraged projects.

Much of the focus of open-source proponents is on the benefits that oc-cur during early-stage development of the open project. It is during this time that collaboration is maximized, the benefit of meritocracy is most keen, and there is sufficient excitement to maintain a community. A potential down-side of the open approach is this early excitement diminishes as the work transitions to maintenance and issue resolution. The requirements of the community change, and the work of visionary developers needs to transi-tion to those willing to support the code long-term. This seems like the Achilles’ heel of open source until one compares it to the closed-source equivalent. It is naive to expect development projects

can help weed out solutions that may not be appropriate.

Panasonic has identified the fol-lowing requirements as a guide for the OpenDOF Project:

˲ Security. IoT solutions must be se-cure to be successful. Not everyone is a security expert, and expecting applica-tion and system programmers to build secure solutions without established frameworks is wishful thinking.

˲ Interoperability. Even if a stan-dard were established and accepted soon, there would still be still hun-dreds of thousands of devices in the field. Further, given the longevity of devices, it is unrealistic to believe any accepted standard would remain un-changed during the lifetime of a device.

˲ Flexibility. The breadth of IoT solutions is truly staggering, and so-lutions must adapt to the differing re-quirements of each solution.

˲ Scalability. Collective industry experience with scalability is limited to applications that interact with hu-mans and on platforms that tend to be refreshed frequently. Imagine having several billion DOS-based devices in the field today. Further, core infrastructure is likely to be unable to deal with the proposed scale of the IoT. Issues like bandwidth sharing and device firm-ware updates need to be resolved.

˲ Reliability. IoT adoption by criti-cal infrastructures will raise the im-portance of system reliability.

These guidelines have been fol-lowed by the OpenDOF designers for almost two decades. The formation of the OpenDOF Project is just the latest in a series of developments related to this technology. In 2005, Panasonic ac-quired emWare, Inc., a company based in Salt Lake City, UT. Since 1996 this group had been working on a device-networking stack that aimed to satisfy many of the requirements previously listed. Panasonic continued to develop this technology until 2015, when the decision was made to release it as open source. This echoed a similar decision made by emWare in 2000. At that time, the President of emWare, Michael D. Nelson, stated: “Our strategy is a bold move: releasing elements of our intel-lectual property as a means for em-Ware to ignite the device networking market. By embracing an open strat-

egy, emWare and other companies will collaborate to engineer more efficient, interoperable device networking solu-tions that no company could produce single-handedly” [3].

Fifteen years later, this sentiment was echoed by Todd Rytting, the CTO of Panasonic North America, when he said the following in the announce-ment of the OpenDOF Project: “Open sourcing a proprietary technology in-vites the open-source community to evaluate, work on, and ultimately im-prove the software. In a market full of incompatible proprietary offerings, this initiative brings a powerful tool to developers and equipment makers to help them create what the market wants in the IoT: interoperable and flexible services and applications le-veraging data from connected devices, and most importantly, value to the cus-tomer” [4].

For more than 15 the idea that open technologies foster collaboration and provide customer benefit has remained the same. The “open source way” is a term used to describe this form of col-laboration. It is focused on the open ex-change of ideas and information, par-ticipation by a variety of stakeholders, the willingness to act quickly and “fail fast,” and an emphasis on community that can build on the best work of its members (an example of meritocracy). This kind of collaboration allows many groups to work together, producing better solutions than any group could produce on its own.

While many markets can benefit

Whether we consider devices interacting with each other or the collection and analysis of data from these devices, a means of communicating information is required.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 61

to develop both client and cloud appli-cations for use not only on Panasonic but also other industry hardware. This agnostic view of development fosters the Panasonic commitment to com-munity and industry.

All of these efforts focus on en-abling customers and partners to do more with Panasonic products and services than they could do before. So, in the end, the decision to create the OpenDOF Project is viewed as a way to help our customers and part-ners—and that always makes good business sense.

In 1932, when Konosuke Matsush-ita decided to “open source” his radio patents, computer software did not exist. But the decision to do so then was just as multi-faceted as the deci-sion to form the OpenDOF Project 83 years later. In both cases, the industry reacted positively. The release in 1932 helped to spur an industry and solidify the growth of what would eventually become the Panasonic of today.

History will judge the relative im-portance of the 2015 OpenDOF Project release and its impact on the IoT. The combined effect of open-source proj-ects in general, and particularly in the IoT, has already been judged, and is not only a success, but a required fea-ture moving forward.

References

[1] Orion, E. Linux is winning enterprise cloud market share at expense of Windows. Inquirer. http://www.theinquirer.net/inquirer/news/2385329/linux-is-winning-enterprise-cloud-market-share-at-expense-of-windows

[2] Blackduck. The Ninth Annual Future of Open Source Survey. 2015. https://www.blackducksoftware.com/future-of-open-source

[3] emWareR Opens Its emNetT Protocol. Appliance Design. Sept. 28, 2000. http://www.appliancedesign.com/articles/88177-emwarer-opens-its-emnett-protocol

[4] Panasonic. Panasonic to Share Intellectual Property to Spur Growth of Internet of Things. March 22, 2015. Press release. http://shop.panasonic.com/about-us-latest-news-press-releases/Content03232015092444644.html

Biography

Bryant Eastham serves as the principal software architect for Panasonic Corporation of North America. He is responsible for establishing the company’s architectural road map and vision for distributed platforms and the Internet of Things. Eastham sets and shares Panasonic’s architectural vision across R&D projects and products developed in North America.

© 2015 Copyright held by Owner(s)/Author(s). Publication rights licensed to ACM.

1528-4972/15/12 $15.00.

inside large companies do not follow similar trends of diminishing atten-tion. What’s more, with closed-source, when the projects diminish there is no external way to revitalize them.

Panasonic understood many of these benefits when making the deci-sion to open source its IoT platform as the OpenDOF Project. If only these reasons were considered, that would seem to make the decision almost to-tally altruistic. Even with its corporate principles like “contribution to soci-ety,” there was more to the story (as there often is when companies leverage open source). In this case, there was constructive pressure from customers to provide open-source solutions. Pan-asonic, particularly in North America, is a B2B company. We collect data from solar farms and in-flight weather sta-tions as well as a variety of other non-consumer products. Partners prefer open source as a form of insurance, and Panasonic’s partners are no differ-ent. An extension of this partnership is the idea of customer-as-a-partner and the rise of the “hackathon” mentality. The logic is by enabling the technology-savvy customer to produce extensions of your products, you create a win-win environment. The company wins by helping to foster a loyal customer base that actively promotes the company’s products; the customer/partner wins by being able to monetize their work. Open source and open APIs are a great way to facilitate this arrangement.

The need to strengthen the custom-er–partner relationship helps to clarify the broader vision inside Panasonic, where the open-source OpenDOF Proj-ect is only a single piece. To under-stand this, a more detailed explana-tion of the technology will help.

At its core, the project defines a distributed object model and messag-ing system built around a distributed object framework (DOF). The object model is built on interfaces that are similar to interface definitions in pro-gramming languages like Java or C#. These interfaces define types, proper-ties, methods, events, and exceptions. Interfaces are implemented and made available on a network by providers who associate the interface with an object. Both the object and the inter-face are identified using a naming

scheme that allows discovery and un-derstanding on the client side, clients can then interact with the objects over the network. This messaging system is integrated with a complete security solution that includes authentication, access control, and encryption.

The OpenDOF Project provides: ˲ Source control, defect tracking,

and continuous integration and test-ing systems.

˲ Native support for Java, C, and C#. ˲ Official builds of the compo-

nents suitable for integration, as well as the ability to build directly into a solution.

˲ Full protocol specifications. ˲ A patent pledge that protects us-

ers against patent claims by contrib-utors.

The Panasonic vision includes the open sourcing of the DOF technol-ogy and specifications (hence the OpenDOF Project), as well as much of the infrastructure required for using and testing it. For example, the tech-nology allows the definition of stan-dardized interfaces that are exposed by devices. Interface definitions can be easily shared with both partners and customers. To this end, Panason-ic has released a flexible platform for sharing these interface definitions for any technology stack.2 The documen-tation of these interface definitions, combined with a common API that le-verages them, is a primary goal of the OpenDOF Project. Together, they allow collaboration to cover the post-launch product life cycle.

Panasonic is also releasing its dis-tributed continuous testing frame-work as a separate open-source project on GitHub.3 As the IoT continues to scale, it will be important for develop-ers to be able to access actual hardware for testing. The Panasonic Distributed Testing Framework allows for a variety of automated testing scenarios, includ-ing both virtual and actual devices.

Today, Panasonic is working to fos-ter community around its products by launching the Solutions World Portal and Marketplace. This development portal allows our community partners

2 https://opendof.org/the-interface-repository-project/

3 https://github.com/PSLCL/testing-framework

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 262

feature

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 63

Toward a Web of SystemsWeb and semantic technologies will form the foundation for ecosystems of machines that interact with each other and with people as never before.

By Florian Michahelles and Simon MayerDOI: 10.1145/2845163

tors—factories, buildings, electric vehicles, and the urban infrastruc-ture—and consumers. This could help to flatten the electricity demand curve and match renewable energy sources with demand peaks. Previ-ously, each of these subsystems had its own norms and standards. Now, their products and solutions must be

T he concept of the Internet reaching out into the real world has been around for more than a decade. This development started by providing real-world things with a machine-readable identifier, such as an RFID tag or a barcode, and associating digital content with these IDs. Driven by the technological advances in processing,

sensing, and communication technologies and their decreasing cost, physical things have become ever more interconnected and pervasive, allowing them to collect real-time data and take control of other connected devices that affect the real world.

On top of this development, we can identify an emerging trend to develop solutions based on physical devices and interconnected digital services both in the consumer and industrial

domains. This will result in connected value chains of energy systems, public infrastructure, automated manufac-turing lines, and healthcare solutions. To leverage the potential borne by this development, every connected system and subsystem needs to be capable to make sense out of the data it receives, processes and shares.

This “digitalization”—a term that has become popular to describe the process of moving to a digital busi-ness—is leading to an ever-greater use of information and communication technologies. Thus, application areas are starting to overlap and interact. For example, smart grids can inter-act with decentralized energy genera-Im

age

by A

ndri

j Bor

ys A

ssoc

iate

s /

Shu

tter

stoc

k.co

m

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 264

feature

hierarchies (vertically) and domains (horizontally), and yields a paradigm that we refer to as the “Web of Systems.”

USABLE SEMANTICS FOR INTEROPERABILITYWe propose to use semantic technolo-gies to add shared meaning to informa-tion that is exchanged between compo-nents of a system and between systems across domains. The basic technologies to codify meaning in an appropriate way have been available for several decades. They have been applied to the World Wide Web in the context of the Seman-tic Web, but have yet to achieve break-through outside isolated application domains. We believe we can utilize se-mantic technologies to connect agents within and across several domains in a pragmatic way. Instead of “inventing” concepts and models for application domains, we propose to translate exist-ing domain standards into machine-readable representations. Furthermore, although we are working with technolo-gies that require huge up-front invest-ments for large future payoffs, we are attempting to apply an iterative develop-ment process where we already demon-strate (limited) added value early on. In the rest of this section, we will first in-troduce the core components of our ap-proach from a more technical perspec-tive, and then give a few examples of how these components are used within several of our current projects.

CORE COMPONENTSFrom a technical perspective, our pro-posed approach rests on three core com-ponents: First, we propose a method of facilitating the interconnection of het-erogeneous devices and services that builds on the emerging activity streams format. Second, using a common se-mantic framework, we enable these dis-tributed agents to share a common un-derstanding of the world, thus enabling them to “speak the same language.” Third, to ensure the created systems are accessible for human users, we propose the use of augmented reality (AR) tech-nologies to elicit semantic relationships and dependencies in the real world.

Activity streams for bridging In-ternet of Things (IoT) silos. Activity streams (AS) originated in the social web domain as a simple format that

mutually compatible, across different domains. However, norms and stan-dards of many different and previ-ously independent application areas must now become integrated with each other and work together seam-lessly, across vertical domains.

Vertical industry suppliers, like Sie-mens, have a strong tradition of com-mitment to standardization in their vertical businesses, such as industrial automation, energy generation and transmission, building control tech-nology, and mobility and medical technology. However, industry needs to actively tackle the described devel-opment toward applications that act across verticals. As an example, in the electric mobility domain, electric vehicles, charging stations, storage batteries, urban infrastructure, and power networks have to exchange in-formation about battery charge state, power availability, and energy system stability in real time. To perform to-gether in a meaningful way, these sub-systems need to truly collaborate. This collaboration goes beyond “syntactic” agreement about exchanged messages. Instead, a common understanding of meaning of shared information will be established on a semantic level. Only then will the system be capable of de-centralized coordination tasks such as using idle vehicle batteries for leveling peak loads and stabilizing power grids.

EMPOWERING MACHINES TO SHARE MEANING As machines will not only communi-cate within, but also across domains, new mechanisms of describing mean-ing to machines have to be developed. Unlike the human-created content on the web, much of the terminology and many processes in industrial domains already are partially structured thanks to standards, norms, agreements, and process descriptions.

Therefore, when aiming to intercon-nect systems vertically within a field and horizontally across domains, one main research question for industry is: How to translate domain-specific information represented in human-readable form into machine-readable shared understanding?

To approach this problem, we pro-pose to build upon established technol-ogies from the web and Semantic Web domains: We add meaning to machine-to-machine communication by estab-lishing an ontology of interlinked terms, entities, and relationships. For example, this ontology would contain a machine-readable definition of what an “electric generator” is and how it relates to the unit “volts.” We believe this approach is viable in many domains that are relevant for industry since these terms, entities, and relationships can be based on exist-ing domain-specific standards. Doing this will allow for communication across

Figure 1. Semantic core components of our approach. Our “Web of Systems Semantic Framework” enables horizontal interoperability while pluggable knowledge packs are responsible for linking concepts vertically within a domain. We propose to base as much of the machine-readable information as possible on readily available human-readable standards documents.

a

c

b b b b b

e

d Human-ReadableStandards Documents

Web of Systems Semantic Framework (WSF)

Knowledge Pack(Smart Grid)

Knowledge Pack(Industrial Automation)

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 65

main concepts such as information about units and dimensions (see Fig-ure 1a). The WSF also contains infor-mation about other abstract concepts that are reusable across domains, such as machine-readable definitions of what constitutes a problem, how such problems relate to states and so forth. For usage in concrete applica-tions, the WSF is extended with knowl-edge packs (KPs) that encode domain-specific information (see Figure 1b). Thus, the domain-specific KPs en-able vertical interoperability between agents within a domain (for instance, an electric car and a charging station; see Figure 1c), and their integration with the WSF ensures horizontal in-teroperability across domains (see Figure 1d). To facilitate the transition to this system and mitigate interoper-ability problems, we propose to base WSF and KP concepts on agreed-upon industrial standards. In these cases, it is thus sufficient to translate stan-dards documents into a machine-readable language, rather than invent-ing new concepts on a clean slate (see Figure 1e).

AR to uncover hidden relationships. One crucial factor to support the adop-tion of semantic technologies by sub-ject matter experts is to make ontolo-gies more tangible, and maybe even allow (limited) changes to the employed domain models at run time. We believe this could be achieved by using AR sys-

allows to link walled gardens such as Facebook or Twitter. It makes informa-tion about (user) activities, which take place in the scope of one of those plat-forms, available to outside partners. We propose to take this one step fur-ther, using AS as the foundation for a connective fabric IoT silos. To test this approach, we created ASbase, a cus-tom-built AS broker that allows clients to easily implement distributed AS-based scenarios [1]. The main purpose of this platform is to collect activities in the AS format and distribute them to interested parties, where it supports both a request-response and publish-subscribe pattern. For instance, clients can subscribe to streams that concern a specific patient in a healthcare set-ting, and also select the concrete data streams they are interested in (such as heart-rate data). To enable this, AS-base implements a filtering language that is based on the query language of the widespread MongoDB NoSQL database system. The openness and extensibility of AS allow users of the format to increase the richness of con-cepts that can be represented by them. When describing several of our current projects in the next section, we present a scenario where AS are used for inte-grating heterogeneous devices that collect health data, analysis services, and a visualization tool for health pro-fessionals within a single vertical silo. Furthermore we showed their applica-

bility to bridging such vertical silos in an experiment where multiple project groups used AS to interconnect initial-ly independent implementations and form cross-vertical mashups [1].

Semantic framework. After laying the foundation for seamless machine-to-machine (M2M) communication using AS, the next crucial step is to make this communication meaning-ful by enabling distributed agents to share a common understanding about exchanged information. We propose to tackle this problem with technolo-gies known from the Semantic Web domain that are applied in a prag-matic way to real-world problems. In particular, we propose the creation of a set of core ontologies, referred to as the “Web of Systems Semantic Frame-work” (WSF) that captures cross-do-

As machines will not only communicate within, but also across domains, new mechanisms of describing meaning to machines have to be developed.

Figure 2: Visualization of wearable health sensor data. Our visualization dashboard displays data streams from heterogeneous wearable health devices in a way that enables doctors to quickly gain an overview of the patient’s health-related activities.

Wearable Health Devices transmit patient data

Doctors customize dashboard To fit diagnostic needs

Patients and doctors gain insights about the patient’s health condition and exercise level

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 266

feature

dencies. For instance, a voltage regu-lation application requires access to voltage data streams. This enables sub-stations to semantically resolve such dependencies and automatically install required additional applications. Us-ing cross-domain knowledge from the WSF, our smart grid KP furthermore relates application functionality to po-tential grid problems—for instance, the KP encodes the fact that a voltage regulation application resolves voltage band violation problems. Together, the WSF and the domain-specific KP thus deliver a semantic layer that greatly fa-cilitates the operation of a secondary substation, and even holds the poten-tial to fully automate the resolution of common issues in electric grids.

Goal-driven manufacturing. In in-dustrial manufacturing, our approach of combining semantic technologies with seamless device communication is valuable with respect to a novel way of controlling manufacturing devic-es. In contrast to process-driven ap-proaches where relationships between manufacturing devices (such as a weld-ing machine and a robot arm) are stati-cally defined, we make use of embed-ded semantic descriptions of device functionality to dynamically create mashups that fulfill the production’s specified goal (such as “produce a red car door,” see Figure 3). The main ad-vantage of our system is its high degree of flexibility, as service mashups can adapt to dynamic environments. Addi-tionally, they are fault-tolerant with re-spect to individual devices becoming unavailable, for instance because they undergo maintenance [3]. The goal-driven control of manufacturing de-vices is particularly valuable to reduce machine tooling times that are an im-portant factor especially when produc-ing small batch sizes. Our approach holds the potential to have manufac-turing lines reconfigure themselves at runtime, based on descriptions of the functionality of individual devices, and even considering “non-functional” properties that influence the process indirectly, such as the required time or monetary cost of a process.

CHALLENGESBy establishing a Web of Systems, we can enable machines to collaborate

tems that overlay semantic information on real-world scenes and thereby en-able humans to directly observe which background information a system is processing. Beyond displaying such rather static information, the usabil-ity of semantics-driven systems could be increased even further by showing runtime data on the AR overlay, such as communication between devices.

CURRENT PROJECTSWe will demonstrate the value and power of our approach in the context of a selection of three projects: the healthcare, smart grid, and industrial automation domains.

HealthViz: Professional visualiza-tion of wearable health sensor data. One concrete example of AS and an AS-base platform is the HealthViz project, in which we investigate the integration of wearable sensors from the consum-er wellness domain into a professional health IT platform [2]. In this scenario, individuals use wearables or services on portable devices (such as activity tracker applications on smartphones) that allow them to collect information about their level of exercise and other physical activities. Health coaches or doctors can then use this information to better assess a patient’s lifestyle, and to monitor and suggest lifestyle changes. Here, AS and ASbase are used to decouple the data collection from concrete devices and services from

the use of the collected data in further stages of the data processing pipeline, such as during aggregation and analy-sis, and also from the final consump-tion by user interfaces that are target-ed at health coaches and doctors.

Our visualization interface (see Fig-ure 2) consumes AS activities and data representations that are linked from them and displays them in a dashboard-style way, which gives the health coach or doctor a quick overview of the most relevant analyses of the recent past. If they choose to do so, they can drill down into the data, in which case more de-tailed representations will be accessed and visualized. The ASbase subscrip-tion API can also be used to have the in-terface dynamically update itself while being used by the health professional.

Intelligent secondary substation for smart grids. While the HealthViz sys-tem focuses strongly on the classical in-tegration of heterogeneous services in an IoT context, we demonstrate the val-ue of our approach of combining a core WSF with domain-specific KP in the smart grid domain. Here, we support the creation of an intelligent secondary substation prototype that comes with the capability of deploying additional functions at run time via an “app store.” Our primary contribution to this project is a domain-specific KP that describes which applications can be deployed on a secondary substation together with metadata such as application depen-

Figure 3. Based on a production goal (a) and functional descriptions of individual devices (b), a Reasoner (c) derives a production plan that can immediately be implemented, (d) and takes into account dynamic context factors such as individual devices being unavailable.

Reasoner

Goal Description

Production Plan

Produce a red car door!

Robots

Conveyors

Welding Machines

Paint Mixers

a

c

d

b

b

b

b

Implementation of Production Plan

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 67

horizontally within domains as well as vertically across domains. Web tech-nologies and semantic technologies are key enablers for this development. However, when implementing this paradigm on a larger scale, a variety of challenges have to be solved.

Engineering: Supporting the effi-cient creation of ontologies from stan-dard documents. One challenge lies in facilitating the process of “translating” today’s merely document-based hu-man-readable to machine-readable on-tologies. In order to deploy semantics on a larger scale, tools and methods have to be developed that overcome the limits of a manual translation process and can support domain experts in the codification of knowledge for its usage by computer systems.

Lifecycle management: Account-ing for upgrades and extensions. Once a domain-specific ontology has been created, it has to be kept up to date to changes and updates. Thus, efficient ways of updating and extending on-tologies, as well as guaranteeing the consistency of distributed knowledge bases, are crucial to drive adoption.

Virtual affordances: Informing us-ers about virtual cause-effect relation-ships. Humans are adept to spotting cues that convey how physical and virtual objects should be used, for in-stance whether a door can be opened by pushing or pulling. However, as physical devices get more connected and become parts of virtual systems, human actions will trigger side effects across virtual and physical spaces. We therefore require new and intuitive ways to inform users about the effects of their actions and the virtual, hid-den, causal relationships. Perhaps, technologies such as AR are well suited to tackle this challenge.

Provenance: Tracing causes and their effects in the Web of Systems. On a more abstract level, we are very interested in tracing cause-effect rela-tionships ex post to gain a deeper un-derstanding of a specific system. For instance, this could allow tracing fluc-tuations in the output of a manufactur-ing plant back to specific machines op-erating below capacity or undergoing maintenance. Beyond industrial auto-mation, a similar system could allow doctors to trace analysis results about

a patient’s health back to the specific measurements that gave rise to these findings, which might be valuable dur-ing diagnosis processes.

Business perspective: Creating a business model that supports sharing ontologies. Ontologies will be valuable if they can be shared across partners and re-used over and over again. For instance, ontologies such as QUDT (www.qudt.org)—which contains in-formation about quantities, units, and data types—are relevant in most domains, and therefore should consti-tute standard building blocks. Thus, organizational structures, business incentives, ways of protecting IP, and managing the distributed semantic knowledge base have to be identified that permit sharing the cost and ben-efits of these initial investments.

DISCUSSION AND CONCLUSION The digitalization of industries has become a reality that requires new concepts of managing the informa-tion exchange between machines. To fully leverage the potential of cou-pling services to products, machines will need to collaborate horizontally across domains. This calls for de-veloping mechanisms to add shared meaning to data that build upon es-tablished domain-specific standards, but make them usable in machine-comprehensible ways.

AS provides a good mechanism to coordinate and synchronize collabo-ration across devices in an emerging (industrial) IoT. Within that space, the development of shared vocabularies is valuable for bridging applications across IoT domains, which should support the creation of a common AS description language. We suggested

using semantic technologies to encode domain-specific standards and con-nect them to established and existing cross-domain ontologies. Finally, we proposed to apply AR technology to provide access to the invisible digital properties of smart devices and to help maintain and curate the correspond-ing semantic models. We also present-ed various examples and first results of our efforts to digitalize processes in the healthcare, smart grid and indus-trial manufacturing domains.

We believe enabling machines to “speak the same language” is pivotal for transforming physical products into digitally connected solutions. Although we have presented our first steps toward a Web of Systems, there is still a lot to do. We invite you to join us, and together we can make the digitali-zation of industry happen.

ACKNOWLEDGEMENTSThis article is based on joined work of the Siemens Web of Things research group in Berkeley, CA. The authors want to thank particularly Jack Hodg-es, Mareike Kritzler, Ralf Mossham-mer, and Dan Yu.

References

[1] Mayer, S., Wilde, E., and Michahelles, F. A Connective Fabric for Bridging Internet of Things Silos. In Proceedings of the Fifth International Conference on the Internet of Things (Seoul, South Korea, Oct. 26-28). IEEE, Washington D.C., 2015.

[2] Ryokai, K, Michahelles, F., Kritzler, M., and Syed, S. Communicating and Interpreting Wearable Sensor Data with Health Coaches. In Proceedings of the Ninth International Conference on Pervasive Computing Technologies for Healthcare (Istanbul, Turkey, May 20-23). IEEE, Washington D.C., 2015.

[3] Mayer, S., Inhelder, N., Verborgh, R., Van de Walle, R., and Mattern, Friedemann. Configuration of Smart Environments Made Simple. In Proceedings of the Fourth International Conference on the Internet of Things (Cambridge, Oct. 6-8). IEEE, Washington D.C., 2014.

Biographies

Florian Michahelles heads the Siemens Web of Things research group. Having worked in the fields of ubiquitous and wearable computing for more than a decade, Michahelles’ focus at Siemens is to leverage the web architecture and semantic technologies for enabling new business opportunities, especially in the fields of wearable sensing and human-robot interaction.

Simon Mayer has a strong background in distributed systems and Web of Things research; his focus at Siemens is to apply his knowledge on (Semantic) Web technologies, functional modeling, and human-computer interaction to enable interoperability and self-configuration across devices in factory and building automation scenarios. Mayer also co-chairs the international Web of Things (WoT) workshop series.

© 2015 Copyright held by Owner(s)/Author(s). Publication rights licensed to ACM.

1528-4972/15/12 $15.00

We can utilize semantic technologies to connect agents within and across several domains in a pragmatic way.

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 268

feature

PROFILE

software on Mac,” he shared.After this first taste in software

entrepreneurship, things snowballed. In 1985, Steve Jobs was ousted from Apple, and went on to found NeXT Computer. At the time, NextStep was a giant leap forward as far as operating systems were concerned. Pryor’s group began developing on NextStep’s system components. They expanded upon a sub-component called “NetInfo,” which was an early version of LDAP. Thanks to a deal with NeXT, Xedoc developed a multi-platform version of NetInfo, which had them involved in many successful corporate sales stateside. They soon had their first taste for how an Australian company might compete in Silicon Valley.

FROM MELBOURNE TO SILICON VALLEY AND BACKFollowing the team’s increased engagement with enterprise software, Xedoc ended up being pivoted into a company called Verve, which developed what was arguably the world’s first dedicated business process management engine. Verve eventually took the team to Silicon Valley. “I moved to San Francisco with my wife and one year-old son in 1997, and we lived there until 2002, through the heart of the dot-com boom. And bust, I guess, on the tail end there. Ultimately, our company was acquired in the heady days of the dot-com era by a newly minted public company called Versata. So that was our transition from desktop utilities all the way through to enterprise software used to build financial systems like loan origination software.”

Pryor confessed the transition was not initially an easy one, and underneath the superficial similarities between Australia and the U.S., there run important cultural differences: “I think

Matthew Pryor Using Tech to Manage Droughts, from Australia to CaliforniaDOI:10.1145/2856300

medical school and becoming a doctor. He bemusedly recalled his father’s initial disappointment, but ultimately he encouraged his son to pursue whatever attracted him the most, and to be in control of his own destiny.

Pryor initially enrolled in a electronic engineering course at the Royal Melbourne Institute of Technology. “My career advisor in high school, in year 11, had suggested to me—this was in 1983—that there really wasn’t much of a future in computer programming. Because by the time I would have graduated, computers would be programming themselves, so there were much better prospects in electronic engineering. I kid you not, this was the actual career advice I was given in 1983,” he recalled. After his first year, he switched to software engineering and didn’t look back.

STARTING ONE’S FIRST COMPANYUpon graduating, Pryor started Xedoc. However, unlike the startup culture pervasive today, founding Xedoc was a mostly spontaneous gesture.

“During university, I had a part-time job at a desktop publishing bureau, which was starting to take off, and I had saved up and bought myself a first generation Macintosh in 1985—the first year that they came into the country. I got into the design and desktop publishing side, met a guy who was building a publishing desktop system, and together with a third guy, we got together and saw that there was a significant opportunity for offering commercial services around Mac programming in particular, because there simply weren’t any companies in Australia that had gone there. Notably, I think we produced the first secure electronic document shredder, imaginatively called ‘Shredder,’ which we sold to the first company to produce virus checking

Matthew Pryor is the CEO of Observant, the Australian company founded in 2003 with the goal of solving water management

problems for the agricultural sector in the 21st century. He started his entrepreneurial life in software engineering fresh out of college, and experienced the boom and burst of the dot-com bubble of the late ‘90s from a front row seat in Silicon Valley, before becoming a trailblazer in agricultural technology (or agtech). Matthew Pryor had many stories to share with XRDS.

OWNING ONE’S CHOICESLike many curious kids with a knack for all things geeky, Pryor’s affair with computer science started at home. “I guess I’ve always been a tinkerer. I would always grab transistor radios and pull them apart, and I had a number of those electronic kits that you would be given as birthday gifts. I always seemed to be given things like that,” he confessed. In high school he was inspired by a friend’s electronic enthusiast magazine to buy a game kit and to solder it up from scratch. Historically speaking, this was still in the early days of computing in Australia, when the community in Victoria was gathered around printed media and Dick Smith’s electronic components shop in Melbourne.

“It worked! And I had never soldered anything in my life before! Pretty much from that time I was hooked with this idea of starting from nothing and making something. Really, your imagination was the limiting factor of what you could do with it,” he enthusiastically remembered.

The experience contributed to Pryor’s decision to study computer science in university, which was a major U-turn from his earlier plans of going to

DEPARTMENT EDITOR, ADRIAN SCOICĂ

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 69

Australian tech companies. There have been intense efforts of replicating the success seen in Silicon Valley in other tech hubs around the world over the last decades, but with mixed results. The main reason according to Pryor, is the symbiotic relationships between startups and educational institutions, which is crucial to entrepreneurship, is not as mature elsewhere as it is in Silicon Valley. Australian businesses also tend to not have the same international ambitions as their American counterparts, at least not from the get-go. The concept of internships, for example, which made Silicon Valley tech companies famous on campuses around the world, has not taken root in Australia.

However, as communication and cross-border collaboration improve, the differences are slowly being iterated away. Pryor’s concluding remark was one of optimism. He expressed pride at his achievements and excitement about the future: “When I look at the market today, and I see people talking about the ‘Internet of Farming Things’ and agtech, I feel lucky to have been involved in this market since the earliest days, and that our vision was very much on the right track. Even if we’re not done, and in terms of market adoption we still have a long way to go, it’s clear we’re on the right track!”

Copyright held by Owner(s)/Author(s).

to site, it became obvious that cameras, remote water sensors, and remote pump start/stop controllers for the underground water wells were perfect applications there,” he explained.

Despite the wealth of opportunity, Pryor confessed Observant was also about a decade early in its business It was particularly hard in their early days to convince farmers to adopt their systems from the onset.

“Today, you look at what we do and you call it the Internet of Things, big data, analytics, or agtech. It’s a legitimate business category now, and it’s attracted billions of dollars of investment in 2014,” he pointed out. Pryor explained when Observant started in 2003, they were mostly in uncharted territory, and the market was entirely different. “Agriculture is a slow adopter of new technology, not because farmers are non-innovative, but because they are very intolerant unreliable products: They can’t control the weather, and they can’t control market prices; so for the things they can control, they want absolute reliability. But once a product has proven to be reliable, then they very aggressively implement it.”

Deploying electronics in the outback taught Observant many hard lessons. “It’s hard to imagine a worse place to deploy tech products, maybe with the exception of a space program. It’s hot, it’s dusty, it’s far from people, it’s far from electricity and communications towers, and it has temperature extremes from the low negatives to the high positives. But coincident with our development as a company, Australia was going through what was referred to as our millennium drought,” he explained, adding the drought catalyzed their motivation.

Fast forward a few years, the experience they gained during the Australian drought found immediate application to California’s drought, now approaching its sixth year. As a result, Observant expanded and established operations in Sacramento, CA in 2013, historically coming full circle.

PUTTING AUSTRALIAN TECH ON THE WORLD MAPDespite success stories like theirs, there aren’t many internationally renowned

a lot of Australians—and this might be true for other English-speaking countries as well—can be lulled into a sense of thinking that doing business in the U.S. is the same as doing business in other places. But in reality in the U.S., and in particular in Silicon Valley, you’re dealing with the pinnacle of the art when it comes to the tech business.”

When I asked him to elaborate on the cultural differences he encountered, he explained the business side dominates in Silicon Valley: “It’s not that it’s malicious, or lacking in moral standard, it’s just that it’s 110 percent business the whole time. I think the most important cultural difference is in the U.S. you are brought up to go after—and get—what you want. Over there, there’s just more an innate sense that negotiating and working out how to move circumstances into your advantage is a skill that kids learn from the earliest age.”

Following the burst of the dot-com bubble, and having spent a good amount of time in the enterprise software business, in 2002 Pryor decided it was time to bank on his experience, leave the Bay Area, and return home to Australia with his family. He took some months off in order to get some breathing space, and started looking around for a new project.

AGRICULTURE MEETS TECHDuring his months off, Pryor met with a friend whose family had big agricultural holdings in the remote, dry regions of Australia. Pryor couldn’t believe how little tech was employed in agriculture. The more involved he got, the more he realized there was a huge opportunity for technology to play a transformative role in driving agricultural production. His ideas led to the founding of Observant, Pryor’s latest company.

Observant’s initial product was centered on water management for cattle grazing operations in arid regions. “These cattle stations are very large, the largest ones we worked on were around 3,500 km2 in size, so keeping water up to the cattle was a significant challenge, and a significant occupational risk both for the staff and the livestock. Rather than having people clock up thousands of kilometers a week driving from site

70 X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 270

end

“Beyond the Internet of Things.”At the SWARM lab, a group with

diverse competencies, ranging from mathematics to computer science, work together to develop novel services for the Internet of Everything (IoE). IoE is an extension of the IoT concept, in

which people, processes, objects, and data are strongly interconnected.

The main activity of the SWARM JOL is, therefore, the study of IoE services that leverage the interactions between people and a multitude of things. Such a swarm of objects and data aims to enhance the competitive assets made available by TIM.

My particular research interests lie in the application of machine learning and artificial intelligence techniques to IoE. I am currently working on a smart notification system. As detailed by Martin Pielot et al. in their 2014 Mo-bileHCI paper on the effect of notifica-tions on users, the number of notifica-tions received nowadays is high and the tolerance to them is typically cor-related with negative emotions, such as stress and feeling overwhelmed. In

T IM, one of the most important telecommunication compa-nies in Italy, has been investing in research and development

for more than 50 years. In 2013, the company started a series of research and innovation laboratories in col-laboration with major Italian universi-ties. The goal of these Joint Open Labs (JOLs) is to encourage a generation of new ideas in an interdisciplinary set-ting.

At Politecnico di Torino in Turin, Italy, TIM launched four JOLs cover-ing mobile systems, service robotics, augmented reality applications, and the Internet of Things (IoT). I am a first year Ph.D. student in the university’s computer and control engineering pro-gram. I am involved with and funded by the SWARM JOL, whose tagline is

SWARM Joint Open Lab Politecnico di Torino, Italy

LABZ

A typical day in the JOL open space at Politecnico di Torino, Italy.

Collaboration among all the members of the SWARM lab is encouraged as a way to enhance progress on the individual research topics.

71X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 71

BACK

IP AddressesThe Internet of Things (IoT) is a big idea and comes with some big challenges. In order to build such an enormous network of sensors and electronic devices that can communicate, the devices need some way to identify and contact each other in the first place. While there are certainly many possible approaches to this issue, in one imagining the IoT devices could identify each other in the same way many devices already do, via Internet protocol (IP) addresses. However, a massive network necessitates a massive number of addresses.

The common IP address scheme in use today is IP version 4 (IPv4), which defines an address as a 32-bit number, allowing for a maximum of 232 unique public addresses. While that may seem huge, due to reserved addresses, inefficiencies of allocation, and the proliferation of devices on the Internet, the number of available addresses is quickly being exhausted. In fact, in February 2011, the Internet Assigned Numbers Authority (IANA) issued the last remaining blocks of addresses to the Regional Internet Registries (RIR). More recently ARIN, the RIR that allocates addresses in the U.S. and Canada, depleted its pool of addresses a few months ago in September 2015.

Fortunately, this exhaustion of IPv4 addresses has long been anticipated, and the replacement scheme, IPv6, defines an address as a 128-bit number. Still, despite being standardized in 1998, offering an enormous address space, simplified routing, and a number of other advantages, IPv6 has seen very slow deployment. The necessary transition to IPv6 will likely to continue slowly and painfully in the coming years, but the potential benefits are well worth the price.

—Finn Kuusisto

IPv4 IPv6Definition Published 1981 1998

Public Addresses 232 (4.3x109) 2128 (3.4x1038)

Address Format Dot-decimal of four octets(93.184.216.34)

Colon-separated hexadecimal(2606:2800:220:1:248:1893:25c8:1946)

Base Header Size 160 bits 320 bits

Last Addresses Issued 2011 -

93.184.216.34

64.238.147.58

64.238.147.76

192.149.252.124

4.31.198.44

192.0.43.8

the IoE domain, with more devices re-quiring the user’s attention, the num-ber of such notifications is expected to grow. My current work is focused on the study of a system that uses machine learning to manage incom-ing notifications according to context awareness and user habits. Such a system will be able to deliver the right notification at the proper time, at the most appropriate place, and to the most adequate person.

The SWARM lab hosts many other activities besides my project. Other Ph.D. students, for example, are work-ing within a European project named INTrEPID1 on the application of IoT devices toward the realization of “smarter” homes. INTrEPID aims to proactively involve users in the energy management of their smart home: The project’s intelligent system can notify the user of behaviors that reduce en-ergy consumption and save money.

The system recognizes user hab-its through the available IoT devices, whether they are in the home or worn by the user, and suggests the most economical moment for carrying out typical activities, like switching on a washing machine. Moreover, the sug-gestions for best practices are custom-ized and only change gradually with time, so the user does not resist adopt-ing them.

Both projects are interconnected by the notifications concept; collabo-ration among all the members of the SWARM lab is encouraged as a way to enhance progress on the individual re-search topics.

For more information, visit the SWARM JOL website: http://jol.tele-comitalia.com/jolswarm/

Biography

Teodoro Montanaro is a Ph.D. student at the Politecnico di Torino. His interests lie in the application of machine learning and artificial intelligence techniques to the Internet of Things.

1 http://www.fp7-intrepid.eu/

IoT enabled driverless cars will gener-ate $1.3 trillion in annual savings in the United States, with more than $5.6 trillions of savings worldwide.

72 X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 272

According to Sensor 22, Benny is Preparing DinnerBY LARA ZUPAN AND MARINKA ZITNIK

HELLO WORLD

O ne winter morning in the near future…

6:30 a.m. A switch sensor in the bed is deactivated.

Pressure mat sensors placed under the bedroom carpet are activated.

6:35 a.m. A sensor in the hall-bedroom door switches to “on.” A sensor in the hall-toilet door switches to “on.”

6:37 a.m. A flush is activated. Capacitance sensors detect an object placed beneath the spout.

6:40 a.m. A sensor in the hall-toilet door switches to “off.” A sensor in the hall-kitchen door switches to “on.”

6:43 a.m. The fridge is opened. An object is removed from the cupboard. Tea is taken down from the shelf. Taps on the sink are turned on. Liquid is poured from the kettle.

Based solely on the pattern of sensor activation you might have inferred the above text most likely describes a morning routine. Your guess was possible because sensors can provide powerful clues about our activity. For example, an activated switch sensor in the bed strongly suggests sleeping, and pressure mat sensors can detect our movement and position. Watching for patterns in how people interact with household devices can therefore help us to recognize various activities that take place in home environments [1].

MEET BENNY Let’s consider data from a sensor system installed in a three-room apartment of a 26 year-old male, who we will call “Benny” [2]. Examples of some of the 14 sensors that were placed in Benny’s home include sensors taped to the groceries cupboard, microwave, and toilet. Data were collected over a 25-day period during which Benny recorded start and end points of his daily activities, such as taking showers and preparing meals.

Figure 1: Shown are the activities that were annotated in the human activity recognition data set [2]. The histogram shows the number of times the activity occurs in the data set. Shown is the output of Distributions widget produced by the Orange workflow shown in Figure 3.

Figure 2: State-change sensors were installed in a home environment. Shown is an activation pattern of a sensor that detected if the fridge was opened (“>= 0” stands for activation, “< 0” indicates no sensor activity). As expected, the sensor in the fridge most often fired when Benny was getting a drink or preparing meals and was inactive when Benny was using the toilet.

73X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2 73

speculate Benny might suffer from a sleeping disorder. Another explanation might be noisy and ambiguous information from multiple sensors; for example, the sensors might have been dislodged or failed.

Next, we would like to get insight

For our purposes, we represent the sensor data using a large number of features. We generate one feature vector per activity by calculating the features from the start to the end time of each activity label. Time slices for which no annotations were available are collected into a separate activity labeled as “unknown.”

We consider two types of data features. The first type reports whether activation of a particular sensor existed during some time period. The second type is whether a particular sensor fired before another particular sensor. The number of “existed” features is equal to the number of sensors present in the system (14) and the number of “before” features is equal to the number of all combinations of sensors and locations in the home environment (14*13/2=91 and 7*6/2=21, respectively). This way, we incorporate temporal and contextual information about sensors. In total, our data set contains 431 sensor activation profiles, where each profile corresponds to a vector of 126 features.

We will explore this data set to see whether it contains any discriminative patterns of sensor activity. We will then consider several prediction models to recognize Benny’s activities.

HAVE AN ORANGE FOR BREAKFASTTo analyze the data we use Orange (http://orange.biolab.si), an open-source data mining and visualization tool [3]. Orange provides a large number of widgets, components for machine learning, which we organize in workflows to perform end-to-end data analysis. Orange is especially appealing because it allows us to interact with our data and design data mining workflows in an intuitive way. Let’s start!

Figure 1 shows our sensor data records eight different activities, whose number of occurrence in the data set varies a lot. For example, there are 114 sensor profiles when Benny used the toilet, which is substantially more than the number of times he prepared dinner. Properly addressing skewed data distributions remains challenging when mining sensor data about human activities. We would expect the number of times Benny prepared dinner would be

approximately the same as the number of times he went to bed. As we are all well aware, a person typically prepares dinner only once per day and goes to sleep the same amount of time—once. A high number of recorded sleeping activities (see Figure 1) may lead us to

Figure 3: This Orange workflow loads the human activity data set, visualizes distributions of various activities and sensor activations, and computes pairwise distances between feature vectors in the data set. The workflow also runs the multidimensional scaling algorithm (MDS) and performs hierarchical clustering of sensor data to discover activity patterns.

Figure 4: This Orange workflow focuses on accurate detection of human activities. The workflow evaluates the ability of various classification algorithms, such as logistic regression, random forest and support vector machines, to predict human activities from sensor activation profiles. Predictive performance is assessed using 10-fold cross-validation and receiver operating characteristic (ROC) analysis.

74 X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 274

prediction algorithms may have trouble differentiating between the two, thus often mistaking one for another.

NO NEED TO CALL 911, SENSOR 42 HAS ITThe Internet of Things may provide extraordinary benefits to how well Benny will live [5]. Already today many everyday devices ring, glow, and buzz to sense Benny’s movements, habits, or, for example, to remind him to take his medications. Now imagine a world where everything around Benny is “smart.” Devices that previously made Benny’s morning coffee and kept his food cold will ensure he eats and sleeps well. The refrigerator will track changes in Benny’s diet while his coffee maker may signal a change in his sleeping habits and morning routine. Services connected to health centers will provide additional eyes, ears, and even a helping hand when Benny needs help at a distance from his family and friends or when he grows old. What a future!

References

[1] Kim, E., Helal, S., and Cook, D. Human activity recognition and pattern discovery. Pervasive Computing 9, 1 (2010), 48-53.

[2] Van Kasteren, T. L. M., Englebienne, G., and Kröse, B. J. Transferring knowledge of activity recognition across sensor networks. In Proceedings of the Eighth International Conference on Pervasive Computing. 2010, 283-300.

[3] Demšar, J., et al. Orange: data mining toolbox in Python. The Journal of Machine Learning Research 14, 1 (2013), 2349-2353.

[4] Fernández-Delgado, M., et al. Do we need hundreds of classifiers to solve real world classification problems? The Journal of Machine Learning Research 15, 1 (2014), 3133-3181.

[5] Gershman, S. J., Horvitz, E. J., and Tenenbaum, J. B. Computational rationality: A converging paradigm for intelligence in brains, minds, and machines. Science 349, 6245(2015), 273-278.

Supplementary Materials

http://github.com/acmxrds/winter-2015

Biographies

Lara Zupan is a high school junior attending Gimnazija Vic in Ljubljana, Slovenia. She has edited several high-school publications and designed Novi Dijak , a student run magazine targeting Slovenian high-school students. She is also a member of the European Youth Parliament. She is interested in graphics design, literature, and visual arts.

Marinka Zitnik is a Ph.D. student in computer science at the University of Ljubljana. She has also done research at the University of Toronto, Imperial College London, Baylor College of Medicine, and Stanford University. Her interests include machine learning, probabilistic numerics, and bioinformatics.

Copyright held by Owner(s)/Author(s).

into the circumstances under which the sensors in Benny’s apartment were fired. The sensor installed in the fridge was most active during the times Benny was getting a drink or preparing meals (see Figure 2). As expected, kitchen sensors were not active when Benny was brushing teeth or using the toilet. We observed similar patterns of sensor activity for other sensors installed in Benny’s apartment.

So far we have seen our sensor features show discriminative power with regard to different activities. The features successfully distinguish between activities that have little in common, such as making dinner and brushing teeth, which take place in different locations (e.g., bathroom) and involve various types of objects (e.g., cabinet). To further examine whether vectors describing the same activity are indeed more similar than vectors belonging to different activities, we performed hierarchical clustering (see Figure 3). We found clusters of sensor profiles corresponded nicely to different activity types. These results show the potential to predict Benny’s activities solely based on noisy sensor information, which we will explore next.

TO SLEEP OR NOT TO SLEEPWe didn’t do this analysis just for the fun of it. Our actual goal is to predict human behavior by referring to the information sensed by the sensors. To make accurate predictions we rely on various algorithms, such as naive Bayes, logistic regression, support vector machines, and random forest [4]. Using a cross-validation procedure, we generally observed that more sophisticated algorithms performed better than simpler ones (see Figure 4). The best performer was logistic regression (AUC=0.810, 10-fold cross-validation), while the worst performer, as expected, was a majority learner.

However, there were instances when any and all of the considered algorithms made incorrect predictions. This happened most often when sensor activations in question described similar activities. For example, the data that indicate Benny making breakfast resembles the data saying he’s preparing dinner. This is why

ACM’s Interactions magazine explores critical relationships between people and technology, showcasing emerging innovations and industry leaders from around the world across important applications of design thinking and the broadening � eld of interaction design.

Our readers represent a growing community of practice that is of increasing and vital global importance.

To learn more about us, visit our award-winning websitehttp://interactions.acm.org

Follow us on Facebook and Twitter

To subscribe:http://www.acm.org/subscribe

INTER AC TION S

Association for Computing Machinery

IX_XRDS_ThirdVertical_V01.indd 1 3/18/15 3:35 PM

ACM Transactions on Parallel ComputingSolutions to Complex Issues in ParallelismEditor-in-Chief: Phillip B. Gibbons, Intel Labs, Pittsburgh, USA

For further information or to submit your manuscript, visit topc.acm.org

Subscribe at www.acm.org/subscribe

ACM Transactions on Parallel Computing (TOPC) is a forum for novel and innovative work on all aspects of parallel computing, including foundational and theoretical aspects, systems, languages, architectures, tools, and applications. It will address all classes of parallel-processing platforms including concurrent, multithreaded, multicore, accelerated, multiprocessor, clusters, and supercomputers.

Subject Areas

• Parallel Programming Languages and Models• Parallel System Software• Parallel Architectures• Parallel Algorithms and Theory• Parallel Applications• Tools for Parallel Computing

For further information or to submit your manuscript, visit topc.acm.org

Subscribe at www.acm.org/subscribe

ACM Transactions on Interactive

Intelligent Systems

ACM Transactions on Interactive Intelligent Systems (TIIS). This quarterly journal publishes papers on research encompassing the design, realization, or evaluation of interactive systems incorporating some form of machine intelligence.

World-Renowned Journals from ACM ACM publishes over 50 magazines and journals that cover an array of established as well as emerging areas of the computing field.

IT professionals worldwide depend on ACM's publications to keep them abreast of the latest technological developments and industry news in a timely, comprehensive manner of the highest quality and integrity. For a complete listing of ACM's leading magazines & journals,

including our renowned Transaction Series, please visit the ACM publications homepage: www.acm.org/pubs.

PLEASE CONTACT ACM MEMBER SERVICES TO PLACE AN ORDERPhone: 1.800.342.6626 (U.S. and Canada) +1.212.626.0500 (Global)Fax: +1.212.944.1318 (Hours: 8:30am–4:30pm, Eastern Time)Email: acmhelp@acm.orgMail: ACM Member Services General Post Offi ce PO Box 30777 New York, NY 10087-0777 USA

ACM Transactions on Computation Theory (ToCT). This quarterly peer-reviewed journal has an emphasis on computational complexity, foun-dations of cryptography and other computation-based topics in theo-retical computer science.

ACM Transactions on Computation

Theory

www.acm.org/pubs

PUBS_halfpage_Ad.indd 1 6/7/12 11:38 AM

76 X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

POINTERSACRONYMS

AIM Automatic Identification and Mobility: A diverse family of technologies that share the common purpose of identifying, tracking, recording, storing, and communicating essential business, personal, or product data. In most cases, AIM technologies serve as the front end of enterprise software systems, providing fast and accurate collection and entry of data.

BL Business Logic: The goal or behavior of a system involving “things” serving a particular business purpose. It can define the behavior of a single thing, a group of things, or a complete business process

M2M Machine to Machine: Wired and wireless devices communicate with other devices of the same type. With a wide range of applications such as industrial automation, logistics, smart grid, smart cities, health, defense etc. M2M brings several benefits to the Internet of Things.

NGN Next-Generation Networks: A packet-based network able to provide telecommunication services and make use of multiple broadband, QoS-enabled transport technologies, in which service-related functions are independent from underlying transport-related technologies.

RFID Radio Frequency Identifier: The use of electromagnetic or inductive coupling in the radio frequency portion of the spectrum to communicate to, or from, a tag through a variety of modulation and encoding schemes to uniquely read the identity of a radio frequency tag.

THE INTERNET OF THINGS

The Internet of Things (IoT) refers to the concept of connecting any device with an on and off switch to the Internet (and/or to each other) leading to a giant network of connected “things” (which also includes people), redefining relationships between people-people, people-things, and things-things. British entrepreneur Kevin Ashton coined the term in 1999, although the first Internet-connected toaster was unveiled at the Interop conference in 1989.1 —Tejas S. Khot

GETTING STARTED

Coursera is offering a specialization track for building devices that can control the physical world. The course covers topics like embedded systems, the Raspberry Pi Platform, and the Arduino environment. The final capstone project lets you apply the skills you learned by designing, building, and testing a microcontroller-based embedded system. This introduction to programming the IoT takes place over five courses and cost $384 USD to enroll.https://www.coursera.org/specializations/iot

“67 Open Source Tools and Resources for IoT” by Scott Amyx With growing support from industry, open-source solutions are becoming increasingly popular, accelerating the innovation and adoption of new technologies. Amyx has compiled a useful list of open-source tools and resources for the IoT, ranging from drivers and middleware to operating systems, APIs, and protocols.http://techbeacon.com/67-open-source-tools-resources-iot

1 http://www.livinginternet.com/i/ia_myths_toast.htm

Interacting in the IoT The IoT has certainly been the subject of plenty of hype, but its potential is real. Less obvious is the requirement for how people interact with connected objects. Many connected things will communicate directly with each other, but at some point, people will need to interact with many of them. For this, there has to be a user interface, most likely in the form of an app. But app-based user interfaces pose several problems. Galen Gruman and Justin Zalewski share guidelines for designing elegant user interfaces in the IoT. http://www.infoworld.com/article/2867356/internet-of-things/beware-this-iot-fallacy-the-headless-device.html http://studioscience.com/interaction-design-within-the-internet-of-things/

IOT RESEARCH

Rethinking a Secure Internet of Things Existing approaches to secure computing systems are insufficient for these new cyber-physical applications, as they have very different trust models and network architectures, bridging pervasive local area networks, personal mobile devices, server storage, and web-based applications. The Secure Internet of Things Project (SITP) is a five-year collaboration between Stanford University, the University of California Berkeley, and the University of Michigan to research fundamentally new and better ways to secure the IoT. http://iot.stanford.edu/

77X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

Connecting the World with Ant-Sized Radios “A Stanford engineering team, in collaboration with researchers from the University of California, Berkeley, has built a radio the size of an ant, a device so energy efficient that it gathers all the power it needs from the same electromagnetic waves that carry signals to its receiving antenna—no batteries required.” The three-year project involved re-engineering not only the components, but also the functions of a radio. http://news.stanford.edu/news/2014/september/ant-radio-arbabian-090914.html

“Computations on the Edge in the Internet of Things” by Andreas Moregård Haubenwallera and Konstantinos Vandikas How does one make sensor data processable and searchable in a human-friendly way? Ericsson Research in collaboration with the Swedish Institute of Computer Science and Uppsala University developed an open-source system called the IoT Framework, released under the Apache 2.0 license. The IoT Framework is a computational engine for quantitative information accumulated by sensors connected via any IP network such as the Internet. The paper discusses details of their approach.http://www.sciencedirect.com/science/article/pii/S187705091500811X

FURTHER READING

The New Hardware Movement Over the past decade or so, hardware development started looking much more like software development—enabling faster and cheaper deployment in ways not known earlier. This has taken computing to newer places, leading to a movement now known as the IoT. The new hardware movement is a primary enabler of the Maker movement, promoting prototyping as a hobby. Jon Bruner, the director of hardware and IoT at O’Reilly Media, shares an overview of how these two transformative ideas correlate.http://radar.oreilly.com/2015/06/why-the-internet-of-things-isnt-the-same-as-the-new-hardware-movement.html

“When Fridges Attack: The new ethics of the Internet of Things” The prevalence of connected devices and flowing data also raises ethical and legal questions. Now that we know we are being monitored in a variety of ways, both voluntarily and involuntarily, is the IoT fostering a utilitarian approach to society? Utilitarianism is the ethical construct based on outcomes of maximum utility. Dr. Peter McOwan, professor of computer science at Queen Mary, University of London, answered some of these queries in a speech given at the 2014 British Science Festival.http://www.theguardian.com/science/alexs-adventures-in-numberland/2014/sep/08/when-fridges-attack-the-new-ethics-of-the-internet-of-things

CONFERENCES

IEEE World Forum on Internet of Things (WF–IOT 2015)University of MilanMilan, ItalyDecember 14–16, 2015http://www.ieee-wf-iot.org/

NDC LondonICC Suites ExCeL ArenaLondon, United KingdomJanuary 13–15, 2016http://ndc-london.com/

International Conference on Internet of Things and Applications (IOTA 2016)Maharashtra Institute of TechnologyPune, Maharashtra, IndiaJanuary 22–24, 2016http://www.iota2016.org/

IoT Nexus USAHotel KabukiSan Francisco, CAFebruary 10–11, 2016http://iot-nexus.com/usa/

IoT Tech Expo Europe 2016London OlympiaLondon, United KingdomFebruary 10–11, 2016http://www.iottechexpo.com/europe/

International Conference on Embedded Wireless Systems and Networks (EWSN 2016)Graz University of TechnologyAustriaFebruary 15–17, 2016http://www.iti.tugraz.at/EWSN2016/

IOT ConferenceHoliday Inn, Munich City CentreMunich, GermanyMarch 14–17, 2016https://iotcon.de/

EVENTS

78 X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

Imag

e by

Ada

m C

ohn

/ Fl

ickr

Michelin Challenge DesignMichelin is hosting a design challenge to foster creative thinking in vehicle design. The theme is “Mobility For All–Designing for the Next Frontier.” Participants are encouraged to explore functional and affordable vehicle designs in emerging markets. Your submission should contain graphic renderings of the vehicle, tire, and wheel assembly as well as an essay. Applications are due by June 1, 2016.http://www.michelinchallengedesign.com/

SF Crime Classification ChallengeKaggle is hosting an online data-science competition in which participants are invited to predict the category of crimes that have occurred based on 12 years of San Francisco crime reports. Winners will receive swag prizes from Kaggle. The deadline for submitting your code and results is June 6, 2016.https://www.kaggle.com/c/sf-crime

ReactEuropeReact.js is an open-source library developed at Facebook for creating web user interfaces, which is growing in popularity. The ReactEurope developer conference will take place in Paris from June 2–3, 2016. Many leaders from the React community will be in attendance. The event features workshops, lightning talks, and a hackathon.https://www.react-europe.org/

Wearable Technology Show 2016ExCeL LondonLondon, United KingdomMarch 15–16, 2016http://www.wearabletechnologyshow.net/

IoT WorldPorte de VersaillesParis, FranceMarch 23–24, 2016http://www.iot-world.fr/

EAI International Conference on Industrial IoT Technologies and ApplicationsGhuangzhau, ChinaMarch 24–26, 2016http://industrialiot-conf.org/

IEEE International Conference on Internet-of-Things Design and Implementation (IoTDI 2016)Technical University of Berlin, Charlottenburg campusBerlin, GermanyApril 4–8, 2016http://conferences.computer.org/IoTDI/

STC 2016LIMICSParis, FranceApril 17–19, 2016http://www.stc2016.org/

CONTESTS & EVENTS

IoT World HackathonThe Internet of Things World Hackathon is a 30-hour , app-building contest that will take place May 10–12 in San Francisco. Two challenges for participants include building the best consumer and industrial IoT project. Sponsors, which include Broadcom and Intel, will provide hardware for participants to use in their project, as well as prizes for the best use of hardware.http://iotworldevent.com/hackathon/

International Conference on the Internet of Things and Applications (IOTA 2016)

Maharashtra Institute of Technology Pune, Maharashtra, India January 22–24, 2016

The number of devices connected via the Internet is rapidly increasing. The Internet of Things (IoT) aims to enhance each and every part of our lives by integrating itself with everything around us. Be it communications, energy, healthcare, business, or government, the IoT paradigm isn’t keeping to itself. With most of the world’s governments supporting digitization, the effects of the IoT will soon transform our daily lives.

This conference aims to bring together distinguished researchers from academia and industry to showcase their research in IoT, sensor networks, hardware, and software. Discussions on various challenges and solutions to the practical realization of the IoT for use in our daily lives are on the agenda. This conference also features workshops, tutorials, and poster paper presentations.

The conference is being held in Pune, India. The city is referred to as the “Oxford of the East,” and offers various cultural attractions. Among numerous historical structures around the city, monuments like the Aga Khan Palace and the Sinhagad Fort are a must see for everyone attending.

For more information, please visit http://www.iota2016.org/

—Darshit Patel

FEATURED EVENT

79X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 2

Imag

e by

Nic

o El

Nin

o /

Shu

tter

stoc

k.co

m

GRANTS AND SCHOLARSHIPS

Leeds Anniversary Research Scholarship: Big Data and Internet of Things Architecture and ProtocolsWebsite: https://www.engineering.leeds.ac.uk/faculty/postgraduate/research-degrees/scholarships/big-data-internet-things.shtmlDeadline: January 15, 2016Eligibility: Honors degree (UK equivalent of 2.1) in electronic/electrical engineering, physics, or mathematics.Benefits: £14,057 Explanation: The scholarship is aimed to develop network architectures and protocols that provide resilience, scalability and low-power consumption in future big data and IoT networks. The Ph.D. scholarship provides an opportunity to work within the Communication Networks and Systems group in the School of Electronic and Electrical Engineering at the University of Leeds.

American Speech Language Hearing Foundation: New Century Scholars Research GrantWebsite: http://www.ashfoundation.org/grants/ Deadline: To be announced early 2016. Eligibility: Applicants must be committed to teacher-investigator careers in academic environments or in external research institutes.Benefits: $25,000Explanation: The program funds researchers doing ground-breaking investigations in the field of communication sciences and disorder for up to two years.

The Google Anita Borg Memorial Scholarship: EMEAWebsite: http://www.google.com/anitaborg/emea/Deadline: To be announced early 2016 Eligibility: Applicants must be female university students studying in any computer science discipline at a

university in Europe, the Middle East, or Africa.Benefits: €7,000Explanation: The program funds female students to pursue advanced studies in computing.

National Science Foundation Scholarships in Science, Technology, Engineering, and Mathematics Program (S-STEM)Website: http://www.nsf.gov/funding/pgm_summ.jsp?pims_id=5257Eligibility: Students who are awarded S-STEM scholarships must be U.S. citizens, permanent residents, nationals, or refugees.Deadline: May 16, 2016Benefits: $10,000–$40,000Explanation: The National Science Foundation awards institutions of higher education scholarships for low-income students at the undergraduate and graduate levels in any science, technology, engineering, or mathematics discipline. The scholarships can run for up to four years.

German Universities of Applied Sciences (UAS7): Study & Internship Program (SIP) in GermanyWebsite: http://www.uas7.org/scholarships/study-a-internship-program.htmlDeadline: February 15, 2016Eligibility: Undergraduate student at an accredited U.S. or Canadian university. Nationality is not a limiting factor. Proficiency in German language is not required.Benefits: €1,000 contribution for travel expenses, full tuition waiver, and other benefits.Explanation: Students have a unique opportunity to spend a full year studying and interning in German universities of applied sciences. Research internships are available in a wide range of fields, from aeronautical engineering to logistics and management to nanoscience.

STC 2016 Laboratoire d’Informatique Médicale et d’Ingénierie des Connaissances en e-Santé (LIMICS)

Paris, France April 17–19 2016

The IoT is near. Phones, wristwatches, clothing, and various other lifestyle gadgets are already connected, and will have a great impact on the health and well-being of users. We are witnessing a paradigm shift toward sustainable healthcare and social welfare systems. However, with increased presence also comes the question: are we safe? The IoT must also be in line with data confidentiality.

The European Federation for Medical Informatics (EFMI) and the French Association for Medical Informatics (AIM) have partnered to develop “Transforming Healthcare with the Internet of Things,” a special topic conference that will focus on various fundamental, applied, and industrial aspects of IoT in health. This event is bound to interest academia, industries, governments, and developers alike. There will be workshops, panels, demonstrations, and presentations on diverse issues affecting IoT in health, from healthcare in education to the use of lifestyle apps in everyday living.

Attendees can also treat themselves with a trip to the Louvre or the Eiffel Tower, both of which are located in the city of Paris.

For more information, please visit http://www.stc2016.org/

—Darshit Patel

FEATURED EVENT

X R D S • W I N T E R 2 0 1 5 • V O L . 2 2 • N O . 280

end

BEMUSEMENT

Treasure Maps

http

://t

hisi

sind

exed

.com

/wp-

cont

ent/

up

load

s/2

015/

10/

card

4719

.jpg)

The Guilt Timer

http

://w

ww

.phd

com

ics.

com

/syn

d/c/

phd1

00

715

vphe

.tif

Hardware Reductionism

http

://x

kcd.

com

/158

8/

SUBMIT A PUZZLE

Can you do better? Bemusements would like your puzzles and mathematical games (but not Sudoku). Contact xrds@acm.org to submit yours!

Puzzle: Door Number …?A criminal gets to pick his punishment by facing whatever is behind the door of his choosing. There are three rooms. The first is full of burning fires, the second is full of assassins with loaded guns, and the third is full of lions that haven’t eaten in three years. Which room is the safest choice?

Find the solution at: http://xrds.acm.org/bemusement/2015.cfm

Association for Computing Machinery

ThinkLoud

CONNECT WITH OURCOMMUNITY OF EXPERTS.

www.computingreviews.com

They'll help you find the best new books and articles in computing.

Computing Reviews is a collaboration between the ACM and ThinkLoud.

C

M

Y

CM

MY

CY

CMY

K

1974 Red Sphere Ad-F2.pdf 4/27/2012 8:51:11 AM

Search NSA to Download

CAREERS at the NATIONAL SECURITY AGENCY

WORKEXTRAORDINARYInside our walls, you will find the most extraordinary people doing the most extraordinary work. Yet it’s not just finite field theory, quantum computing or RF engineering. It’s not just discrete mathematics or graph analytics.

It’s all of these and more – rolled up into an organization that leads the world in signals intelligence and information assurance.Inside our walls you will find extraordinary people, doing extraordinary work, for an extraordinary cause:

The safety and security of the United States of America.

U.S. citizenship is required for all applicants. NSA is an Equal Opportunity Employer and abides by applicable employment laws and regulations. All applicants for employment are considered without regard to age, color, disability, genetic information, national origin, race, religion, sex, sexual orientation, marital status, or status as a parent.

Computer/Electrical EngineeringComputer ScienceCybersecurityInformation AssuranceMathematicsForeign LanguageIntelligence AnalysisCryptanalysisSignals AnalysisBusinessFinance & AccountingPaid Internships, Scholarships & Co-op

APPLY TODAY

WHERE INTELLIGENCE GOES TO WORK®

14CNS-06_8.25X10.875(live8x10.625).indd 1 10/16/15 10:46 AM