PENCIL/LIGHTBULB: ©ISTOCKPHOTO.
COM/HSTRONGART
LAPTOP: ©ISTOCKPHOTO.COM/FAD1986,
REQUIREMENTS
■ IEEE Student Member or Graduate Student
Member (graduate or undergraduate level)
for the time of service (two years).
■ Strong interest in the professional, social,
and technical concerns of students.
■ Desire to see those concerns addressed
in IEEE Potentials in an objective, timely
fashion.
BENEFITS
■ Communicate ideas to students.
■ Work with other IEEE members from
academia and industry.
■ Show prospective employers your
professional involvement.
RESPONSIBILITIES
■ Provide input and feedback on the
editorial content of IEEE Potentials.
■ Write an editorial for each issue.
■ Read and evaluate articles.
■ Serve on the IEEE Student Activities
Committee (SAC), which oversees the
student programs and IEEE Student
Branches established at leading universities
and colleges worldwide.
■ Attend two IEEE Potentials editorial board
meetings and two SAC meetings each year.
Digital Object Identifier 10.1109/MPOT.2018.2884178
Student Member or Graduate
Student Member to serve as the
Student Editor of IEEE Potentials.
If you’d like to apply, simply com-
plete and mail this application.
Be sure to include a copy of your
resume and your answers to the
essay questions. We look forward
to hearing from you.
ESSAY QUESTIONS
Please answer the following essay questions (in a separate document):
1) In 50 words or fewer, explain why you would like to be the next Student Editor
of IEEE Potentials.
2) As Student Editor, what would your goals and objectives be?
3) List three areas of major concern to student engineers and technology professionals.
In approximately 200 words, explore one area in detail.
4) List four editorial topics you would write about as IEEE Potentials Student Editor.
Mail or e-mail to:
IEEE Potentials, 445 Hoes Lane, Piscataway, NJ 08854 USA
E-mail: potentials@ieee.org
Name _______________________________________________________
Mailing address ________________________________________________
City ____________________________ State ________________ Zip _____
Country ______________________________________________________
Permanent address _____________________________________________
City ____________________________ State ________________ Zip _____
Country ______________________________________________________
Phone number _________________________________________________
E-mail _______________________________________________________
I am an undergrad ________________ or a grad _________________ student
Institution ____________________________________________________
Graduation date ________________________________________________
Don’t forget to submit this information and a resume with your application!
Apply by 24 December 2018
 The magazine for high-Tech innovaTors

January/February 2019
Vol. 38 No. 1

theme: smart Consumer

eleCtroniCs systems

6 Smart consumer electronics systems

saraju p. mohanty

8 Power supplies for consumer electronic devices

santanu mishra

14 S-CHIRP: Securing communications

in lightweight peer-to-peer networks in the IoT on the Cover:
Unboxing the technology
shaya Wolf, rafer cooley, and mike borowczak behind smart consumer
electronics systems.

20 A neighbor trust-based mechanism

boxes: ©istockphoto.com/bet_Noire
braiN: ©istockphoto.com/heNrik5000

to protect mobile networks

Gurveen Vaseer, Garima Ghai, Dhruva Ghai,
and pushpinder s. patheja

Departments
26 Proof of authentication:
IoT-friendly blockchains & Columns
Deepak puthal and saraju p. mohanty 3 editorial
4 the way ahead
5 gamesman solutions
30 Drones: Augmenting our quality of life
christos kyrkou, stelios timotheou, panayiotis kolios,
48 gamesman problems

theocharis theocharides, and christos panayiotou

37 A method for localizing the eye pupil

for point-of-gaze estimation
Narayan panigrahi, kiran Lavu, sai krishna Gorijala,
peter corcoran, and saraju p. mohanty
MIssIon statEMEnt: IEEE Potentials
is the magazine dedicated to undergraduate
and graduate students and young profes-
sionals. IEEE Potentials explores career
strategies, the latest in research, and im-
portant technical developments. Through

43 E-marketing via augmented reality: A case its articles, it also relates theories to prac-
tical applications, highlights technology’s

study in the tourism and hospitality industry global impact, and generates international
forums that foster the sharing of diverse
ideas about the profession.
Neda shabani, arslan munir, and azizul hassan

Digital Object Identifier 10.1109/MPOT.2018.2850458

IEEE PotEntIals Januar y/Februar y 2019 ■ 1

IEEE PoTEnTIAlS—The magazine for high-Tech innovaTors

Sandra “Candy” Robison, President, Murty Polavarapu, Member Sebastian Corrado, Region 9
EDIToRIAl boARD IEEE-USA Development scorrado@ieee.org
editor-in-chief Forrest D. Wright, President, Standards Sergio Benedetto, Rajesh Ingle, Region 10
Vaughan Clarkson, University of Assoc. Member-at-Large ingle.rb@gmail.com
Queensland Martin Bastiaans, Member & Jill Gostin, Member-at-Large
student editor Geographic Activities Regional student
Cristian Quintero, Universidad Susan “Kathy” Land, Technical Activities Representatives
Distrital Fracisco José de Caldas
ADvISoRy CoMMITTEE Kayla Ho, Region 1
Division Directors Vaughan Clarkson, Chair (Potentials EIC) kho02@nyit.edu
associate editors Renuka P. Jindal (I) Mary Ellen Randall (MGA Past Chair)
John Benedict Boggala, Amazon Jacob Culleny, Region 2
F. Don Tan (II) J. Patrick Donohoe (SAC Chair)
Raymond E. Floyd, jacobculleny@comcast.net
Vijay K. Bhargava (III) Cecelia Jankowski (MGA Managing
IEEE Life Senior Member Jillian Johnson, Region 3
Jennifer T. Bernhard (IV) Director)
Zhijia Huang, Bank of America jjohns81@cbu.edu
John W. Walz (V)
Christopher James, John Y. Hung (VI) Benjamin Strandskov, Region 4
University of Warwick Bruno C. Meyer (VII)
MgA STuDEnT ACTIvITIES stran1b@cmich.edu
Jay Merja, MUVR Technology Dejan S. Milojicic (VIII)
CoMMITTEE Jessica Teeslink, Region 5
Sharad Sinha, Nanyang Technological Alejandro “Alex” Acero (IX) Elisa Barney Smith, Chair jessica.teeslink@mines.sdsmt.edu
University, Singapore Toshio Fukuda (X) ebarneysmith@boisestate.edu Mariella Saviola, Region 6
corresponding editors Elizabeth Johnston, Vice Chair msaviola@sandiego.edu
Region Directors lise.johnston@ieee.org Mohammad Jamiul Alam, Region 7
Cátia Bandeiras, Instituto Superior
Babak Beheshti, Region 1 Pablo Herrero, Past Chair jmjalam@gmail.com
Técnico
Katherine J. Duncan, Region 2 pablo.herrero@ieee.org Ana Inacio, Region 8
Syrine Ferjaoui, National Engineering
Gregg L. Vaughn, Region 3 Preeti Bajaj, Branch Chapter inesinacio@ieee.org
School of Sousse
Bernard T. Sander, Region 4 Representative, preetib123@yahoo.com
Athanasios Kakarountas, University Cristian Quintero, Region 9
Robert C. Shapiro, Region 5 Robert Burke, Branch Chapter Student
of Thessaly cristianquintero@ieee.org
Sachin Seth, Texas Instruments Kathleen A. Kramer, Region 6 Representative, Pasan Pethiyagode, Region 10
Sri Niwas Singh, Indian Institute Maike Luiken, Region 7 robert.burke@ieee.org
Margaretha Eriksson, Region 8 pasan.uom@gmail.com
of Technology Kanpur Dinko Jakovljevic, Young Professionals
Teofilo Ramos, Region 9 Representative,
IEEE PERIoDICAlS Kukjin Chun, Region 10 jakovljevic.dinko@windowslive.com
MEMbER & gEogRAPHIC
MAgAzInES DEPARTMEnT Vaughan Clarkson, Potentials EIC
ACTIvITIES DEPARTMEnT
445 Hoes Lane, HEADquARTERS STAff Cecelia Jankowski, Managing Director
v.clarkson@ieee.org
Piscataway, NJ 08854 USA Stephen Welby, Executive Director Cristian Quintero, Potentials John Day, Director, Member Products
Craig Causer, Managing Editor Michael Forster, Publications Student Editor, qcristianesteban and Programs
Geraldine Krolin-Taylor, Senior Jamie Moesch, Educational @hotmail.com Lisa Delventhal, Manager, Student
Managing Editor Activities Youmna El-Bitar, MGA/SAC/SPAA Chair and Young Professional Programs
Janet Dudar, Senior Art Director Konstantinos Karachalios, Standards youmna.elbitar@gmail.com Christine Eldridge, Administrative
Gail A. Schnitzer, Associate Art Activities Robert Vice, IEEE USA SPAC Chair Assistant, Student Services
Director Cecelia Jankowski, Member & robert.vice@gmail.com Shareyna Scott, Student Branch
Theresa L. Smith, Production Geographic Activities Liz Burd, TAB Representative, Development Specialist
Coordinator Cherif Amirat, Chief Information lizburd@newcastle.edu.au Kristen Mahan, Program Specialist
Mark David, Director, Business Officer Prasanth Mohan, IEEEXtreme Project Young Professionals
Development—Media & Advertising Donna Hourican, Staff Executive, Kelly Werth, Program Specialist
Lead, prasanthemy@gmail.com
+1 732 465 6473 Corporate Activities Student Activities
Felicia Spagnoli, Advertising Thomas Siegert, Business
Production Manager Administration & Chief Financial REgIonAl STuDEnT
Peter M. Tuohy, Production Director Officer ACTIvITIES CoMMITTEE CHAIRS IEEE HKn REPRESEnTATIvE
Kevin Lisankie, Editorial Services Director Charles Rubenstein, Region 1 Kathleen Lewis
Karen Hawkins, Chief Marketing
Dawn M. Melley, Staff Director, c.rubenstein@ieee.org kmlewis@mit.edu
Officer
Publishing Operations Mary Ward-Callan, Technical Activities Drew Lowery, Region 2
Chris Brantley, IEEE-USA dlowery@gmail.com InDuSTRy REPRESEnTATIvES
Victor Basantes, Region 3 R. Barnett Adler
IEEE boARD of IEEE MEMbER & gEogRAPHIC victor_basantes@hotmail.com b.adler@ieee.org
DIRECToRS ACTIvITIES boARD Nevrus Kaja, Region 4 Peter T. Mauzey
José M.F. Moura, President and CEO Martin Bastiaans, Chair nkaja@umich.edu p.mauzey@ieee.org
Toshio Fukuda, President-Elect Francis Grosz, Chair-Elect Anthony (Tony) Maciejewski, Region 5 Prijoe Philips Komattu
James A. Jeffries, Past President Mary Ellen Randall, Past Chair aam@colostate.edu prijoe.philips@gmail.com
William P. Walsh, Secretary Deborah Cooper, Treasurer Elizabeth Johnston, Region 6 John Paserba
Joseph V. Lillie, Treasurer Cecelia Jankowski, Secretary lise.johnston@ieee.org John.Paserba@meppi.com
Theodore W. Hissey, Director Emeritus Ron Jensen, Geographic Unit Mahsa Kiani, Region 7 Gowtham Prasad
Vice Presidents Operations mahsa.kiani@gmail.com smartgowtham@gmail.com
Witold M. Kinsner, Educational Activities Michael Lamoreux, Information Efthymia Arvaniti, Region 8 Robert Vice
Samir M. El-Ghazaly, Pub. Services & Prod. Management earvaniti@ieee.org robert.vice@gmail.com

IEEE prohibits discrimination, harassment, and bullying. Promoting Sustainable Forestry

For more information, visit http://www.ieee.org/web/aboutus/whatis/policies/p9-26.html. SFI-01681

IEEE Potentials (issN 0278-6648) (ieptDF) is published bimonthly by the institute of
electrical and electronics engineers, inc. Headquarters address: 3 park avenue, 17th
Floor, New York, NY 10016-5997. phone: +1 212 705 7900. Change of address must be
received by the first of a month to be effective for the following issue. please send to ieee
operations center, 445 hoes Lane, piscataway, NJ 08854. annual subscription, for ieee
student members, first subscription Us$5 included in dues for U.s. and canadian student
members (optional for other student members). prices for members, nonmembers, and
additional member subscriptions are available upon request. Editorial correspondence
should be addressed to ieee potentials, 445 hoes Lane, piscataway, NJ 08854. respon-
sibility for contents of papers published rests upon authors, and not the ieee or its mem-
bers. Unless otherwise specified, the ieee neither endorses or sanctions any positions
or actions espoused in IEEE Potentials. all republication rights including translations are
reserved by the ieee. Copyright and Reprint Permissions: abstracting is permitted with
credit to the source. Libraries are permitted to photocopy beyond the limits of U.s. copyright
law, for private use of patrons, articles that carry a code at the bottom of the first page, pro-
vided the per-copy fee indicated in the code is paid through the copyright clearance cen-
ter, 222 rosewood Drive, Danvers, ma 01923. For other copying, reprint, or republication
permission, write to IEEE Potentials at piscataway, NJ. all rights reserved. copyright © 2019
by the institute of electrical and electronics engineers, inc. printed in U.s.a. subscrip-
tions, orders, address changes: ieee operations center, 445 hoes Lane, piscataway,
NJ 08854, phone: +1 732 981 0060. other publications: ieee also publishes more than 30
specialized publications. advertising Representative: ieee potentials, 445 hoes Lane,
piscataway, NJ 08854, phone: +1 732 562 3946. IEEE Departments: ieee operations
center (for orders, subscriptions, address changes, and educational/technical/ standards/
publishing/regional/section/branch services) 445 hoes Lane, piscataway, NJ 08854,
Usa. operations center +1 732 981 0060; Washington office/professional services +1 202
785 0017. headquarters: telecopier +1 212 752 4929, telex 236-411.
periodicals postage paid at New York, NY, and at additional mailing offices. Postmaster:
send address changes to ieee potentials, ieee, 445 hoes Lane, piscataway, NJ 08854,
Usa. canadian publications agreement Number 40030962. return Undeliverable cana-
dian addresses to: Fort erie, oN L2a 6c7 canada. canadian Gts #125634188.
priNteD iN the U.s.a.
Digital Object Identifier 10.1109/MPOT.2018.2850383

2 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals


The Future Is Now
I
remember those childhood days during my vacations
from school, when my mother would not let me go out
to play with my friends if I did not first help with
some of the housekeeping tasks. Sweeping and mop­
ping were the things that took most of my time, and I
had to do them at least twice every day. In those days, I
just wanted to complete my tasks as soon as possible
so I could go out and play with my
friends. This is not as much of a
problem for the kids of today be­
cause they have vacuum cleaner
robots to do the work. discover what is coming in
I was born in 1996, a year when
16 MB of random access memory
was a big deal and Wi­Fi didn’t
exist; neither did Google and Net­
flix, companies that were founded
some years after I was born. The
world has changed significantly
in the last 20 years, which is nothing compared with the
time the last couple of previous revolutions took.
These past 20 years have been full of releases of dra­
matic inventions conceived to make our lives easier, such
as affordable personal computers for work, portable mu­
sic players for fun, digital cameras to take more photos,
and, most recently, smartphones that are able to do all of
these things and more. This is just the beginning of an
entire revolution, including voice commands to activate
Digital Object Identifier 10.1109/MPOT.2018.2876533
Date of publication: 28 December 2018
editorial
by Cristian Quintero
mechanisms and personal assistants ready to arrange
our schedules.
In this issue of IEEE Potentials, you will discover what
is coming in the world of smart consumer electronics de­
signed to make our lives easier. You can see the work of
hundreds of engineers who are thinking about what we
will need in the next couple of years. This may be one of
our most exciting issues because
you can find what is on the horizon
In this issue of IEEE in terms of new technology.
Potentials, you will In a couple of months, I will leave
my parents’ house and carry out all
of the necessary household chores
the world of smart in my own apartment, but you know
consumer electronics what? I am not worried at all. I have
designed to make our my vacuum cleaner robot so I won’t
have to sweep or mop, I own Internet
lives easier. Protocol cameras so wherever I am I
can check that everything is OK in
my apartment, and I have my Google Home to ask for any­
thing I need, whenever I need it. Of course, I will have to do
some things, like cook, but this is fine, at least for now.
What do you think? How much time will it take us until
we live like they do in the animated sitcom The Jetsons and
be able to have our breakfast just by pushing a couple of
buttons? How much time will pass until we have a cool as­
sistant like J.A.R.V.I.S. from Iron Man?
about the author
Cristian Quintero (cristianquintero@ieee.org) is the stu­
dent editor of IEEE Potentials.
IEEE PotEntIals Januar y/Februar y 2019 ■ 3
the way ahead

Opportunities for
Student Members
by Elisa H. Barney Smith

W
elcome 2019! What is your New Year’s reso-
lution? Or maybe a better way of looking at
it is not through a single resolution, but
what opportunities lie ahead for 2019, and
which will be your points of focus? There are many
opportunities for IEEE Student Members. The IEEE is
rolling out some administrative changes. Officers
likely have heard of these changes from multiple
channels already. While they most directly impact
Student Branches and Section officers, the goal is to
open up opportunities to improve
the IEEE membership experience
for all Student Members.
Do you know how the IEEE is
structured? Your IEEE Student
Branch is more than a club; it is
part of the IEEE global structure.
more active, you will get
The IEEE is divided by technical and
geographical areas. Geographically,
there is a Section in which your uni-
versity resides (and all student lead-
professional development
ers should know what that Section
is). The Student Branch is officially a
subunit of that Section, just like sev-
eral other units, such as technical
professional IEEE Members
chapters and affinity groups (e.g.,
Women in Engineering and Young
Professionals). Sections are now required to have a Section
Student Representative as a voting member of its Executive
Committee. Students continue to be invited to attend Sec-
tion planning meetings.
Have you heard of vTools? As Student Members, you
should have received invitations from your Section to at-
tend meetings and events scheduled and targeted toward
the professional IEEE Members in your Section. These
are organized through a tool called vTools (http://sites
.ieee.org/vtools/). What does this mean? Listing events
Digital Object Identifier 10.1109/MPOT.2018.2877248
Date of publication: 28 December 2018
in vTools is more than just additional reporting require-
ments; it will allow you to tell professionals about your
Student Branch meetings. This is a great opportunity to
increase your interaction with local professionals. Ask
your Section leaders if they want to speak to your Stu-
dent Branch about the engineering work they are doing
in their job. Inquire about suggestions from other Section
members to be speakers. The Student Branch at my uni-
versity has been especially creative in the past couple of
years in utilizing the Internet, video cameras, and class-
room projectors to invite speak-
Student Branches can ers from around the world to talk
about technical topics. As a result,
always use more active we have seen an increase in pro-
members, and by being fessionals attending branch meet-
ings. vTools events are showcased
in IEEE Collabratec and can be
more out of your promoted on websites and eNotic-
membership in both es. IEEE Members and the public
may see and register for upcoming
events entered in vTools. Training
and networking for vTools is available through the
opportunities with the Center for Leadership Excellence
(https://ieee-elearning.org/CLE/).
Sections run on a calendar-year
of your Section. operations schedule. You pay your
membership on a calendar year.
As part of the movement to tighten the relationship be-
tween Student Branches and Sections, the annual Stu-
dent Branch reporting due date has been changed to 1
February. This replaces the Annual Report that had been
due yearly in November. By the time this article appears,
that 1 February deadline will be rapidly approaching, so
I hope you have started working on it. For students in
the southern hemisphere, this should work great for you.
For those in the northern hemisphere, this might sound
strange, but it will work, especially after the first year.
For now, think and plan ahead. Report your spring 2019
(Continued on page 48)

4 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals

 gamesman solutions

solution #1: Rock the Boat The speed of a body at any given point
When the man walks from one end of the in an elliptical orbit is given by the formula
lightweight boat to the other, the cen-
2 1
ter of gravity of the man-and-boat v = n a r - a k,
system remains fixed because no
external force acts on it. So, when w h e r e n = 3.986 # 10 14 m 3 /s 2 i s
the man moves one way, the boat the standard gravitational pa -
moves the other. When the man rameter. Due to the rotation of
stops, the boat stops. the elevator with the rotation of
the Earth, we also have v = ~r,
solution #2: Break the Chain where ~ = 7.292 # 10 -5 rad/s.
You should remove the fourth and Therefore, we want to solve the
11th paper clips from the chain. equation

solution #3: Pocket Change 2 2 2nR

na r - r + R k =
Numbers—© CaN stoCk Photo/123dartist,
~r = .
r ^r + R h
aNdroid—© CaN stoCk Photo/kirstyPargeter
The largest amount of change is US$1.19:
three quarters (or a half dollar and a quar-
ter), four dimes, and four pennies. Squaring both sides and rearranging terms, we have

solution #4: Juan For the Road 2nR

r 3 ^r + R h = .
Juan is the name of Pedro’s father. ~2

solution #5: space Elevator This is a quartic equation that is easily solved numerical-
A space elevator is usually conceived as extending up to ly. As a simple approximation, if we set r = t - _ 1 4 i R
geostationary orbit and beyond. If you were to catch the then r 3 ^r + R h . t 4, i.e., there is no t 3 term, so
elevator up to the geostationary level—an altitude of
35,786 km—and step off, you would appear to float near- 2nR 2nR R
by the elevator. The rotational speed of the elevator at t. 4
, which implies r . 4
- 4
~2 ~2
that level balances the gravitational pull so that you 2nR 5R
enter a circular orbit around the Earth that matches the and, therefore, altitude . 4
- 4 .
~2
rotation of the Earth. You have thrown yourself at the
ground and missed. But it’s possible to step off at a lower Substituting the values for n, R, and ~, we find an ap-
level and still miss the ground. At a lower level, you enter proximate altitude of only 23,299 km is necessary if you
an elliptical orbit with the center of the Earth at one want to throw yourself at the earth and miss by stepping
focus of the ellipse. Suppose your distance from the cen- off a space elevator. The numerical solution of the quartic
ter of the Earth is r when you step off the elevator, and equation gives the altitude as 23,412 km.
the semimajor axis of the ellipse is a. The point at which As a postscript, we note that what we have derived is
you step off the elevator is the vertex of the ellipse, i.e., a simplified version of the so-called Molniya orbit in the
the farthest point of the orbit from the Earth. At all equatorial plane. Molniya orbits are named for a constel-
points on the ellipse, you want to be farther from the lation of Russian telecommunication satellites whose or-
center of the Earth than the Earth radius, R, so that you bits are highly elliptical but, usefully, appear to hang over
miss the ground. We’ll assume a spherical Earth with a fixed point relative to an Earth observer for an extended
R = 6, 371 km. So we want r + R $ 2a. To find the period of time. This is just as you would appear to hang
smallest possible value of r, let’s set r + R = 2a. above an earthbound observer looking up from the foot of
the space elevator as you fell.
Digital Object Identifier 10.1109/MPOT.2018.2872280
Date of publication: 28 December 2018

IEEE PotEntIals Januar y/Februar y 2019 ■ 5

 Smart Consumer Electronics Systems
©iStockPhoto.coM/lvcandy
I
n 2017, I was invited by IEEE
Managing Editor Craig Causer
and IEEE Potentials then-Edi-
tor-in-Chief (EIC) Sachin Seth
to put t o g e t he r a consumer-
Digital Object Identifier 10.1109/MPOT.2018.2850539
Date of publication: 28 December 2018
6 ■ J a n u a r y / F e b r u ar y 2019
Smart consumer
electronics systems
electronics (CE)-related s p e c i a l
issue of IEEE Potentials. I started
thinking about a possible t heme.
During that sa me t ime period, I
was discussing renaming an IEEE
conference, the IEEE International
Symposium on Nanoelectronic and
Information Systems, for which I
IEEE PotEntIals
Saraju P. Mohanty
©iStockPhoto.coM/MF3d, Brain— ©iStockPhoto.coM/BuBaone
am the Steering Committee chair.
A fter much d i s cussion, we came
up with the idea to r en a me it the
IEEE International Symposium on
Smart Electronic Systems.
To convince people, I simplisti-
cally explained that nanoelectron-
i c a nd i n for mat ion systems a re
0278-6648/19©2019IEEE
essentially smart electronic systems.
So, I thought, why not make the fo-
cus of this IEEE Potentials theme
smart CE systems?
The idea become more concrete
when I was a panelist for the IEEE In-
ternational Conference on CE 2018 and
presented on the theme of energy and
security tradeoffs in CE systems.
The key message I delivered is that
tradeoffs among security, energy
consumption, latency, nonrecurring
design cost, and recurring operational
cost are needed for ef fective current
generation in Internet of Things (IoT)-
enabled CE system design. This can
be achieved by the appropriate use
of hardware and software compo-
nents in CE systems.
What are smart CE systems?
Smart CE systems (or, simplistically,
smart electronic systems) are envi-
sioned to be energy, security, and
response smart. These three aspects
and the design tradeoffs among
them are the key for next-generation
CE. Energy smart ensures minimal
energy consumption that allows lon-
ger battery life and smaller energy
bills. Security smart deals with the
security, privacy, or protection of CE
systems as well as that of the data
or media that these systems cap-
ture, process, or store. Response smart
refers to accurate sensing, intelligent
processing to gather knowledge or
information from the data, and fast
actuation or response based on the
information. Optimal combinations
of hardware and software modules
need to be explored for energy, secu-
rity, and response smartness of elec-
tronic systems.
From the hardware perspective,
a typical electronic system is an
analog/mixed-signal system on a
chip (AMS-SoC) containing ana-
log, communications, digital, radio-
frequency, memory, sensors, and
energy-source components as well as
firmware, system software, middle-
ware, and application software. We
are in an era of smart environments
such as smart cities, homes, health
Optimal combinations of hardware and software
modules need to be explored for energy, security,
and response smartness of electronic systems.
care, and transport systems that are
combinations of cyberphysical sys-
tems (CPSs) realized using the IoT
and working collectively to provide in-
telligence or smartness. As a result,
to pursue the efficient realization
of smart electronics, the IoT, CPSs,
and smart environments, there is a
need to research new hardware, firm-
ware, middleware, and software fac-
ets that will interact with one other.
This IEEE Potentials special issue
makes an effort to bring this con-
cept to the forefront.
scanning this issue
The articles in this issue were in-
vited from selected established re-
searchers and reviewed for quality.
They include:
■■“Power Supplies for Consumer
Electronic Devices” discusses the
power supplies that are needed
in every power-driven CE system.
■■“S-CHIRP: Securing Communica-
tions in Lightweight Peer-To-Peer
Networks in the IoT” presents a
protocol for secure communica-
tion for heterogeneous IoTs with
round-robin protection.
■■“A Neighbor Trust-Based Mech-
anism to Protect Mobile Networks”
introduces a neighbor trust-based
security scheme that prevents
malicious attacks in a mobile ad
hoc network.
■■“Proof of Authentication: IoT-
Friendly Blockchains” presents a
new consensus algorithm that
can make it possible for block-
chains to run in the IoT with min-
imal computational requirements.
■■“Drones: Augmenting Our Quality
of Life,” discusses drone technol-
ogies in terms of research op-
portunities, and the article dem-
onstrates, through case studies,
the implications of social, econom-
ic, and scientific impacts.
IEEE PotEntIals
■■“A Method for Localizing the Eye
Pupil for Point-of-Gaze Estimation”
presents a method to estimate
the point of gaze on a screen by
implementing nonintrusive eye
tracking that can be used for
fatigue detection and attention
tracking. As a result, it has diverse
applications, such as in smart
cars and smart health care.
■■“E-Marketing via Augmented
Reality: A Case Study in the
Tourism and Hospitality Indus-
try,” focuses on the use of aug-
mented reality in e-marketing
with an emphasis on the hospital-
ity and tourism industry.
looking forward
I would like to sincerely thank IEEE
Managing Editor Craig Causer and
current IEEE Potentials EIC Vaughan
Clarkson. I thank all of the contrib-
uting authors, without whom this
issue could not have become possi-
ble. I sincerely hope that this special
issue of IEEE Potentials will be excel-
lent reading for many students and
researchers as well as general read-
ers around the world.
about the guest editor
Saraju P. Mohanty (saraju.mohanty@
unt.edu) is a professor at the Uni-
versity of Nor t h Te x a s, D e nt o n .
His research in sma r t elect ron ic
systems has been funded by the
National Science Foundation, Semi-
conductor Cor porat ion, and the
U.S. Air Force. He has authored 280
research articles and three books
and holds four U.S. patents. His
Google Scholar h-index is 29 and
i10-index is 90. He is the editor-in-
chief of IEEE Consumer Electronics
Magazine. He serves as the chair of
the Technical Committee on Very
Large Scale Integration of the IEEE
Computer Society.
Januar y/Februar y 2019 ■ 7
 Smart Consumer Electronics Systems
©iStockphoto.com/lvcandy

Power supplies
for consumer
electronic devices
Santanu mishra

A
device used for noncom-
mercial purposes in enter-
tainment, communication,
and home offices is called
a consumer electronics de­
vice (CED), which incorporates elec-
tronic circuits for operation. By this
definition, computers, cell phones,
mobile devices, light-emitting diode
(LED) lights, TVs, cameras, gaming
stations/consoles, and various other
home elect ronics a re exa mples
of CEDs.
Most CEDs require dc power to
drive the electronic circuits. A utility-
connected house receives a single-
phase ac input. This supply can be
230 V/50 Hz or 110 V/60 Hz depend-
ing on geographical location. As
shown in Fig. 1, an ac/dc power sup-
ply is used to interface the ac grid
input to a CED (in this case, a moth-
erboard of an LED TV). Inside a CED,
there are various analog and digital
electronic circuits and integrated cir-
cuits (ICs) to process electronic infor-
©iStockphoto.com/Romanokopny

mation. These ICs require dedicated

power supplies because the volt-
age, current, and dynamic require-
ments are different. Therefore, there
are multiple dc/dc power supplies,
known as point­of­load converters on
the IC card or motherboard.
With renewable energy [e.g., roof-
top solar photovoltaic (PV)] becoming for use in homes. There is also an ef- multinational companies to stream-
popular, dc supply is directly available fort to standardize dc power distri- line this effort.
bution and CEDs that work with this In this article, a general overview
Digital Object Identifier 10.1109/MPOT.2018.2850478
standard. A consortium known as of power converters used in CEDs is
Date of publication: 28 December 2018 Emerge Alliance works with several discussed. Basic network theory and

8 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals 0278-6648/19©2019IEEE

control engineering will be used to ex-
plain the fundamental operation of a
converter and its application.
Power converter classification
Based on input/output profile
A power converter or power supply
is used to efficiently convert and
control power from one voltage/
current/frequency level to another.
Based on the profiles of input and
output voltages, a power converter
can be classified into four types.
1) dc/dc converter [Fig. 2(a)]: Its input
and output voltages both are uni-
polar in nature. Depending on the
magnitude of voltages, they are
further divided into three types:
a step-down, or buck type, where
the magnitude of the output vol-
tage is smaller than the input
voltage; a step-up, or boost type,
where the magnitude of the output
voltage is higher than the input
voltage; and an up-down, or buck-
boost type, where the magnitude
of the output voltage can be either
smaller or higher than the input
voltage. Buck-type dc/dc convert-
ers are very common in CED ap-
plications to power digital ICs
inside a CED. Boost-type dc/dc
converters find applications in
universal serial bus charging
ports, sensor power, and powering
portable speakers, among others.
Similarly, buck-boost-type dc/dc
converters find applications in
smart devices, bias supplies, and
so on.
2) dc/ac inverter [Fig. 2(b)]: It has
a unipolar input voltage and a
bipolar periodic ac voltage. They
are quite common in uninter-
rupted power supply design to
provide battery backup for grid
supply. Solar inverter to interface
PV source to grid also belongs to
this class of converters.
3) ac/ac converter [Fig. 2(c)]: Its input
and output voltages are both bi-
polar. In modern applications,
ac/ac converters can be realized
by cascading an ac/dc converter
and a dc/ac converter.
4) ac/dc rectifier [Fig. 2(d)]: It has a
bipolar periodic ac input and
a unipolar dc output. All types of or step-down ratio between the input
ac adapters belong to this family and output. Figure 3(b) and (c) de-
of converters. picts the way an off-line isolated con-
From this classification, it is clear verter provides protection against in-
that, as far as block-level operation is sulation failure of equipment. In an
concerned, every converter is a non- isolated dc/dc converter, the trans-
linear block because its output con- former provides insulation against
tains a frequency component other direct short to ground [Fig. 3(b)]. In
than its input frequency. comparison, a nonisolated dc/dc
converter [Fig. 3(c)] doesn’t provide
Isolated versus nonisolated protection against insulation failure.
Sometimes in off-line applications, a
converter is required to galvanically Power converter building blocks
isolate the utility grid input and the A power converter consists of two
dc output. This is generally achieved major building blocks: a power stage
by using an isolated dc/dc converter. and a control stage (see Fig. 4). The
The basic structure of an isolated power stage is responsible for pro-
dc/dc converter is shown in Fig. 3(a). cessing the power between the input
It is realized by cascading a dc/ac and output. The control stage con-
inverter, a high-frequency transformer trols the converter parameters, e.g.,
(ferrite-core-type transformer), and maintains the fixed-output voltage if
an ac/dc converter. High-frequency the input voltage varies. The control
transformers make the overall con- stage doesn’t have any power pro-
verter design smaller by reducing cessing capability and is generally
the transformer size. realized using low-power electronic
Isolated dc/dc converters are used circuits. Both mixed-signal ICs and
to improve safety, enhance noise im- fully digital ICs are common in the
munity, and achieve a large step-up implementation of the control circuit.
dc/dc dc/dc
ac/dc Converter 1 Converter 2
Supply 3.3-V
230 V, 50 Hz ac dc
110 V, 60 Hz
ac/dc
Supply 19.5-V
dc
ac/dc
Supply 19.5-V
dc
FIG1 the several types of power supplies or power converters used to supply power to
the cEd.
vi, ii vo, io
dc/dc vi, ii vo, io, fo
dc/ac
Converter
vi vo vi Inverter
vo 0
(a) (b)
vi, ii, fi vo, io, fo vi, ii, fi vo, io
ac/ac ac/dc
Converter vi Inverter
vi vo 0 0 vo
0
(c) (d)
FIG2 the power converters are classified based on input and output voltage profile. (a)
the dc/dc (green) has a unipolar input and a unipolar output. (b) the dc/ac (blue) has a
unipolar input and an ac output. (c) the ac/ac (yellow) has an ac input and an ac output
(the frequency and magnitude of the ac may be different). (d) the ac/dc (gray) has an ac
input and a unipolar output.
IEEE PotEntIals Januar y/Februar y 2019 ■ 9
There are various sensors that feed pulsewidth modulated signal to drive and control implementation decides
the converter voltage/current in- the converter. the source type. If the output voltage
formation into the control circuit. From a terminal behavior point of of the converter is regulated, the con-
Based on this information, the con- view, a converter works like a volt- verter behaves like a voltage source.
trol stage generates a square-wave age or current source. The feedback This essentially means that, similar
to a voltage source, the output voltage
of the converter remains constant ir-
HF respective of variation in load current,
Transformer input voltage, and other parameters.
dc/ac ac/dc dc Similarly, if the output current of the
dc
Stage Stage converter is regulated, the converter
(a) behaves like a current source. In this
ac/dc case, the load resistance decides the
dc/ac
ac/dc
output voltage.
Figure 5 shows an example in which
Vgrid
a step-down dc/dc converter is used

Insulation
Failure
as a current source. In this case, the
inductor current is fed back to the
controller and regulated to a desired
Isolated dc/dc Converter reference value. Because of the pres-
(b) ence of a controller, it is possible to
dc/dc implement sources with square-wave
Vgrid > 50 V
output as well. When the converter is
+ operating as a source, it must follow
Insulation
Failure

Electric the rules of circuit theory, as depicted

Shock
in Fig. 6(a).
Nonisolated dc/dc Converter
some important design factors
(c)
Output power
FIG3 (a) an isolated dc/dc converter. (b) isolated converters provide protection against
The power rating of a converter de-
insulation-failure-based ground faults. (c) nonisolated dc/dc converters do not provide
protection against insulation-failure-based ground faults. hF: high frequency. pends on the load it is designed to
support. The overall objective is to
reduce cost and improve the efficiency
of the converter.
Power Stage When the load current is in the
Io order of tens to hundreds of milli-
Vin amperes, a simple linear regulator or
Load
switch capacitor circuit is preferred.
They are mainly used for bias appli-
cation. However, as the load current
Compensator level increases beyond a few amperes,
PWM Vo, Io, etc. a switching converter is the most op-
– timal solution because it provides the
+ Reference best tradeoff between efficiency and
Control Stage component count.
Here, another factor that affects the
(a)
power supply design choice is the ra-
Io tio of the input and output voltages. If
+ + + + the input-to-output ratio is high, even
if the load current is low, a switching
Vo Load Vo Load converter is the most optimal solution.
Io This is because, in a linear regulator,
– –
with a high input-to-output ratio, the
(b) voltage drop across the semiconduc-
tor device is very high, which leads to
FIG4 (a) the basic building blocks of a dc/dc converter. (b) the electrical equivalent higher power loss. When the load cur-
circuit of a converter. pWm: pulsewidth modulator. rent is very high, it is advantageous

10 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals

to use multiple parallel modules, as ■■Circuit design, printed circuit board (MOSFETs) are mostly preferred as
shown in Fig. 6(b). This type of design layout, and electromagnetic in- switches in CED power processing
leads to better efficiency, energy den- terference become critical. circuits. Compared to silicon (Si)-
sity, and modularity to the power con- The selection of switching frequen- based MOSFETs, gallium nitride
version system. cy depends on switch type. Metal–oxide– (GaN)-based MOSFETs can operate at
semiconductor field-effect transistors a much higher switching frequency
Dynamics and control
As far as control function is concerned,
there are primarily two aspects of Power Stage vsw L /rL Vo
power converters used in CEDs that
are important: 1) control, in that the Ro
+
converter output follows a desired Vin
Gate
input reference signal, and 2) regula-
Driver
tion, in that the converter output (volt- Co
age or current) remains within desired
limits with change in input voltage,
load, or other parameters.
Vramp –
These concepts are depicted in Fig. 7. Vpwm
As shown in Fig. 7(a), the converter IL
+
output follows the variation in refer-
ence signal. Similarly, the response of
converter output voltage to a dynamic Iea Current Iref
change in load current for a server Compensator
power supply is shown in Fig. 7(b). Control Stage
The output regulation performance (a)
(<100  mV) during load change is evi-
dent. Here, control over the regulated
output as a function of current is dem- 100
onstrated. At higher current, the out-
put voltage is regulated at a smaller
value compared to the output voltage
(V)

at a lighter current. It helps in the 50

reduction of the size of the filter ca-
pacitor because the converter output
Iref
is now allowed to vary during dyna- IL
mic operation. 0
0 0.5 1 1.5 2 2.5 3 (ms)
High switching V (IL) V (IREF) Time
frequency operation (b)
The switching frequency is the rate
at which the switches of the convert- FIG5 an example of a controlled buck converter as a current source. (a) the schematic
and (b) response of the source to a given reference (l ref) .
er are turned on and off in a second.
In modern designs, the switches of
the converter are operated at 100 s of
kilohertz to several megahertz. The
I2
advantages of higher switching fre-
quency operation of a converter are Io Io Load
Vin Converter 1 Vo
the following. I1
■■It leads to smaller magnetic and
filter capacitor size. This results V2 Converter 2
in higher power processing den-
Load
sity of the converter. Vo Vo V1
■■The converter can be designed to
have faster dynamic response. (a) (b)
The drawbacks of increasing the
switching frequency are the following. FIG6 (a) the network theory rules need to be followed by a dc/dc converter operating
■■It leads to higher power loss in as an independent source. (b) the paralleling of dc/dc converters is a very common
the switches. technique to realize higher-power modules.

IEEE PotEntIals Januar y/Februar y 2019 ■ 11

 Tek Stop Tek Stop

Output Voltage Regulation Limit

Vo
Vo

2
Vref Load Step Up
1

Io 100 A

1 Reference Step Change

4
Load Step Down
Ch1 50 mVΩ M 200 µs A Ch4 320 mV
Ch1 200 mV Ch2 1.00 V M 100 µs A Ch1 512 mV 500 mV
Ch4

(a) (b)

FIG7 the examples of converter dynamics. (a) the control of the output voltage using a reference signal and (b) the regulation of a con-
verter output with a change in load current. the feedback loop forces the converter to maintain an output voltage within a specific limit.

hertz switching frequency range are

98 reported with GaN-based devices.
Figure 8 shows the comparison
97 Buck Converter with GaN of efficiency of a buck converter with
Devices (EPC2012)
96 Si- and GaN-based MOSFETs as a
function of switching frequency. At
Efficiency (%)

95
higher switching frequency, GaN-
94 based converters have a distinct effi-
Buck Converter with ciency advantage. Due to a very high
93
Si Devices (IRF7492) speed of operation, inherent to GaN-
92 based MOSFETs, their packaging
plays an important role in harness-
91
ing this feature.
90
100 200 300 400 500 600 700 800 900 1,000
Switching Frequency (kHz)
application examples

FIG8 a comparison of the efficiency of a synchronous buck converter with Si and LED dimming circuit
Gan moSFEts. The illumination of an LED string
can be boosted by increasing the
average current through it. The pow-
er converter driving the LED string
I1 (commonly known as an LED dri ­
Ia ver) works as a current source to
ipwm
LED String

achieve the dimming function in LED

0 lights. The LED driver is a controlled
dc/dc converter.
fpwm > 200 Hz dc/dc 1
Analog
As shown in Fig. 9, there are two
dc/dc 2 Dimmer ways to realize a variable current
PWM Dimmer source. In the first method, the aver-
age current through the LED string
is regulated, which is known as ana­
FIG9 the basic technique used in an lEd dimming circuit. log dimming. In the second method,
the average current is controlled
(in multimegahertz) due to low par- ing loss compared to Si-based MOS- by forcing a square-wave current
asitic capacitances in GaN devices for FETs, especially in the megahertz through the LED string. In this case,
the same on-resistance of the switch. switching frequency range. Efficient the pulsewidth of the square wave
GaN devices have a much lower switch- power converters in the multimega- determines the average current. The

12 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals


Vbat
Feedback
Low SOC
Control
Icrg
+ 13
Vref High SOC
Optimal Charger
(a)
FIG10 (a) the basic control of an optimal battery charger and (b) an experimental battery-charge profile of a 36-ah, 12-v lead
acid battery.
frequency of the square wave needs to
be much higher than 200 Hz so that
it can’t be recognized by the naked
eye. The second method is more pop-
ular in LED dimming applications.
Battery chargers
The method of charging is an impor-
tant factor that decides the life of a
battery. There are various charging
methods available, including constant
current (CC), constant voltage (CV),
boost charging, pulsed charging, trick-
le charging, and float charging.
Normally, to improve the life of a
battery, it is charged with a CC when
the battery state of charge (SOC) is
low and with a CV when the battery
SOC is near 100%. In CV mode, the in-
ternal resistance of the battery limits
the battery-charge current. There-
fore, the charger needs to work as a
current source when battery SOC is
low, and it should work as a voltage
source when battery SOC is high.
Figure 10(a) depicts an optimal
battery-charger circuit where the
same converter is made to operate
like a voltage and current source. The
tracking of battery SOC is achieved
by monitoring the battery terminal
voltage (V bat), using a feedback con-
trol circuit as discussed in the “Power
Converter Classification” section. For
current source operation, the con-
verter is fed back with battery cur-
rent, which is regulated to Icrg. For
voltage source operation, the convert-
14.5
14
13.5
Vbat (V)
12.5
Battery 12
11.5
11
0 428
er output voltage is feedback into the
controller. An experimental plot of
battery-charging behavior is shown
in Fig. 10(b). Between 0 and 900 min,
the converter charges the battery as
a current source when battery SOC is
below 90%. After this, the converter
works like a voltage source and its
current output is controlled by the in-
ternal resistance of the battery.
Read more about it
• S. Mishra and O. Ray, “Advanc-
es in nanogrid technology and its in-
tegration into rural electrification in
India,” in Proc. 2014 Int. Power Elec­
tronics Conf., Hiroshima, Japan, pp.
2707–2713.
• N. Mohan, T. M. Undeland, and
W. P. Robbins, Power Electronics:
Converters, Applications, and Design,
3rd ed. Hoboken, NJ: Wiley, 2003.
• R. White, “The amazing server
power supply,” IEEE Power Electron.
Mag., vol. 2, no. 3, pp. 62–64, 2015.
• C. Jain and B. Singh, “Solar
energy used for grid connection: A
detailed assessment including fre-
quency response and algorithm com-
parisons for an energy conversion
system,” IEEE Ind. Appl. Mag., vol. 23,
no. 2, pp. 37–50, 2017.
• R. W. Erickson and D. Maksi-
movic, Fundamentals of Power Elec­
tronics, 2nd ed. New York: Springer-
Verlag, 2001.
• Intersil Americas. (2018). Power
Management ICs ISL8273M, ISL80113,
IEEE PotEntIals
CC Mode CV Mode
Icrg 1.8
Vref
1.6
1.4
1.2
Ibat (V)
1
0.8
0.6
0.4
0.2
0
729 998 1,478
Time (min)
(b)
ISL99140 datasheets. [Online]. Avail-
able: www.intersil.com
• S. K. Mishra and K. D. T. Ngo,
“Dynamic characterization of the
synthetic ripple modulator in a tight-
ly reg u lated d ist r ibuted power
application,” IEEE Trans. Ind. Elec­
tron, vol. 56, no. 4, pp. 1164–1173,
2009.
• R. R. Duarte, G. F. Ferreira, M A.
Dalla Costal, and J. Marcos Alonso,
“Performance comparison of Si and
GaN transistors in a family of syn-
chronous buck converters for LED
lighting applications,” in Proc. 2016
Industry Applications Society Annu.
Meeting, Portland, OR, pp. 1–7.
• R. Rosen, “Dimming techni-
ques for switched-mode LED driv-
ers,” Texas Instruments, Dallas, TX,
Rep. SNVA605, 2011.
• R. Singh and S. Mishra, “A mag-
netically coupled feedback-clamped
optimal bi-directional battery char-
ger,” IEEE Trans. Ind. Electron., vol.
60, no. 2, pp. 422–432, 2013.
• S. Mishra, “Power converter sys-
tems for consumer electronics de-
vices,” in Proc. 2016 IEEE Int. Symp.
Nanoelectronic and Information Sys­
tems, pp. 72–75.
about the author
Santanu Mishra (santanum@iitk
.ac.in) is a professor at the Indian
Institute of Technology, Kanpur, India.
Januar y/Februar y 2019 ■ 13
©iStockphoto.com/lvcandy
Smart Consumer Electronics Systems

S-CHIRP: Securing
communications in
lightweight peer-to-peer
networks in the IoT
Shaya Wolf, Rafer cooley, and mike Borowczak

A
lthough you may think of
our technological world as
a giant web of elegantly
interconnected devi­
ces, the reality is
a collection of messy overlap­
ping networks, all commu­
nicating si mult a neously

_W
ky
on disparate infrastructural

eS
h
/c
backb o n e s. To communicate

m
o
.c
to
o
within the chaos today, speci f ic

h
kp
c
to
components may be forced to share

iS
©
t—
o
space on dedicated local networks,

,i
ve
ti
ea
or, in many cases, they must rely on

R
c
ee
/B
m
complex switching and routing meth­
o
.c
to
o

ods to transfer information.

h
kp
c
to

As more devices are added to

iS
©
i—
i-F

peer­to­peer (P2P) networks in highly

W

specialized, lightweight, and local­

ized environments, the security and
isolation of communication within a
network remains a major design chal­
lenge. By examining several commu­
nication problems in specific domain
applications, a secure communica­
tion protocol, secure communication
for heterogeneous Internet of Things
(IoT) with round­robin protection
(S­CHIRP), is used to explore and de­
fine potential minimalistic solutions.
Iot overview
An enormous network of simple hard­
ware sensors, controllers, and actua­
Digital Object Identifier 10.1109/MPOT.2018.2850600
Date of publication: 28 December 2018
tors continues to per­
meate the infrastruc­
ture of our 21st­century
civilization. Growing in
popularity, the IoT requires
lightweight communication
protocols across a variety of different
devices to enable its pervasive and
extended reach. Without lightweight
protocols, users struggle to feasibly
connect all of their devices together
without unintended communication
overlap. People don’t want to acciden­
tally give control of all of their Web­
connected light bulbs to a neighbor,
but they also don’t want to spend a
day figuring out how to connect them
all together. Wang, Lee, and
Murray (2017) found that
because network resources,
standards, and communica­
tion patterns vary greatly between
devices, IoT platforms are limited
and have unfulfilled potential.
As the number of IoT devices con­
tinues to grow, human­to­machine
(H2M) interactions are quickly repla­
ced by machine­to­machine (M2M)
interactions. Take, for example, an
IoT sensor network that measures
crop soil properties and communi­
cates directly with a set of IoT water
sprinklers and IoT fertilizer tanks
to alter the fertilizer­to­water ratio

14 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals 0278-6648/19©2019IEEE

being supplied. Figure 1 shows a more Because network resources, standards,
consumer­level scenario related to
lawn­maintenance devices. To scale, and communication patterns vary greatly
M2M networks need to i mplement between devices, IoT platforms are limited
lightweight communication protocols and have unfulfilled potential.
that would still allow neighbors to
have their own individual networks
while potentially sharing a common protocols, such a s S ­ CH I R P, are interactions between different devic­
infrastructural backbone. seg mented into cycles. During one es, or communication rounds, form
With the immense volume of IoT complete cycle, every device needs to a complete cycle. With bidirectional
M2M communication, efficient solu­ communicate with every other device communication, each device com­
tions must be developed that ensure in an organized fashion that each de­ municates with exactly zero or one
privacy from potential intruders as vice agrees upon. The many unique other device during each round of
well as other nearby IoT networks.
The S­CHIRP protocol is one mecha­
nism for secure and efficient com­
munication within IoT networks. The
protocol enables organized pairwise
communication between IoT devices
using only a limited amount of in­
ternal data. S­CHIRP caters to M2M
device communication and group se­
paration, without the need of a cen­
tralized agent. H2
M

M2M communication in the Iot

S­CHIRP focuses on lightweight de­
vices that would constitute a mesh M2
M
or swa r m P 2P net work. T h is in­
cludes unmanned aerial vehicle
(UAV) swarm networks, blockchain
com putational nodes, sensor net­ FIG1 h2m versus m2m. on the left, a human interacts with one single lawn
works, ingestible biomedical sensors, maintenance device. on the right are five interconnected lawn maintenance devices.
smart home networks, file­sharing
structures, electric smart meter sys­
tems, field com mun icat ion net­
works, and similar platforms. With
enhanced security and the absence
of third­party interests, decentralized
networks are gaining traction among
users who are not willing to trade
privacy for convenience. The differ­
ence between centralized and decen­
tralized networks is visualized using
UAVs in Fig. 2.
After deployment, the communi­
cation within a decentralized network
is dominated by M2M interactions.
Without the need for a server to or­
ganize the clients, S­CHIRP enables
faster and more secured networks.
As multiple devices are added to
networks, they begin to form groups,
or cliques, based on logical or securi­
ty constraints. Network protocols dic­
tate how devices in a specific clique FIG2 the mission of four Uavs is coordinated through constant communication to and
send messages to each other without from a centralized base station on the left. on the right, the four Uavs are coordinated
interfering with others. Round­robin through decentralized communication and consensus.

IEEE PotEntIals Januar y/Februar y 2019 ■ 15

 Efficient solutions must be developed that
ensure privacy from potential intruders as well
as other nearby IoT networks.
communication. The effectiveness
of a protocol is determined by the
number of rounds per cycle required
to establish conversations between
all devices. S­CHIRP accomplishes
this in N rounds, where N is the ma­
ximum number of devices in a clique.
Each device only needs to know a
few things to communicate with­
out a server: first, the current round
(R), second, their unique identifier
within their clique (I), and finally,
the maximum number of devices in
their clique (N ).
With this information, a device can
quickly and easily obtain the identi­
fier of its targeted communication
partner device (T). This lightweight
communication is achieved by requir­
ing every device to compute a com­
municate target using a three­param­
eter equation (Borowczak and Purdy
2018). For S­CHIRP to be successful,
it must solve the major challenge in
the adoption and use of lightweight
IoT devices. Specifically, it must en­
able scalable, serverless M2M com­
munication between devices when
multiple distinct groups are present.
Problems and challenges in
M2M clique communication
When multiple groups, or cliques, of
IoT devices exist on the same net­
work backbone, M2M communica­
tion protocols must overcome several
issues. Overlapping networks create
issues in maintaining secure com­
munication within a clique and en­
0 1 2 0
0 1 0 1
Inactive Inactive
Clique 1
FIG3 the several communication scenarios between collocated cliques in m2m networks.
16 ■ J a n u a r y / F e b r u ar y 2019
suring efficiency. Four common
issues encountered when networks
overlap are described in the follow­
ing sections and illustrated in Fig. 3.
Inactive communications
Inactive communications arise in two
ways. First, a device calculates a tar­
geted communication partner who is
out of reach. Second, a device calcu­
lates a targeted communication part­
ner for a device that does not current­
ly exist in network. In both scenarios,
when a device attempts to communi­
cate with an unavailable device, the
resulting communication is inactive.
Inactive communications are not
security threats, but they reduce
overall communication efficiency.
Valid communications
A valid communication connects two
devices in the same clique. Devices
in the same clique must talk to each
other through valid M2M communi­
cations efficiently and securely.
There can be no interference with
the connection (intentional or not),
and these connections must be reli­
able regardless of the number or dis­
tribution of cliques within the same
network backbone.
Cross-clique communication—
invalid communications
Cross­clique communications occur
when a device communicates outside
of its clique. When a device calcu­
lates its targeted communication
1 0 1 0 1 vidual’s identifier after being divided
0 1 0 1 0 1
Valid Invalid Collision
Clique 2 Empty Node
IEEE PotEntIals
partner identifier, it might find that
there are multiple devices in the
vicinity with that same identifier.
Invalid communications are a two­
way cross­clique communication and
occur when devices in two cliques
compute the same targeted partner
identifier. Consider soil sensors, e.g.,
that dictate to automatic sprinkler
systems when to water a lawn. An
invalid communication would be
when two neighbors’ soil sensors both
communicate with the wrong sprin­
kler systems at the same time.
Cross-clique communication—
collisions
In contrast to invalid communica­
tions, collisions are one­way cross­
clique communications, where two
devices from different cliques calcu­
late the same target communication
partner. Consider the soil sensors
again. A collision occurs when two
neighbors’ soil sensors both simulta­
neously communicate to one sprin­
kler system. One sprinkler system
has competing information, while the
other has no information. Cross­
clique communications poses a secu­
rity threat by enabling attackers to
generate traffic within the same
space as a target network and per­
form denial of service attacks. The
objective of S­CHIRP is to minimize
the likelihood of these communica­
tion issues while maintaining a light­
weight M2M communication protocol.
a proposed solution for secure
M2M communication in cliques
Exact per-clique device counts
Consider the following method for an
individual device to calculate a com­
munication target. Let the target
equal the remainder of the current com­
munication round offset by the indi­
the number of devices in the group.
Take, e.g., a room full of people where
two constraints are in place.
First, each person needs to talk
with every other person in the room,
and second, each person can only
talk to one other person at a time.
This communication must be done
in rounds, where pairs of people talk
each round. Once everyone in the
room has talked to each other per­
son once, there has been one cycle of
communication. If everyone needed
to talk with every other person mul­
tiple times, it stands to reason that
you could determine a way to get ev­
eryone to talk to each other once,
then repeat that process as many
times as you need. Without any di­
rection, it could be very clustered and
not very efficient.
Perhaps your first intuition is to
have someone at the front of the room
give instructions on who to com­
municate with next. However, with
S­CHIRP, this same type of commu­
nication is organized without the
need for a centralized controller. It
does this by having each node cycle
through the rest of the nodes by using
modular arithmetic. If applied to this
room full of people, S­CHIRP would
facilitate one cycle of communication
in N rounds, where N is the number of
people in the room.
However, the communication fol­
lows the same pattern each cycle,
much like the room where ever y
person needs to talk to each other
person multiple times. Therefore the
first communication pairs in each
cycle are the same. Specifically, ev­
ery time a new cycle starts, the first
round of communication will have a
node with index 0 communicate with
a node of index 1. With overlapping
cliques, nodes will be looking for a
nonunique node index and could
accidentally (or intentionally) com­
municate to a device in a different
clique. This attack vector is greatest
when clique sizes (N ) are common
multiples of each other. In these situ­
ations, cycles begin at the same time
with similar communication pat­
terns. A representation of this over­
lap for two cliques of size four and
two is given in Fig. 4.
S­CHIRP presumes that cliques
know nothing about each other. If
each device knows the total number
of devices within its clique, they can
communicate based on that count
and the round they are on. Because
the identifiers assigned to devices
are determined within each clique,
there exists the potential for two
Decentralized networks are gaining traction
among users who are not willing to trade privacy
for convenience.
devices, in two different cliques, to to account for a desired population
have the same identifier, thus allow­ and number of cliques, there will be
ing for a communication collision. In some cliques containing devices that
the worst case, the current round is may not exist in that network. These
the same across the network, and placeholders increase the number of
there are multiple nodes with the inactive communications as actual
same identifier, the only distinguish­ devices attempt to communicate with
ing factor in calculating the targeted the nonexistent devices.
communication partner among the
cliques is the maximum clique size. Distinct prime device count
Devices can calculate the same By setting the device count of the
target communication partner iden­ cliques seen in the earlier example to
tifier as a device in a neighboring the nearest unique prime number,
clique, i.e., when the device counts of the network benefits from a decrease
different cliques are multiples of each in communication overlap. To avoid
other. A clique size of two, e.g., will the overlap seen in the exact per­
overlap frequently with a clique of clique device count solution, choos­
size four because they are communi­ ing a unique prime number device
cating in the same pattern. Figure 5 count can keep cycles from overlap­
demonstrates these cliques. A mali­ ping because they will take different
cious node, which can listen to com­ times to complete cycles, and smaller
munications to get the node count, cliques cannot be multiples of larger
can use this overlap to infiltrate the cliques. However, there is still a chance
clique using a slightly stronger sig­ for overlap in later rounds, specifi­
nal than one of the other members. cally at the least common multiple of
Because there might not be a perfect two clique counts. This concept is
distribution of prime clique sizes modeled in Figs. 6 and 7.
Clique 2 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8
Clique 1 1 1 1 1 2 2 2 2 3 3 3 3 4 4 4
Round 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
FIG4 clique 1 has four rounds per cycle (blue, green, orange, pink), one per node.
clique 2 has two rounds per cycle (blue and green). cycle numbers are shown in white.
cycle overlaps occur in the first two rounds, recurring at a four-round offset.
0 1
0 1 Rounds 1, 5, 9 0 1
Rounds 2, 6, 10 Rounds 1, 3, 5 1 0
Rounds 3, 7, 11 Rounds 2, 4, 6
2 3 Rounds 4, 8, 12 2 3
(Inactive)
Rounds 1, 5, 9...
(Every Four)
Clique 1 Clique 2
FIG5 the exact per-clique device count model: normal behavior of isolated cliques (left
and center panels), and combined collocated behavior (far-right panel).
IEEE PotEntIals Januar y/Februar y 2019 ■ 17
 Biosensors are proving to be the health-care tech
of the future, enabling ubiquitous monitoring
systems no larger than a common pill.
selected case study
applications
The dominance of a lightweight M2M
communication in P2P networks is
exemplified in a snapshot view of
case study applications. Figure 7
captures a visual representation of
several of the applications that are
also described in this section. Fur­
ther examples of M2M communica­
tion in IoT networks can be explored
in the numerous works surrounding
smart cities (Mohanty, Choppali, and
Kougianos, 2016), edge computing
(Corcoran and Datta, 2016), and, of
course, wearables (Wei, 2014).
Smart home networks
Mesh networks are rapidly taking a
foothold within the home appliance
market. Smart thermostats and other
devices already operate in harmony
to control everything from tempera­
ture to ambiance. Densely populated
neighborhoods contain many vary­
ing­sized cliques depending on how
many devices have been installed
within a particular household.
UAV swarms
As UAVs become cheaper to produce,
the potential for hundreds, if not
thousands, of UAV agents to interact
with one another rises. According to
Atherton (2017), UAV swarms are
gaining traction both in civilian and
military domains. Drawing inspira­
tion from bird flight patterns and
insect swarm patterns, such bio­
morphic systems as UAV swarms
move and act in ways to avoid colli­
sions while achieving an end goal.
Their ability to share minimally sized
messages is critical.
Sensor networks
Sensor networks, like crop­watering
sensors, are generally deployed with
a large number of computationally
constrained nodes. As stated by Shi
and Perrig (2004), sensor nodes are
susceptible to physical capture, eaves­
dropping, and collusion. Without a
secure, lightweight communication
protocol like S­CHIRP, these networks
could easily be influenced by mali­
cious actors.

0 1 Ingestible biomedical sensors

0 1 0 1
Rounds 1, 6, 11 According to Bachmann et al. (2012),
Rounds 2, 7, 12 1 0 biosensors are proving to be the health­
2
Rounds 3, 8, 13 care tech of the future, enabling ubiq­
2 3 2 3
Rounds 4, 9, 14 Rounds 1, 4, 7
Rounds 2, 5, 8
uitous monitoring systems no larg­
Rounds 5, 10, 15 4 2
4 Rounds 3, 6, 9 er than a common pill. The sensors
Rounds 1, 16... are tiny and lightweight, requiring
(Every 15)
a communication mechanism to
match. Because each node requires
Clique 1 Clique 2 Empty Node
exceptional computation perfor­
mance with serious size constraints,
FIG6 the distinct prime device counts model with cliques sizes four and two, using there is little room left for a robust
device counts of five and three. communication protocol. Further, as
portions of the population begin in­
gesting these sensors, it becomes
extremely important to ensure that
their sensors do not have overlap­
ping communications.

Blockchain computational
nodes
Blockchain computational nodes
currently have access to a large amount
of computing power. However, as
blockchain technology grows in pop­
ularity, so does the desire to make
it lightweight and portable. Even
with large amounts of computational
power, nodes require reliable means
of communication. Given that block­
FIG7 the overlapping home-sensor networks become highly overlapping in densely chain relies on a serverless architec­
populated regions (far left), a decentralized Uav swarm (top center), sensor networks ture, a serverless communication
(bottom center), and ingestible biomedical sensors (far right). protocol, such as S­CHIRP, lends

18 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals

itself to future advances in decen­
tralized autonomous organizations.
Conclusion
The dominance of M2M communica­
tion in IoT networks can be controlled
using a decentralized communication
solution, S­CHIRP, that scales to mul­
tiple cliques of devices while remain­
ing lightweight and device agnostic.
In addition to providing a communi­
cation strategy for pairwise commu­
nication in single­clique worlds, a
solution for multiple cliques leverag­
es the existing strategy by decreas­
ing the likelihood of cycle collision
by increasing per­clique device counts
to distinct prime numbers. Several
specific use cases highlight the benefit
and need for ultralightweight com­
munication protocols.
There are limitations to S­CHIRP.
Networks generally change in ap­
pearance over time, and population
size and clique counts are likely to
vary over time. A complete long­term
solution must operate efficiently given
changes in both of these parameters.
Finally, the security implications of
a nonkeyed decentralized commu­
nication protocol must be taken into
account. Although inactive commu­
nications cause a drop in efficiency,
they do not pose a security threat like
invalid communications.
Many other problems exist across
M2M communication. One in partic­
ular is the conflicts that occur when
a device attempts to communicate
with another already active device.
Achieving communications between
specific IoT devices requires a strat­
egy for protecting the network from
connected devices of foreign intrud­
ers. Borowczak and Purdy (2018)
propose that S­CHIRP could provide
a layer of security for IoT swarm com­
munications by introducing a per­
mutation in round ordering.
Future experiments could also in­
clude the generation of new cliques
over time as new cliques of devices
enter a transmission space. In a re­
al­world situation, cliques may also
merge or distance themselves from
each other. Ultimately, there remains
a significant amount of work in de­
veloping minimal M2M protocols for
A serverless communication protocol lends itself
to future advances in decentralized
autonomous organizations.
lightweight IoT networks that are ca­
pable of maintaining secure clique
isolation while also minimizing com­
putational overhead.
Read more about it
• M. Borowczak and G. Purdy,
“S­CHIRP: Secure communication
for heterogeneous IoTs with round­
robi n protect ion,” i n Proc. 2018
IEEE Int. Conf. Consumer Electronics
(ICCE), 2018, pp. 1–6.
• W. Wang, K. Lee, and D. Mur­
ray, “A global generic architecture
for the future Internet of Things,”
Serv. Oriented Comput. Applicat.,
vol. 11, pp. 329–344, Sept. 2017.
• C. Bachmann, M. Ashouei, V.
Pop, M. Vidojkovic, H. De Groot, and
B. Gyselinckx, “Low­power wireless
sensor nodes for ubiquitous long­
term biomedical signal monitoring,”
IEEE Commun. Mag., vol. 50, no. 1,
pp. 20–27, 2012.
• E. Shi and A. Perrig, “Design­
ing secure sensor networks,” Wire-
less Commun., vol. 11, pp. 38–43,
Dec. 2004.
• S. P. Mohanty, U. Choppali,
and E. Kougianos, “Everything you
wanted to know about smart cities:
The Internet of Things is the back­
bone,” IEEE Consum. Electron. Mag.,
vol. 5, no. 3, pp. 60–70, 2016.
• P. Corcoran and S. K. Datta,
“Mobile­edge computing and the
Internet of Things for consumers:
Extending cloud computing and
services to the edge of the network,”
IEEE Consum. Electron. Mag., vol. 5,
no. 4, pp. 73–74, 2016.
• J. Wei, “How wearables inter­
sect with the cloud and the Internet
of Things: Considerations for the
developers of wearables,” IEEE Con-
sum. Electron. Mag., vol. 3, no. 3, pp.
53–56, 2014.
• K. D. Atherton. (2017, June).
The future of the air force is fighter
pilots leading drone swarms into
batt le. Popular Science. [Online].
Available: https://www.popsci.com/
IEEE PotEntIals
future­air­force­fighters­leading­
drone­swarms
• E. Aljarrah, M. B. Yassein, and
S. Aljawarneh, “Routing protocol of
low­power and lossy network: Sur­
vey and open issues,” in Proc. 2016
Int. Conf. Engineering MIS (ICEMIS),
2016, pp. 1–6.
• A. Castiglione, P. D’Arco, A. De
Santis, and R. Russo. (2017, Sept.).
“ S e c u re g roup com mu n ic at ion
schemes for dy namic heteroge­
neous distributed computing,” Fu-
ture Gener. Comput. Syst. [Online].
74, pp. 313–324. Available: http://
w w w.sciencedirect.com/science/
article/pii/S0167739X15003830
• S. Sattolo, “An algorithm to ge­
nerate a random cyclic permutation,”
Inform. Processing Lett., vol. 22, no. 6,
pp. 315–317, 1986.
about the authors
Shaya Wolf (swolf4@uw yo.edu)
earned her B.S. degrees in computer
science and mathematics at the
University of Wyoming, Laramie, in
2017. She is currently a student
pursuing her Ph.D. degree in com­
puter science at the University of
Wyoming Cybersecurity Education
and Research Lab, Laramie.
Rafer Cooley (rcooley2@uw yo
.edu) earned his B.S. degree in com­
puter science at the University of
Wyoming, Laramie, in 2017, where
he is currently a student pursuing
his Ph.D. degree in computer science.
He is a member of the University of
Wyoming Cybersecurity Education
and Research lab.
Mike Borowczak (mike.borowc
zak@uw yo.edu) earned his Ph.D.
degree in computer science and
engineering from the University of
Cincinnati, Ohio, in 2013. He is the
director of the Cybersecurity Edu­
cation and Research Center and
Lab and a faculty member in com­
puter science at the University of
Wyoming, Laramie.
Januar y/Februar y 2019 ■ 19
 Smart Consumer Electronics Systems
©iStockphoto.com/lvcandy

A neighbor trust-based
mechanism to protect
mobile networks
Gurveen vaseer, Garima Ghai, dhruva Ghai, and pushpinder S. patheja

M
obile nodes in a mobile
ad hoc network (MANET)
form a tempora l l i n k
between a sender and
receiver due to their con­
tinuous movement in a limited area.
This network can be easily attack­
ed because there is no orga ni zed
identity. This article discusses the
MANET, its various associated chal­
lenges, and selected solutions. As a
case study, a neighbor trust­based
security scheme that can prevent
ma licious attacks in a MANET is
discussed in detail. The security
scheme identifies each node’s behav­
ior in the network in terms of pack­
ets received and forwarded. Nodes
are placed in a suspicious range, and
if the security scheme detects mali­
cious function continuously, then it
is confirmed that the particular node
is the attacker in the network.

ManEt: a brief overview

A MANET is a collection of indepen­
dent nodes that are mobile and able
to communicate with each other ©Stockphoto.com/SaSinparakSa

wirelessly without the presence of

any centralized coordinator. Each
node functions as a host as well as a
router. Figure 1 shows the composi­
tion of a MANET where three differ­
ent, independent nodes communicate
with one another in varying environ­
Digital Object Identifier 10.1109/MPOT.2018.2850599
Date of publication: 28 December 2018
ments through another independent
node. MANETs are designed keeping
in mind the arbitrary organization of
electronic devices, such as mobile
phones and laptops, where the net­
work topology changes dynamically
and unpredictably. This unique fea­
ture of MANETs enhances the capa­
bility of stand­alone electronic devic­
es and enables their integration in
advanced applications, such as the
Internet of Things. Several existing
security solutions for wired networks
are ineffective and inefficient for a
MANET environment because the
transmission takes place in an open

20 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals 0278-6648/19©2019IEEE

space, making the MANETs more
susceptible to security attacks.

the challenges of a ManEt

Figure 2 depicts the classification
of malicious attacks in a MANET. Node C
Node A
Attacks can broadly be of two types:
passive and active. Passive attacks
do not disrupt the normal operation
Node B
of the network but prey on impor­
tant data in the network mainly
through eavesdropping and the
analysis of traffic that f lows in
the network, so that the attackers
gather me a n i n g f u l data. Active
attacks are carried out by mali­ FIG1 a representation of a manEt.
cious nodes that reside inside/out­
side the network to gat her data
and modify it. These attacks are
carried out by malicious nodes on Malicious Attacks on Networks
va r ious net working layers, i.e.,
physical, data link, transport, and
network, wherein each attack has
specific characteristics and behav­ Passive Attacks Active Attacks
ior that causes harm to the vulnera­
ble network.
Figure 3 gives an overview of the
importance of an attack­free en­ Network Transport Data Link/
Traffic
vironment for safe phones, smart Eavesdropping Layer Layer Physical
Analysis
Attacks Attacks Layer Attacks
grids, and industrial control systems
showing intrusion detection systems
(IDSs) and intrusion prevention sys­
tems in place. There are four major Sybil U2R Blackhole IP Probe SYN Vampire DoS
types of active attacks: denial of Spoofing Flooding
service (DoS), probing, vampire, and
user­to­root (U2R). In a DoS attack, FIG2 the taxonomy of malicious attacks on networks. ip: internet protocol;
the system is flooded with illegiti­ Syn: synchronize.
mate requests and responses, mak­
ing it inaccessible to legitimate users
and, hence, crashing the server. In
a probe attack, a device or an ac­ Cyberattacks
tivity is maliciously introduced into Intruder
the system, allowing it to gain ac­
cess and damage the system. In a
Safe Phones
U2R attack, the attacker gains con­
trol of the victim’s machine, acquir­
ing super­user privileges. A vampire
attack is a type of DoS attack that Safe Smart
Public IDS + Attack-Free Grids
consumes the node’s energy and Network IPS Environment
decreases the network’s efficiency
and reliability.
Safe
The importance of cybersecurity Industrial
in today’s world and the complexity Control
of security mechanisms have been Systems
Knowledge
widely discussed in the existing lit­ Base
erature. A scheme in which two MA­
NETs were merged using a method
ca lled merging using monitoring, FIG3 the importance of an attack-free environment. ipS: intrusion prevention system.

IEEE PotEntIals Januar y/Februar y 2019 ■ 21

 A MANET is a collection of independent nodes scheme that can detect probe, DoS,
vampire, and U2R attacks. The sec­
that are mobile and able to communicate urity mechanism checks its neigh­
with each other wirelessly. boring node for malicious activity,
i.e., whether the node deviates from
its specific behavior. The main aim
of the protocol is routing, which is
detection, and rehabilitation is avail­ trol systems has been presented. Fur­ designed to build and maintain a
able. An adaptive acknowledgment thermore, a comparison mechanism table of route information.
intrusion detection with node de­ for vampire attacks using routing The flow of a specific scheme is
tection techniques in MANETs has protocols on the network layer has shown in Figs. 4–6. A collection of
b e en proposed. A mu lt i l ayer de­ been performed. addresses of intermediate nodes be­
tection technique to identify DoS tween the sender and receiver is creat­
attacks by designing cross­layer a specific scheme ed. It is compared with receiver­nodes
architecture to control such attacks for secure ManEts addresses until there is a match.
is a lso available. Addit iona lly, a This section presents an intrusion The data are broadcasted from the
multiagent IDS in large switched prevention algorithm for a MANET sender to receiver using ad hoc on­
networks to protect industrial con­ that uses a neighbor trust­based demand distance vector (AODV) pro­
tocol. A preventer (watcher) node is a
special node that is capable of tracing
the established route, observing the
Start activity of nodes, and blocking the
malicious activity generator node
under the specified range. The pre­
M: Set of Nodes venter node uses two primitives: sus­
S: Sender Node P M, R: Receiver Node P M
picious and blocking. If the node’s be­
W: Watcher Node P M, I: Intermediate Node Addresses P M
havior is found to be abnormal (which
E: Energy of Node, C: Consumed Energy, eth: Energy Threshold
does not meet the functionality of
eth: Consumed Energy of All Nodes in Path/2
AODV: Routing Protocol, Rr: 550 Radio Range normal behavior), it is designated as
a suspicious node and under full sur­
veillance of the preventer node. Fur­
Execute Prevention Module thermore, if the activity of the node
W Watch Activity of Node in Rr is found to be malicious, that node is
blocked by the preventer node.
The preventer node plays an im­
Has
No portant role in watching all of the at­
I Updated ID
as R? tacks that have been mentioned. The
activity of a U2R attack is traced by
Yes
Step 2 data­forwarding criteria in which the
I Receive Data and Not intermediate node updates its self
Forward to Next Hop or Receiver identification (ID) as the receiver ID.
Data are not forwarded to the receiv­
er node, which drops by the LOOP
Is (a condition wherein the incoming
data drop No and outgoing node is the same, i.e.,
Yes
by loop? the node forwards itself to the next
node), implying that the preventer
node has identified the activity as a
Is U2R attack and, as a result, blocked
I Set as U2R Attack energy of current
No node > total energy
the attacker node.
Call (Block the Node,
consumed by previous A vampire attacker node con­
Record Time, Broadcast)
nodes in the sumes extra energy of the node. When
path? the preventer node finds that the re­
Stop Yes quired energy for the data transmis­
sion of a particular path is greater
Step 3
than the defined energy threshold,
it checks the energy consumption of
FIG4 a flowchart of the presented algorithm (step 1). each node in the path. W hen the

22 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals

node runs extra threads other than t he overhead. T he prevent ion 3) NRL probe attack and prevention
for data for warding or consumes scheme reduces overhead as the case: The data packet loss is min­
higher energy than all other nodes preventer node repairs the local imized by identifying attacker
combined, it is classified as a vam­ route in which the attacker re­ nodes and blocking them and
pire attack and is blocked by the sides and generates a new route channelizing a new path for the
preventer node. to the destination node. remaining nodes.
A probe attacker node behaves
correctly while the routing mod­
ule executes but later drops or cap­
tures the data of the source node. Step 2
This activity is traced by the pre­
venter node, and it blocks the at­ S Established Route by I Node to R
tacker nodes. A DoS attacker floods Send (Data, S, R)
the network with a number of junk
Does
packets to overload t he system I not match
Is
so that legitimate requests are not TCP/UDP headers
there a data drop No
fulfilled. The preventer node checks and packet is not network
or data capture
the data packet, and if it does not packet type?
and node not
match standard Transmission Con­ forwarded?
trol Protocol (TCP)/User Datagram Yes
Yes
Protocol (UDP) headers, it further
I Set as Probe Attack I Under Suspicious
checks the data rate. If the data Call (Block the Node, Record Time, Broadcast) W (Store I Data, Data Rate,
rate is higher than normal, it iden­ Number of Receivers, Time)
tifies the node as a faulty node and No
Stop
blocks it.
Is
Response of a ManEt data unknown
No and data rate > normal data rate
under the specific
and receiving time > standard
security scheme time pattern?
This section de s c r ib e s the res ­ I Set as DoS Attack
Call (Block the Node, Record Time,
ponse of the si mu l ated net work
Broadcast)
under conditions of attack as well
Yes
as after applying the prevention
mechanism. The responses report­
ed are throughput, normal routing FIG5 a flowchart of the presented algorithm (step 2).
load (NRL), and end­to­end delay.
An AODV protocol is considered
for routing between t he sender
and receiver. Step 3
1) NRL DoS attack and prevention
case: NRL is the ratio between
the number of routing packets W Check the In Activity
to the total data received. Fig­
ure 7(a) shows the performan­
ce of the network under a DoS Is
attack w ith a nd w ithout the In executing extra
prevention scheme. The scheme threads not for
communication? No
stops the unwanted congestion
delay by sea rching for a new
path for the nodes, thereby re­ Yes
ducing overhead a nd remov­ I Set as Vampire Attack
i n g u nw a nted f lood i n g f rom Call (Block the Node, Record Time, Broadcast)
the network.
2) NRL U2R attack and prevention
case: Figure 7(b) shows the net­ Stop
work performance with and with­
out the prevention scheme. U2R
results in f looding, enhancing FIG6 a flowchart of the presented algorithm (step 3).

IEEE PotEntIals Januar y/Februar y 2019 ■ 23

 The security scheme identifies attacker nodes tion scheme checks whether the
packet has updated itself as
through their malicious behavior and minimizes t he root node and, in spite of
packet dropping and routing overhead. being idle, is not forwarding the
data to the next hop. After ensur­
ing these conditions, if the data
4) NRL vampire attack and preven­ packets are tested for UDP/TCP packets are dropped, then that
tion case: For a vampire attack, behavior. If there is a delay in node is blocked and an alternate
the preventer node identifies the data rate, they are blocked path is devised.
the attacker node and blocks by the preventer node, and a new 7) Throughput analysis probe at ­
it. A new path is then detected in path for transmitting legal pack­ tack and prevention case: The
which resources do not consume ets is devised. Figure 7(c) shows performa nce of the a lgor it h m
increased energy, thereby reduc­ the performance of throughput against a probe attack is shown
ing NRL. in the DoS attack and preven­ in Fig. 8(b). Due to packet drop­
5) Throughput analysis DoS attack tion scenario. ping by the attacker, the receiver
and prevention case: The through­ 6) Throughput analysis U2R attack does not gain all of the data. The
put is measured in number of and prevention case: Figure 8(a) attacker wholly consumes the
packets transmitted per unit includes the throughput perfor­ data and also wastes the valu­
time. The throughput per for­ mance of the network under U2R able energ y resources of t he
mance is improved after apply­ att ack w it h a nd w it hout t he MANET. Packets are checked if
ing the prevention scheme, as secur it y scheme. The preven­ they are dropped or not forwarded.

30 7 2,000
25 6

Data Received (kb/s)

5 1,500
20
NRL (%)

NRL (%)

4
15 1,000
3
10
2 500
5 1
0 0 0
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100
Simulation Time (s) Simulation Time (s) Simulation Time (s)
NRL-DoS NRL-U2R Throughput-DoS
NRL-DoS-Prevention NRL-U2R-Prevention Throughput-DoS-Prevention

(a) (b) (c)

FIG7 the nrl and throughput (doS). the (a) nrl for doS, (b) nrl for U2r, and (c) throughput for doS.

1,200 1,400 2,000

Data Received (kb/s)

Data Received (kb/s)

Data Received (kb/s)

1,000 1,200
1,000 1,500
800
800
600 1,000
600
400
400 500
200 200
0 0 0
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100
Simulation Time (s) Simulation Time (s) Simulation Time (s)
Throughput-U2R Throughput-Probe Throughput-Vampire
Throughput-U2R-Prevention Throughput-Probe-Prevention Throughput-Vampire-Prevention
(a) (b) (c)

FIG8 the throughput (U2r, probe, vampire) analysis. the throughput for (a) U2r, (b) probe, and (c) vampire.

24 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals

 TABLE 1. A network performance summary.
PARAMETERS
pdF (% data received)
dropped data (number of packets)
average end-to-end delay (μs)
If no specific reason is provid­
ed for these actions, the nodes
are blocked.
8) Throughput analysis vampire at­
tack and prevention case: It is
checked that no extra threads
are used for communication in
the path. Also, the times of previ­
ous and current nodes are calcu­
lated, and if it is found that the
time taken by the node to move is
more than that of its threshold,
then it is identified as an attack­
er and blocked. Figure 8(c) de ­
picts the throughput analysis
in the vampire attack and pre­
vention scenario.
9) Average end­to­end delay analy­
sis: Average end­to­end delay is
the average time taken to transmit
packets from the sender to receiver
nodes. End­to­end delay depends
on queue delay, bandwidth, pro­
cessing, and contention. Table 1
shows the average end­to­end
delay in different attacks as well as
the prevention scenario. The pre­
vention methodology minimizes
delay by providing a secure and
efficient path. The delay in the
normal case is 0.33 μs. Table 1
shows the network performance
under various attacks and their
prevention schemes.
Conclusion
We present an algorithm to prevent
probing, DoS, vampire, and U2R
attacks. The security scheme identi­
fies attacker nodes through their
malicious behavior and minimizes
packet dropping and routing over­
head. The security scheme first con­
siders nodes in the suspicious range.
Nodes that demonstrate malicious
functions are defined as attackers.
These attacker nodes are blocked
DoS ATTACK/ VAMPIRE ATTACK/ PROBE ATTACK/
PREVENTION PREVENTION
81.6/89.76 82.11/90.97
946/763 698/503
0.52/0.51 0.45/0.36
after confirmation in the network.
For future research, this work will
be tested in different network sizes
to check scalability. We will also look
at preventing multiple attacks in the
network simultaneously.
Read more about it
• M. H. Davis, U. Lang, and S.
Shetye, “A cybermodel for privacy
by design: Building privacy pro­
tection into consumer electronics,”
IEEE Consum. Electron. Mag., vol. 4,
no. 1, pp. 41–49, 2015.
• M. L. Rajaram, E. Kougianos,
S. P. Mohanty, and U. Choppali,
“Wireless sensor network simula­
tion frameworks: A tutorial review,”
IEEE Consum. Electron. Mag., vol. 6,
no. 2, pp. 63–69, 2016.
• D. R. Raymond and S. F. Mid­
kiff, “Denial­of­service in wireless
sensor networks: Attacks and de­
fenses,” Pervasive Comput., vol. 7,
no. 1, pp. 74–81, 2008.
• J. Pullagura and V. Rao, “Sim­
ulation­based comparison of vam­
pire attacks on traditional MANET
routing protocols,” Inform. Commun.
Technol. Sustainable Develop., vol.
9, pp. 501–509, Nov. 2017.
• A. Alsumayt, J. Haggerty, and
A.Lotfi, “Detect DoS attack using
MrDR method in merging two MA­
NETs,” in Proc. IEEE Int. Conf. Ad­
vanced Information Networking and
Applications Workshops, 2016, pp.
889–895.
• J. Vuorinen and P. Tetri, “Par­
adoxes in information security,”
IEEE Potentials, vol. 35, no. 5, pp.
36–39, 2016.
• S. S. Roy, D. Puthal, S. Sharma, S.
P. Mohanty, and A. Y. Zomaya, “Build­
ing a sustainable Internet of Things,”
IEEE Consum. Electron. Mag., vol. 7,
no. 2, pp. 42–49, 2018.
IEEE PotEntIals
U2R ATTACK/
PREVENTION PREVENTION NORMAL
82.02/90.06 26.67/88.27 90.53
855/828 1,163/739 780
0.52/0.46 0.65/0.63 0.33
• A. A l­Roubaiey, T. Sheltami,
A. Mahmoud, E. Shakshuki, and
H. Mouftah, “A ACK: Adaptive ac­
knowledgment intrusion detection
for M A NE T w ith node detection
en hancement,” in Proc. IEEE Int.
Conf. Advanced Information Net­
working and Applications, 2010, pp.
634–640.
• R. Shrestha, K. H. Han, D. Y.
Choi, and S. J. Han, “A novel cross
layer intrusion detection system
i n M A N E T,” i n P ro c . I E E E In t .
Conf. Advanced Information Net­
working and Applications, 2010,
pp. 647–654.
• Y. Gong, Y. Fang, L. Liu, and
J. Li, “Multi­agent intrusion detec­
tion system using feature selection
approach,” in Proc. IEEE Int. Conf.
Intelligent Information Hiding and
Multimedia Signal Processing, 2014,
pp. 528–531.
about the authors
Gurveen Vaseer (gurveenv@oriental
university.in) is a research scholar
at the Oriental University Indore,
India, who has been pursuing re ­
search on cybersecurity since 2016.
Garima Ghai ( ga r imag ha i@
orientaluniversity.in) is a professor
with the Department of Computer
Science and Engineering at Oriental
University Indore, India.
Dhruva Ghai (d h r u v a g h a i@
orientaluniversity.in) is a professor
with the Department of Electronics
and Communication Engineering at
Oriental University Indore, India.
Pushpinder S. Patheja (pspatheja@
gmail.com) is a director of student
welfare, and associate professor in
the School of Computing Science
Engineering at the Vellore Institute
of Technology, Bhopal, India.
Januar y/Februar y 2019 ■ 25
 Smart Consumer Electronics Systems
©iStockphoto.com/lvcandy
T
his article introduces the
concept of proof of authen­
tication (PoAh) for the light­
weight implementation of
blockchains in the Inter­
net of Things (IoT). The PoAh can
replace existing consensus algo­
rithms, such as proof of work (PoW),
proof of stake (PoS), and proof of ac­
tivity (PoA), for resource­ and energy­
Digital Object Identifier 10.1109/MPOT.2018.2850541
Date of publication: 28 December 2018
26 ■ J a n u a r y / F e b r u ar y 2019
Proof of authentication:
IoT-friendly blockchains
deepak puthal and Saraju p. mohanty
constrained infrastructures, such
as the IoT.
Blockchain in the Iot:
the challenges
The IoT is based on the vision to
connect physical devices to the In­
ternet and access remote data to
control the distanced physical world.
The building blocks of the IoT are
objects or embedded devices that
can be connected to the Internet.
The IoT includes sensing, computa­
IEEE PotEntIals
Blue chainS—©iStockphoto.com/cyBrain, White chain—©iStockphoto.com/3dSculptor
tion, communication, identification,
and semantics. A critical require­
ment of the IoT is that the things in
the network must be interconnected.
Implement ing blockcha ins in
the IoT to secure the infrastructure
in a distributed ma nner is a big
challenge (Fig. 1). Consider ing the
resource­constraining devices of the
IoT, blockchain implementation looks
impossible due to the energy re ­
quirement for PoW. However, PoW is
the backbone for blockchains, and,
0278-6648/19©2019IEEE
without it, blockchains cannot have a
distributed form. As a result, the cur­
rent blockchain technology cannot be
applied to the IoT.
Due to the wide range of IoT ap­
plications in the modern world and
blockchains as the only distr ib­
uted security architecture, we can­
not ignore the intersection of these Da
Transition History Stored in a Block of Data
technologies. IoT applications focus
on energy­efficient computing and
1 2 3 n
real­time decision making. This mo­
tivates us to explore lightweight con­
sensus algorithms as a blockchain
transaction verification and valida­
tion method for the IoT. FIG1 the blockchain transactions network.

PoW
Miners within a bitcoin network must
maintain and record on the same dis­
tributed ledger to secure bitcoins and
maintain consistency. However, with
millions of decentralized nodes and
no central server to maintain the net­
work, how can this be done? Bitcoin’s
solution to this problem is PoW. In
traditional transaction methods, a
trusted third party is always involved
to keep a transaction record to main­
tain the balance. However, the decen­
tralized trustless consensus main­
tains the transaction without the
help of third­party services. With the
concept of bitcoin or digital curren­
cies, individual users in the network
hold the distributed ledgers, i.e., the
blockchain (Fig. 2). The users can
track the transaction information
without the help of a trusted third
party. They do not need an authenti­
cation by a third party to validate
their transactions.
Distributed ledgers require an ex­
pensive computational calculation to
solve mathematical puzzles to vali­
date the trustless transactions. The
expensive computational calculation is
also known as a miner, and the process
of validating the transactions is called
PoW. The miner has two important
roles: 1) validate the transactions by
avoiding potential network threats
and 2) calculate reward points. Blocks
contain a number of transactions, and
miners apply PoW to evaluate in­
dividual transactions, as shown in
Fig. 2. The miner that solves the block
first gets a reward point. To achieve
Prev-Hash PoW
Trx-1 Trx-2 … Trx-n
Block – i
Proof
(Include Bits)
Hash
FIG2 the poW transactions. prev-hash: previous hash; trx: transaction.
reward points, the miners compete
with each other to solve the mathemat­
ical problem. After finding a solution,
the miner broadcasts to the network
to update the blockchain and receive
a reward, i.e., cryptocurrency. In a
real­world problem, the mining pro­
cess is essentially an inverse to a hash
function. In the standard blockchain,
the parameters update fortnightly, and
new blocks generate every 10 min.
PoW works on a distributed con­
sensus basis, where a miner needs a
great deal of energy. In a single bitcoin
transaction, electricity consumption is
equivalent to the household electric­
ity use of 1.5 households for one day
in the United States. By 2020, the bit­
coin transaction network will consume
close to the amount of electricity used
in a country the size of Denmark. Due
to the computational and economic
difficulties of PoW, several applications
use a slight modification of PoW to use
a blockchain. The concept of PoW is not
Prev-Hash PoW
Trx-1 Trx-2 … Trx-p
Block – i + 1
Challenge
(Prev-Hash)
only used for bitcoins but also for sev­
eral other applications, including ethe­
reum. One of the common terms for
the algorithm used by ethereum is PoS.
Pos
The basic concept of PoS is proof of
the ownership of digital currency
from PoW. Coinage does not play a
crucial role for bitcoins, whereas the
PoS concept originated for bitcoins
to help prioritize the transactions.
PoS gives the same level of confiden­
tiality, the most critical requirement
for monetary transactions.
The PoS transaction, also known
as coinstake, introduces a new type of
block (Fig. 3). The term kernel is the
Kernel Input Stake Output
Stake Input (Pay to Stake
Stake Input Owner)
FIG3 the structure of poS (coinstake)
transactions.

IEEE PotEntIals Januar y/Februar y 2019 ■ 27

first input of coinstake and required to
meet the target hash protocol and ap­
ply a stochastic process to generate
PoS blocks. Coinstake follows limit­
ed search space to compute hash val­
ue instead of unlimited search space
like PoW. As a result, PoS is energy
efficient compared to PoW. The hash
target that the PoS kernel must meet
is a target per unit coinage consumed
in the kernel. This system ensures
that miners who have not been able to
provide a solution to the cryptograph­
ic puzzle will have a higher chance of
Prev-Hash
Trx-1 Trx-2
Block – i
FIG4 the poah in block transactions.
Cloud
Edge
Fog
creating a block as the coinage of their
currency increases.
Poa
In the PoS scheme, coins continue to
grow in coinage even if the partici­
pant is offline. This creates a prob­
lem where miners go online every
several weeks, create blocks, and
then go offline afterward. This signif­
icantly reduces the number of con­
currently online nodes, reducing the
security of the coin. Another problem
with this scheme is that it discourag­
PoAh Prev-Hash PoAh
… Trx-n Trx-1 Trx-2 … Trx-p
Block – i + 1
es the exchange of the coin, as each
exchange would reset the coinage,
leading to hoarding of the currency.
PoA was designed to reduce the
impact of these problems. The idea
is to reward a fraction of the PoW re­
ward to active peers, where having a
higher stake in the currency increases
the chances of winning the reward. A
miner who solves a PoW puzzle broad­
casts the solution to the network. Ev­
ery node receiving the block is able
to derive a number of ticket numbers
from the solution. A peer who owns
the coin corresponding to the ticket
will put a signature onto the block and
be able to receive a part of the reward.
A peer who is offline will not be able to
sign the block and, therefore, will not
receive the reward. The PoA scheme
aims to incentivize peers to remain
on the blockchain, thereby improv­
ing its security by using this award
mechanism. This scheme also makes

Prev-Hash PoAh

Prev-Hash PoAh
Trx-1 Trx-2 … Trx-p

Hash PoAh Trx-1 Trx-2 … Trx-p

T1 T2 T3

Hash PoAh
Hash PoAh
T1 T2 T3
T1 T2 T3
IoT

Hash PoAh Hash PoAh

T1 T2 T3 Hash PoAh
T1 T2 T3 Hash PoAh
T1 T2 T3
Hash PoAh T1 T2 T3

T1 T2 T3 Hash PoAh
Hash PoAh
T1 T2 T3
T1 T2 T3

FIG5 the blockchain in fog computing.

28 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals

spending of the currency less punish­
ing, as holding onto currency for long
periods of time does not improve the
probability of receiving a reward.
Poah for the Iot
The consensus algorithms, such as
PoW, PoS, and PoA for blockchains,
are introduced for bitcoin applica­
tions. The presented PoAh follows a
traditional blockchain working model
with lightweight block verification.
The initial step of the miner in a
network is to validate the block, fol­
lowed by evaluation of the hash value,
whereas PoAh aims to authenticate
the blocks following the same trans­
action method as blockchains (Fig. 4).
The miners can be trusted nodes of
the network and used for authentica­
tion. All of the network nodes must
maintain and record on the same
distributed ledger, and they can track
the transaction information.
The trusted nodes in the network
authenticate the blocks to add into
the distributed ledger. This includes
two steps for authentication: 1) au­
thenticate the block and source of the
block and 2) upon validating the au­
thenticated block by trusted nodes,
increase the trust value by one unit
for those who have authenticated the
block first. This is followed by all of
the nodes in the network updating
the distributed ledger. In this pro­
cess, individual transitions are veri­
fied from the block. Any miner who
performs false authentication loses
a unit of trust value and becomes
a normal node in the network after
a certain number of false authenti­
cations. PoAh can avoid the inverse
hash computation for energy­efficient
distributed secure communications
and computing in the IoT.
Fog computing is integrated with
the IoT for scalable computing and
communications. Fog computing inte­
grates clouds and edge infrastructure
with the IoT, where the cloud works as
the fully trusted infrastructure to store
security properties. Blockchains in fog
computing can bring a novel architec­
ture for storing IoT device authentica­
tion properties in a trusted cloud and
keeping their references on edge devic­
es to work as a miner in the network to
TABLE 1. A comparison of different consensus
algorithms of blockchain.
PoW
energy consumption high
computation requirements high
latency high
Search space high
evaluate PoAh (Fig. 5). The edge devices
work as a partially trusted device. This
will maintain the decentralized secu­
rity framework in the network.
Traditional IoT device deployment
can be followed to register devic­
es with fully trusted parts, such as
the cloud (Fig. 5). PoAh can also be
integrated with these concepts for end­
to­end secure infrastructure building.
A comparison classification of dif­
ferent blockchain consensus algo­
rithms and their properties is pre­
sented in Table 1.
Conclusion
PoAh removes the reverse hash func­
tion from PoW to lightweight the
process. As a result, blockchains can
efficiently integrate with resource­
constrained networks such as the IoT
and related applications. This also
works efficiently in hierarchical net­
work and fog computing scenarios.
In the future, we aim to implement
PoAh into a real­time IoT and fog test
bed to measure the overall perfor­
mance and efficiency of the network.
Read more about it
• D. Puthal, N. Malik, S. P. Mohan­
ty, E. Kougianos, and C. Yang, “The
blockchain as a decentralized security
framework,” IEEE Consum. Electron.
Mag., vol. 7, no. 2, pp. 18–21, 2018.
• D. Puthal, N. Malik, S. P. Mohan­
ty, E. Kougianos, and G. Das, “Every­
thing you wanted to know about the
blockchain,” IEEE Consum. Electron.
Mag., vol. 7, no. 4, pp. 6–14, 2018.
• C. Dwork and M. Naor, “Pric­
ing via processing or combatting
junk mail,” in Proc. Annu. Int. Cryp-
tology Conf., 1992, pp. 139–147.
• M. Jakobsson and A. Juels,
“Proofs of work and bread pudding
protocols,” in Proc. Secure Informa-
tion Networks, 1999, pp. 258–272.
IEEE PotEntIals
PoS PoA PoAh
high high low
high high low
high high low
low na na
• S. Nakamoto. (2008). Bitcoin: A
peer­to­peer electronic cash system.
www.bitcoin.org. [Online]. Available:
https://bitcoin.org/bitcoin.pdf
• Investopedia. (2018). Proof of
stake. [Online]. Available: https://
w w w.investopedia.com/terms/p/
proof­stake­pos.asp
• S. King and S. Nadal. (2012).
Ppcoin: Peer­to­peer crypto­currency
with proof­of­stake. [Online]. Avail­
able: https://peercoin.net/assets/
paper/peercoin­paper.pdf
• F. Tschorsch and B. Scheuer­
mann, “Bitcoin and beyond: A tech­
nical survey on decentralized digital
currencies,” Commun. Surveys Tuts.,
vol. 18, no. 3, pp. 2084–2123, 2016.
• N. N. Kshetri, “Can blockchain
strengthen the Internet of Things?”
IEEE IT Prof., vol. 19, no. 4, pp. 68–
72, 2017.
• A. Al­Fuqaha, M. Guizani, M.
Mohammadi, M. Aledhari, and M.
Ayyash, “Internet of Things: A sur­
vey on enabling technologies, proto­
cols, and applications,” IEEE Com-
mun. Surveys Tuts., vol. 17, no. 4, pp.
2347–2376, 2015.
• D. Puthal, M. S. Obaidat, P. Nan­
da, M. Prasad, S. P. Mohanty, and A. Y.
Zomaya, “Secure and sustainable load
balancing of edge data centers in fog
computing,” IEEE Commun. Mag., vol.
56, no. 5, pp. 60–65, 2018.
about the authors
Deepak Puthal (deepak.puthal@uts
.edu.au) is a lecturer (assistant pro­
fessor) in the faculty of engineering
and information technologies at
the University of Technology Syd­
ney, Australia.
Saraju P. Mohanty (s a r aju
.mohanty@unt.edu) is a professor at
the University of North Texas, Denton.
Januar y/Februar y 2019 ■ 29
©iStockphoto.com/lvcandy
Smart Consumer Electronics Systems
D
rone technology that has
the potential to disrupt and
augment our quality of life
is swiftly evolving. Drones
are rapidly growing in po­
pularity and are used in various
applications, such as agriculture,
emergency response, border control,
asset inspection, intelligent trans­
portation, and many more areas.
This is primarily due to the rapid
Digital Object Identifier 10.1109/MPOT.2018.2850386
Date of publication: 28 December 2018
30 ■ J a n u a r y / F e b r u ar y 2019
Drones: Augmenting
our quality of life
christos kyrkou, Stelios timotheou, panayiotis kolios,
theocharis theocharides, and christos panayiotou
advances in mobile embedded com­
puting that allow for various sen­
sors and controllers to be integrat­
ed into drone platforms, enabling
them to sense and understand both
their internal state and the exter­
nal environment. In this article,
we showcase relevant drone tech­
nologies, explore research opportu­
nities, and demonstrate through
three use cases how research can
drive these disruptive systems for­
ward within our social, economic,
and scientific activities.
IEEE PotEntIals
©iStockphoto.com/Jelena83
Droning on
The term drone is h a p h a z a r d l y
thrown around by many outlets and
ubiquitously appears in numerous
conversations between consumers,
whether it is related to their holiday
shopping lists, hobbies, jobs, or even
when they discuss the news. In its
original context, a drone refers to
an unmanned aircraft either pre­
programmed with a flight plan or
remotely controlled. The term, in
fact, originates from its two literal
meanings: specifically, male bees or
0278-6648/19©2019IEEE
a m on ot onous bu z z i ng sou nd,
for which multirotor drones are
now famous.
More recently, drone has evolved
into a catch­all term that means any
unmanned robot. This includes ro­
bots designed for water, land, and
air use. In the context of this article,
however, we strictly refer to aerial
drones. Whether called unmanned
aerial vehicles (UAVs) or remotely pi-
loted aircraft systems, their formal
name, drones are becoming more
mainstream. Although they are still
considered to be in their infancy
stage in terms of mass adoption and
usage, they have already surpassed
rigid traditional barriers in indus­
tries that otherwise seemed impen­
etrable by similar technological in­
novations and advancements.
Furthermore, during the past few
years, drones have become central
to the functions of various business­
es a nd gover n ment a l orga n i z a ­
tions and managed to pierce through
areas where certain industries were
either stagnant or lagging. More im­
portantly, however, they are now
considered among the most disrup­
tive technologies that are expected to
augment our quality of life.
Typically, drones are classified as
fixed wing (with longer range, higher
altitude flight capabilities, and less
power consumption) or multirotor
(better hovering and maneuvering
abilities but lesser range and alti­
tude and more power consumption).
Just like other disrupting technolo­
gies, the military has initially been
the major benefactor of drones.
However, once flight technologies
advanced to the point where multiro­
tor drones became extremely easy to
operate, able to maintain strict safety
standards, and equipped with sen­
sors and other electronically assisted
safety features that warranted their
safe passage over multiple environ­
mental conditions, the civilian sector
started adapting them. In fact, the
photo­ and video­producing indus­
tries were the first beneficiaries of
this innovative technology. Further­
more, a gradual reduction in their
price has only magnified this global
phenomenon. Drones are now listed
Whether called unmanned aerial vehicles or
remotely piloted aircraft systems, their formal
name, drones are becoming more mainstream.
as toys, tools, hobby items, entertain­
ment, and gadgets and have long left
the photo and video industry as their
primary commercial application.
The commercial usage of drones
is gaining steady momentum, and
multiple industries are w o r k i n g
with drones as part of their daily
regula r busi ness funct ions. It is
estimated that the market for com­
mercial and civilian drones will ex­
pand at a compound annual growth
rate of 19% between 2015 and 2020,
compared with a 5% growth on the
military side, according to BI Intel­
ligence, Business Insider’s premium
research service.
Multirotor drones have dominated
the commercial market in the last few
years, mainly due to their affordabil­
ity and ease of taking off and land­
ing, whereas fixed­wing drones fea­
ture different characteristics and, as
such, target specific high­end mar­
kets beyond the military and come at
a hefty price. Multirotor drones fea­
ture characteristics that are suitable
for a wider range of applications, such
as mapping and surveying and asset
inspection, due to their ability to hov­
er, maneuver in narrow and difficult
terrain conditions, and take off and
land in very limited space.
Gradually, onboard sensors, such
as gyroscopes, accelerometers, mag­
netometers, and altimeters, have been
integrated into the drone’s flight­con­
trol computer. Drone flight capabili­
ties have, therefore, been augmented
to accommodate various application
needs. Coupled with customizable
computing solutions that can run
various data analysis and pattern­
recognition algorithms, a drone is now
an intelligent, partially autonomous,
and partially self­aware system that
can function either individually or as
part of a swarm.
To showcase the potential of these
systems and present the inherent
research challenges and emerging
opportunities, we provide a brief
IEEE PotEntIals
overview of the technology that pow­
ers these intelligent drones. We also
attempt to propose a road map and
directions toward identifying current
challenges and opportunities that re­
search in intelligent drones is facing.
Furthermore, we showcase how our
own work is currently developing in­
telligent drones to be used as tools for
first responders in emergency situa­
tions, monitor power lines, and oper­
ate as a large­scale, real­time traffic­
monitoring tool.
technological considerations
for drone applications
Sensor technologies have greatly
advanced through the years and are
often the secret sauce inside drones
that enable their use in numerous
remote­sensing applications (Fig. 1).
These sensor technologies provide
drones with the ability to sense and
understand their environment and
become self­aware of their surround­
ings and the context that they are
flying in to potentially take evasive
and reactive actions.
A drone platform can carry a va­
riety of payload equipment consisting
of relatively simple subsystems (such
as a fixed camera with fixed lenses
having a mass as little as 200 g) to
more advanced payloads (such as a
video camera system with adaptive
zoom support and pan and tilt func­
tions). Larger drone platforms can
carry a combination of different types
of sensors (e.g., thermal imagers and
radar scanners, among others) to col­
lect various data elements that can be
processed and analyzed in real time
to provide enhanced capabilities.
sensors that aid the
drone in flying
These sensors are primarily used to
assist in the drone flight by measur­
ing the state of the drone and guiding
its motion with respect to the environ­
ment, such as its position in space,
velocity, and orientation.
Januar y/Februar y 2019 ■ 31
 A drone is now an intelligent, partially autonomous,
and partially self-aware system that can function
either individually or as part of a swarm.
■■Accelerometer: Accelerometers
are used to determine the posi­
tion and orientation of the drone
in flight.
■■Gyroscope: The gyroscope detects
angular velocity in three axes, so
it can estimate the rate of change
of angle in pitch, roll, and yaw.
■■Digital compass: The compass
provides the drone with a sense
of direction. It provides magnetic
field data in three dimensions.
■■Barometer: The barometer helps
the drone with navigation and
measures the altitude.
■■Global Positioning System (GPS):
The GPS sensors offer access to
location information from geosta­
tionary satellites. At least three
satellites are needed to determine
longitude and latitude coordi­
nates and one satellite to deter­
mine the altitude of the UAV.
Typically, satellite data is fused
with other readings to provide a
higher level of accuracy, precision,
and stabilization.
sensors for environment
perception and data collection
External sensors and payloads can
be integrated into a drone to enhance
its understanding of the environ­
ment, collect data for offline pro­
cessing, and make decisions on how
to navigate and act in missions.
■■High-definition cameras: Most
drones are equipped with camer­
as for providing intelligence
about particular situations, espe­
cially because drones can hover
over remote areas that are diffi­
cult for people to access. Camera
resolution, frame rate, and field
of view are important parameters
for each application. Camera sen­
sors can be used to provide com­
puter vision and scene analysis
to the drone.
■■Thermal sensors: Thermal vision
cameras construct images or video
from heat (infrared thermal radia­
tion), invisible light that is also
part of the electromagnetic spec­
trum. Most thermal cameras pro­
duce a video output in which color
encodes the temperature reading
of the scene. Thermal energy is
radiated off nearly every material,
hence, thermal vision cameras can
capture images and temperature
variations emitted from people and
objects and are used for a wide
range of applications.
■■Lidar: Lidar is a sensing technolo­
gy where the environment is usu­
ally scanned with a pulsed laser
beam to measure the relative dis­
tance of the objects detected by
the sensor. Thus, the drone can
be enhanced with obstacle detec­
tion and avoidance for safe navi­
gation during both day and night.
Potential drawbacks include the
high cost and bulky form factor.
■■Radar: Radar systems work by
sending out radio waves and using
the echoes that bounce back to
create topographical maps or avoid
obstacles. In contrast to lidar and
cameras, they can be designed to
rapidly scan large areas and are
less affected by weather, smoke,
and dust. However, the data col­
lected by radar sensors are only
accurate up to a certain extent and
cannot distinguish between differ­
ent classes of objects.
■■Telecommunication sensors:
Drones equipped with telecommu­
nication hardware such as Wi­Fi
antennas may provide clues about
someone’s position. In emergen­
cy situations, such equipment
can also be used on the drone to
quickly re­establish cellular signal
coverage over areas that experi­
ence an outage.
■■Electronic sniffers: Sniffers are
devices that sample air and provide
information about nuclear, toxic, or
any kind of chemicals existing in
the area to monitor pollution.
The different drone payloads used
for each application can have an
impact on the drone characteristics
and requirements. Drones can have
a very limited payload capacity that

(a) (b) (c)

FIG1 examples of drone sensor technologies. (a) a high-definition camera equipped on a drone is the most standard payload used
today. (b) drone obstacle avoidance systems with ultrasonic sensors and cameras are becoming more widespread. (c) lidar sensors
with reduced cost will be a more frequently used payload in future drones.

32 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals

largely depends on the power of the
motors and propeller combination,
as the weight of the payload can neg­
atively affect the flight time. In addi­
tion, the capacity of the drone’s main
battery plays an important role when
the payload needs to be powered by
the drone itself, thus compromising
the overall flight time.
Another choice to consider is the
drone’s level of autonomy, which can
vary from fully autonomous operation
to fully controlled by a remote pilot. The
fully autonomous operation requires a
multitude of sensors to observe and
understand the environment as well
as a suite of algorithms to interpret
the data from the sensors. The drone
would have to make decisions about
where to fly and how to classify objects
captured by its camera.
In addition, there needs to be suf­
ficient computing power for real­time
processing on the edge on tiny, light
computers. Alternatively, process­
ing can be offloaded to the cloud.
However, this presents a tremendous
challenge on the communication links
both in terms of bandwidth and pri­
vacy/security, limiting the application
spectrum to nonlatency­critical tasks.
Although, when the flight and plan­
ning are done by a human operator,
the requirements can be relaxed and,
hence, the drone can be equipped
with only the necessary application­
specific sensors. As it becomes cheap­
er and easier to customize commercial
drones, by utilizing emerging innova­
tive embedded and mobile sensing
and computational solutions, new
opportunities will arise to facilitate re­
search and innovation in a wide array
of niche applications.
Research opportunities
and challenges
In these early stages, research has
focused on the low­level controls that
could be applied by resource­con­
straint onboard computers for alti­
tude stabilization and the maneu­
vering of individual drones. Over
time, additional effort has been put
into more advanced tasks, including
indoor localization and navigation.
To achieve these tasks, research has
focused on optimizing embedded
A drone platform can carry a variety of payload
equipment consisting of relatively simple
subsystems to more advanced payloads.
system algorithms for remote sens­
ing, digital signal processing, and
fault­tolerant control, all coupled
together to achieve good real­time
performance.
Furthermore, as cameras are now
standard equipment in most drones,
a large body of research is focused on
utilizing advanced computer­vision
algorithms to make drones fly auton­
omously, track a target through ob­
stacles, or deliver a payload to a cer­
tain location. These technologies have
matured to a level that has now made
it possible to investigate how complex
missions can be achieved either by a
single drone or a swarm working in
collaboration. Coordinated teams of
drones can attain significantly greater
capabilities than what is possible by a
single agent, and, thus, this research
track has recently received consid­
erable attention. Broadly speaking,
these missions can be classified into
the following four main categories
(Fig. 2): data collection and surveying,
monitoring and tracking, temporary
infrastructure, and delivery of goods.
A classic problem that arises in
surveying and collecting data over a
spatial field of interest is how to cover
the area faster with the least number
of resources. Research challenges in­
clude task assignment, scheduling,
and path planning, considering the
physical resource limitations of the
drones including the total number of
drones, their battery levels, and their
communication ranges. Problems
that arise in monitoring and track­
Data Collection Monitoring and
and Surveying Tracking
FIG2 the four main categories of drone-based mission types.
ing missions consider the derivation
of accurate models of the underlying
dynamics and investigate strategies
that generally seek to minimize
the tracking error for all detected
events/objects.
Drone deliveries have the po­
tential to significantly reduce the
cost and time of making last­minute
goods shipments. Research in this
domain has focused on the manage­
ment of job allocations to minimize
the overall cost subject to delivery
time constraints taking into con­
sideration the limitations in battery
and payload capacity. Drone deploy­
ments have also been considered to
provide over­the­top services includ­
ing the setup of an aerial communi­
cation platform, illuminate fields for
night operations, and make voice
announcements across the horizon.
To accomplish this, problems of node
placement for coverage and topology
control are being looked at to address
the communication and computation
challenges that arise.
In all mission types, the ability of
a drone to analyze visual data and
determine its location and next ac­
tion can have tremendous benefits
and increase its capabilities. Be­
yond the use of computer­vision
algorithms for object detection and
recognition tasks, another area of
focus is vision­based navigation and
control. Systems that rely on the GPS
can achieve little better than 3­m
resolution, which is not adequate
for urban areas or indoor settings. In
Temporary Delivery
Infrastructure of Goods

IEEE PotEntIals Januar y/Februar y 2019 ■ 33

extreme cases, drones could be sent
into earthquake­damaged buildings
to look for survivors, which would re­
sult in avoiding all sorts of obstacles
and moving into free space without
relying on external connectivity. To
accomplish these feats, a drone re­
quires a complex system of cameras
as well as advanced algorithms to
figure out where it is, where the ob­
stacles are, and how to navigate this
area based on its understanding of
the environment.
In carrying out any of the main
mission categories, recent research
has also demonstrated that swarms
of drones can be more effective. Data
collection and search­and­rescue
missions are faster and more efficient
with a team of drones to pool data and
provide redundancy in case some ma­
chines fail. But the use of more vehi­
cles also adds complexity, and drones
working together must be able to
communicate with one another and
make collective decisions.
It should also be emphasized that
many of these problems are inter­
related and, thus, need to be jointly
investigated to optimize mission ob­
jectives. Searching for and tracking
objects across the field, e.g., could re­
sult in competing objectives that need
to be resolved in an intelligent way.
Clearly, searching is necessary to de­
tect events/objects before accurately
tracking them. At the same time, a
drone may need to abort tracking a
particular object if exploration could
reveal a new set of objects in the
area. We focus our research effort on
investigating problems that involve
such complex mission objectives and
seek to devise intelligent solutions
(a)
FIG3 vehicle detection using drones. (a) vehicles detected using motion, (b) vehicles detected using still images, and (c) combined
detection of vehicles.
34 ■ J a n u a r y / F e b r u ar y 2019
that approach these problems in a
holistic perspective.
The success of drone deployments
across the aforementioned applica­
tion categories has created safety
and security issues not just to the
aviation sector but also every other
aspect of social life. There are great
challenges in effectively addressing
these issues, mainly due to the fast
evolution of technology. Current ef­
forts focus mainly on regulating op­
eration through both legislative and
technological means. As part of the
envisioned technology, which is our
primary focus in this article, novel
air­traffic control solutions are be­
ing researched to monitor and con­
trol flights, and various ground and
aerial systems are being considered
for intercepting rogue drones.
Case studies of
drone applications
We are currently working on several
interesting projects using drones.
Based on our work, we present three
cases that illustrate the intrigu­
ing and challenging applications of
drones in real­world problems: 1) road
transportation network monitoring,
2) power­line inspection, and 3) civil­
defense protection.
Road transportation
network monitoring
Monitoring the state of road networks
is fundamental for efficient traffic
management. Current sensing tech­
nologies provide different types of
measurements, such as veh icle
counts at a single point, traffic den­
sity over a road stretch, and speed and
location from moving vehicles. None­
(b)
IEEE PotEntIals
theless, such technologies suffer from
high installation time and cost and
capture only part of the spatiotempo­
ral traffic evolution, whereas different
technologies measure only a fraction
of the traffic parameters of interest
(e.g., traffic flow, speed, vehicle density,
turning ratios at intersections).
Drones are emerging as a prom­
ising technology that can overcome
t hese limitat ions because drone
technology is noninvasive, requires
no installation cost, and provides
frequent spatiotemporal measure­
ments for the entire network. None­
theless, using drones for traffic mon­
itoring leads to several challenges
that need to be dealt with regarding
their deployment, such as maintain­
ing the line of sight and developing
computer­vision algorithms for the
detection of traffic entities (e.g., pe­
destrians, vehicles, buses).
In this context, recent research
efforts have been geared toward cre­
ating software tools to address these
challenges by formulating and solv­
ing appropriate optimization prob­
lems for the efficient deployment of
drones to ensure line of sight for all
roads of the network and by develop­
ing accurate and real­time algorithms
for the detection of vehicles using
data from onboard cameras. Vehicle
detection is achieved by analyzing
the v isual data through a n inte­
grated computer vision processing
pipeline that dynamically extracts
roads segments and robustly detects
vehicles in the resulting network by
combining the results from two dif­
ferent approaches (Fig. 3). The first
uses motion algorithms to find mov­
ing vehicles, whereas the second
(c)
employs convolutional neural net­
works to detect still vehicles using
deep­learning techniques. Finally,
postprocessing the results helps in
the derivation of individual vehicle
trajectories and computation of traf­
fic parameters of interest.
Power-line inspection
Providing an uninterrupted supply
of electricity is an essential everyday
commodity. Nonetheless, properly
inspecting and maintaining the vast,
aging, and often inaccessible power­
line network of a country to detect or
prevent the occurrence of faults is a
difficult and costly task. Drones pro­
vide an attractive alternative to man­
ual inspection that can increase pro­
ductivity and reduce the inspection
time and cost. They help by patrol­
ling the lines using visual equipment
to detect issues that need to be fixed.
This is a very tricky task because
power lines are too thin and not eas­
ily distinguishable in poor lighting
and background conditions, obsta­
cles may block the drone path, and
faulty components may be diffi­
(a)
FIG4 a drone-based power line inspection. (a) the detection of an overheated insulator, (b) a destroyed insulator, and (c) power lines
as viewed by a thermal camera.
(a)
FIG5 a drone as a future first responder can (a) assist civil-protection units in organization and planning, (b) provide situational aware-
ness of disaster-stricken areas, and (c) locate disaster survivors with thermal cameras.
cult to detect even with manual ex­
pert inspection.
For this problem, part of our recent
work focuses on developing methods
and software tools for drone­based
power­line monitoring with minimum
intervention from the operator to de­
tect and report cut power lines or
poles, broken insulators, vegetation,
or other items touching or near the
lines (Fig. 4). Drones are programmed
to autonomously take off, follow the
shortest path to go over each pole, and
then return and land in a safe and
timely manner. High­definition, infra­
red, and thermal cameras are used to
take closeup images and video, which
are processed in real time to detect
and geotag faults using computer­
vision algorithms. Multiple sensing
technologies and computer­vision al­
gorithms allow for the detection of dis­
tinct faults. The use of thermal cam­
eras, e.g., helps with the identification
of overheated insulators, and different
computer­vision algorithms help with
the detection of power lines and veg­
etation in different backgrounds (e.g.,
pavement or rough terrain).
(b)
(b)
IEEE PotEntIals
Civil-defense protection
Drones are considered to be future
autonomous first responders for
civil­defense protection. At their cur­
rent stage, drones support firefight­
ing operations, enhance the capa­
bilities of civil­protection forces, and
improve the response of rescue teams.
In this context, drones support sever­
al missions related to the early detec­
tion of abnormal situations, assist
disaster preparedness for various
physical or manmade disasters (i.e.,
forest fires, floods, earthquakes, or
landslides), and assist search­and­
rescue operations.
Our recent work is geared toward
developing software and hardware
technologies to support a multitude
of civil­defense applications (Fig. 5).
Using high­definition and thermal
cameras, drones locate disaster sur­
vivors, such as trapped civ i lia ns
after an earthquake or shipwreck
survivors in the sea. Search­and­
rescue operations also benefit from
autonomous deliveries using cus­
tomized robotic arms to pick up and
drop emergency response packa ges.
(c)
(c)
Januar y/Februar y 2019 ■ 35
 Almost every aspect of our lives will be
impacted by the presence and usage
of drones in one way or another.
Another technology we are currently
working on is the establishment of
a temporary infrastructure to sup­
port various functionalities, such as
communication connectivity, us­
ing a network of Internet­enabled
drones or area illum i nat ion us­
ing drones equipped with special
lighting equipment. Finally, con­
structing three­dimensional maps
of a ter rain can help emergency
managers assess the impact of a
disaster on different regions and pri­
oritize operations.
Conclusion
Drones and their underlying en­
abling technologies are advancing at
a rapid pace. In the not­too­distant
future, not only will your grocery
shopping be delivered by a drone, you
will commute via a flying autonomous
vehicle. The development of hundreds
of more uses of drones are under
way due to the multiple investments
pouring into this promising indus­
try. Almost every aspect of our lives
will be impacted by the presence
a nd usage of drones in one way
or another.
It is a ver y exciting time to be
working in this area of research,
with many real­world problems and
interesting applications that can
be solved using innovative drone
technolog y. Specifically, there is
great promise in developing novel
swarm optimization t e c h n ique s
and autonomous­control algorithms
so that a group of drones can col­
laboratively execute a task much
more efficiently.
Also, there will be many oppor­
tunities to utilize machine learning
for improving the drone perception
capabilities by taking advantage of
large amounts of data that are be­
coming available for numerous ap­
plications. F i n a l l y, as drones are
integrated more and more into our
society, security and safety aspects
will take a central role in forming
36 ■ J a n u a r y / F e b r u ar y 2019
how drones can be used in many
consumer applications.
acknowledgment
This work was funded by the Euro­
pean Union’s Horizon 2020 research
and innovation program under
grant agreement 739551 (KIOS CoE)
and from the Republic of Cyprus
through the Directorate General for
European Programs, Coordination,
and Development.
Read more about it
• D. Joshi. (2017, Aug. 8). Com­
mercial unmanned aerial vehicle
(UAV) market analysis—Indust r y
t rends, compa n ies a nd what you
should know. Business Insider. [On­
line]. Available: http://www.business
insider.com/commercial­uav­market­
analysis­2017­8
• P. Petrides, P. Kolios, C. Kyrkou,
T. Theocharides, and C. Panayiotou,
“Needs assessment of un manned
aerial vehicles in emergency re­
sponse,” in Proc. Smart Blue City:
First Euro-Mediterranean Conf. and
Exhibition, 2016, pp. 1–16.
• P. Petrides, P. Kolios, C. Kyrkou,
T. Theocharides, and C. Panayiotou,
“Disaster prevention and emergency
response using unmanned aerial
systems,” in Smart Cities in the Medi-
terranean. New York: Springer­Ver­
lag, 2017, pp. 379–403.
• C. Kanellakis and G. Nikolako­
poulos, “Survey on computer vision
for UAVs: Current developments and
trends,” J. Intelligent Robot. Syst., vol.
87, no. 1, pp. 141–168, 2017.
• C. Kyrkou, S. Timotheou, P. Ko­
lios, T. Theocharides, and C. G. Pan­
ayiotou, “Optimized vision­directed
deployment of UAVs for rapid traffic
monitoring,” in Proc. 2018 IEEE Int.
Conf. Consumer Electronics (ICCE),
Las Vegas, NV, pp. 1–6.
about the authors
Christos Kyrkou (kyrkou.christos@
ucy.ac.cy) earned his B.Sc., M.Sc.,
IEEE PotEntIals
and Ph.D. degrees in computer engi­
neering from the University of Cyprus
in 2008, 2010, and 2014, respective­
ly. He is a senior researcher at the
KIOS Research and Innovation Cen­
ter of Excellence at the University of
Cyprus, Nicosia.
Stelios Timotheou (timotheou
.stelios@ucy.ac.cy) earned his B.Sc.
degree from the Electrical and Com­
puter Engineering School of the
National Technical University of Ath­
ens, Greece, in 2005 and his M.Sc.
and Ph.D. degrees from the Electri­
cal and Electronic Engineering De­
partment of Imperial College London
in 2006 and 2010, respectively. He
is an assistant professor in the
Department of Electrical and Com­
puter Engineering and the KIOS
Research and Innovation Center of
Excellence at the University of Cy­
prus, Nicosia.
Panayiotis Kolios (pkolios@ucy
.ac.cy) earned his B.Eng. degree in
telecommunications engineering
from King’s College, London, in 2008.
He was awarded a Ph.D. degree from
King’s College in 2012. He is a mem­
ber of the KIOS Research and Inno­
vation Center of Excellence at the
University of Cyprus, Nicosia.
Theocharis Theocharides (ttheo
charides@ucy.ac.cy) earned his B.Sc.
and Ph.D. degrees in computer engi­
neering from Pennsylvania State
University, State College, in 2002 and
2005, respectively. He is an associate
professor in the Department of Elec­
trical and Computer Engineering
and the director of research at the
KIOS Research and Innovation Cen­
ter of Excellence at the University of
Cyprus, Nicosia.
Christos Panayiotou (christosp@
ucy.ac.cy) earned his B.Sc. and
Ph.D. degrees in electrical and com­
puter engineering from the Univer­
sity of Massachusetts, Amherst, in
1994 and 1999, respectively. He is
an associate professor with the
Department of Electrical and Com­
puter Engineering and deputy direc­
tor of the KIOS Research and Inno­
vation Center of Excellence at the
University of Cyprus, Nicosia.
 Smart Consumer Electronics Systems
©iStockphoto.com/lVcandy

A method for localizing

the eye pupil for
point-of-gaze estimation
narayan panigrahi, kiran lavu, Sai krishna Gorijala, peter corcoran, and Saraju p. mohanty

©iStockphoto.com/Vladimir arndt

T
he estimation of the point
of gaze in a scene present­
ed on a digital screen has
many applications, such as
fatigue detection and atten­
tion tracking. Some popular appli­
cations of eye tracking through gaze
Digital Object Identifier 10.1109/MPOT.2018.2850540
Date of publication: 28 December 2018
estimation are depicted in Fig. 1.
When estimating the point of gaze,
indentifying the visual focus of a
person within a scene is required.
This is known as the eye fix or point
of fixation. Finding the point of gaze
involves tracking different features
of human eyes. Various methods
are available for eye tracking, some
of which use special contact lenses,
whereas others focus on electrical
potential measurements. Optical
tracking is a nonintrusive techni­
que that uses a sequence of image
frames of eyes that have been recorded
using video­capturing devices. This
technique is popularly known as
video oculography.
All of these techniques only work
successfully if two or more features

0278-6648/19©2019IEEE IEEE PotEntIals Januar y/Februar y 2019 ■ 37


Fatigue
Detection
Pupillometry
Training Effectiveness
FIG1 the applications of gaze-point estimation.
of the eyes are accurately detect­
ed. This increases the computation­
al complexity, thereby limiting the
speeds of eye­tracking systems. Some
commercial eye­tracking equipment
available on the market includes the
Eyelink­1000, SensoMotoric Instru­
ments, and Tobii glasses. The cost
of these systems is quite high, and,
therefore, they can only be used for
research purposes. In this article,
a low­cost system is proposed that
uses a modified Web camera to ob­
tain fixation points using the center
of the pupil. The system is capable of
generating 25–30 frames/s which is
sufficient to compute the position and
size of a pupil in the pixels.
Pupil
Iris
FIG2 an eye captured using passive
imaging.
(a)
FIG3 (a) the bright-pupil effect and (b) the dark-pupil effect.
38 ■ J a n u a r y / F e b r u ar y 2019
Attention
Tracking
Computer-Based
Gaming Simulation
Gaze-Point
Estimation
Applications Psychological
Studies
Measurement of
state-of-the-art gaze estimation
Different eye­tracking algorithms,
applications, and methods have been
reported. Many interesting applica­
tions of real­time eye­gaze tracking
for gaming designs and consumer
electronics systems are available.
Eye tracking can be done using such
methods as two­dimensional (2D)
regression and three­dimensional
(3D) modeling of the eye. The 3D
methods implement a geomet r ic
model of the eye to estimate the point
of visual axis and scene intersection.
These systems typically use a single
camera, although models utilizing
two or more cameras for fixation are
also available. Artificial neural net­
works used for determining the
mapping function that converts the
pupil­glint vector into gaze points are
also reported.
A system that detects a human face
in an image and then detects the eyes
and estimates the center of the pupil
based on the image has been pro­
posed. We present a system that uses
only a single feature, the center of the
pupil, to estimate the point of gaze.
(b)
IEEE PotEntIals
a specific eye-tracking method
The first step in obtaining the point of
gaze is to acquire the coordinates of
the center of the pupil based on snap­
shots of the eye of a subject. There are
two popular ways of performing this:
passive and active imaging.
Passive imaging uses the sur­
rounding visible light to capture im­
ages of t he eyes, as depicted in
Fig. 2. These images are processed to
detect various eye features, such as
eye corners and the pupil.
Active imaging uses an illumina­
tion source along with the imaging
device. If the illumination source
emits light in the visible spectrum,
it causes discomfort to the user and
limits the duration for which the ex­
periment can be conducted. If the
source is coaxial to the optical axis
of the eye, infrared (IR) light gets re­
flected from the retina and illumi­
nates the pupil, leading to the bright­
pupil effect depicted in Fig. 3(a). If the
illumination source is offset from
the optical path, the retro reflec­
tion from the retina is directed away
from the camera, causing the dark­
pupil effect depicted in Fig. 3(b).
The bright­pupil effect is difficult to
obtain due to the requirement that
the illumination source needs to be
coaxial to the optical path. This re­
quires a specialized IR illumination
source. However, the dark­pupil ef­
fect can be produced easily, and, as
a result, it is a widely used technique
for eye tracking.
Detection of pupil
The main feature required for any
eye tracker is to detect the coordi­
nates of the center of the pupil. How­
ever, to identify the relative motion of
the pupil, at least one other feature
that stays fixed during eye move­
ments is required. This feature can
be an eye corner or a glint formed by
the IR illumination source. The vec­
tor of the pupil­glint or pupil­eye
corner can be used to obtain a point
of regard on a screen, allowing the
user’s head some slightly free range
of movement. However, if the head
pose of the viewer is stabilized, coor­
dinates of the center of the pupil
alone are sufficient to obtain the eye
fix. This reduces the computational
complexity required to identify other
features of the eye, such as glints
and eye corners.
Detecting the pupil location is a
key feature of any eye­tracking mech­
anism. The accuracy of the system
largely depends on the precise detec­
tion of the center of the pupil for all
positions of the eyes. Detection of the
pupil is easier when using the bright­
or dark­pupil effect [Fig. 3(b)]. For
simplicity, this discussion is limited
to the dark­pupil effect.
The first step is to acquire an im­
age of the eye using an IR camera.
The dark regions in this frame con­
sist of eyelashes, a part of the eye­
brow, and the pupil. Of all these dark
regions, the pupil has a well­defined
shape and is the largest connected
component in the entire image. The
second step is to obtain a binary im­
Eye tracking can be done using such methods
as two-dimensional regression and
three-dimensional modeling of the eye.
age that consists of dark regions other dark spots from lashes and
only. This is achieved by applying the eyebrows.
an intensity threshold transforma­ The next step is to find the largest
tion to the image. A threshold value connected component, i.e., the pupil
is chosen that represents the value in this image. One way is to process
of intensity of color in each pixel that the entire image to detect the largest
is to be considered as dark. If all of connected component. This method is
the three values of any pixel, i.e., suitable when the head position is not
red, green, and blue (RGB) are less fixed. However, if the head position is
than that threshold value, then that stable, a small region can be chosen
pixel is considered as dark. Thus, all within which the pupil can be locat­
of the pixels with RGB values above ed, and only that region will be pro­
the threshold are painted in white, cessed in every frame. This reduces a
and all of the pixels with RGB values significant amount of computational
less than or equal to the threshold effort and the time required to detect
are painted in black. Now, the binary the pupil. To find the largest con­
image contains the pupil and a few nected component, a unified scan and

Start

Adjust Camera

Obtain Snapshot of Eye, Apply Intensity Threshold Transformation, and Scan ROI

Find Largest Connected Component (Pupil) in ROI and Centroid of Pupil

Yes
Is System Calibrated?
No
Display Calibration Point

Yes No
Fixation?

Yes
Store Data More Points?

No
Obtain Mapping Function and Optimization Function

Display Random Point

Obtain Point of Fixation and Calculate Point of Gaze, Optimize to Reduce Error

Yes
Process More Points?

No
End

FIG4 the proposed method for pupil detection. roi: region of interest.

IEEE PotEntIals Januar y/Februar y 2019 ■ 39

 To compute the actual coordinates for the
corresponding pupil coordinates, one must
calculate the value of the polynomial coefficients
using regression techniques.
array­based search algorithm can be goal of detecting the pupil coordi­
used. Once the pupil is detected, the nate is to calculate a point of gaze
centroid of the pupil will give the coor­ from it. For this purpose, a personal
dinates of its center. The bright­pupil calibration routine is run for every
effect follows the same principle ex­ session. A calibration routine typi­
cept for the fact that the pupil appears cally displays a few points or dots
bright, and, hence, the threshold ef­ in a particular order or a random
fect is to be applied in such a way that manner. The users are instructed to
only bright regions remain. The rest of fixate on those points, and, while
the procedure is the same as that of the doing so, the corresponding pupil
dark­pupil effect. The proposed method coordinates for each point is record­
is depicted in the flowchart in Fig. 4. ed. These data are used to obtain a
modeling function that gives the
Obtaining the point of gaze relation between pupil coordinates
from pupil coordinates and actual screen coordinates. The
The pupil coordinates will provide screen coordinates can be related to
the location information of the pupil the pupil center coordinates using
in any captured frame. However, the two separate functions. The func­
Support for
Stabilizing
Head Movement
50 cm
Camera
(a) (b)
(c) (d)
FIG5 the experimental setup for gaze estimation. (a) the actual experimental setup.
(b) the head fixated with camera. (c) the head fixated with the camera setup for eye-
gaze estimation. (d) the graphical user interface of the eye-gaze estimator with a map
on the screen.
40 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals
tions are second­order polynomials
of the pupil center coordinates.
To compute the actual coordinates
for the corresponding pupil coordi­
nates, one must calculate the value
of the polynomial coefficients using
regression techniques. The two most
popular are polynomial affine trans­
formation using matrix inversion
and the least­squares technique.
In polynomial affine transforma­
tion, four equations are obtained by
substituting the recorded data by
displaying four different points on
the screen and the corresponding
pupil coordinates. These equations
are solved using a matrix inversion
method to obtain unknowns. The
other method is to display any num­
ber of screen coordinates and pupil
coordinates and obtain the values of
coefficients using the least­squares
method. The actual exper imenta l
setup is shown in Fig. 5(a) and (b).
a case study
The experimental setup consists of a
screen displaying the scene in which
the fixation graph is to be identi­
fied. A customized Web camera was
mounted on a chin rest to obtain
snapshots of the eyes. The camera
was modified to detect IR light by
removing the IR blocking filter from
its lens assembly. The chin rest acts
as a support to stabilize the subject’s
head movement. The screen is placed
at a distance of 50 cm from the sub­
ject’s eye, as in Fig. 5. The overall dis­
play resolution of the scene is 800 ×
600 pixels. The scene is 22 cm high
and 35 cm wide. This provides a visu­
al cone with a 25° vertical and 39°
horizontal field of view, and the reso­
lution of the visual angle is 26 pixels
per degree vertically and 21 pixels per
degree horizontally. IR light­emitting
diodes (LEDs) were used to illuminate
the eyes of the subject.
The first step in calibration was to
match the coordinate system of the
eye with the coordinate system of the
screen. To achieve this, a crosshair is
displayed on the screen. The subject is
instructed to fixate on the center of the
crosshair, and the camera is adjusted
such that the center of the pupil is
exactly at the center of the crosshair
 0 0
2 3 1 2 2 3
1
12 100 1 10 10 11 12
100 10 11 11 3
12
200 8 9
200 8 1 10 2 11123 13 9 13 4

y Axis
17 4
y Axis

8 17 913 4 17
17
15 5 13 4
300 7 16 6 14 300 8
16 16 15 15 14
16 15 14
14
400 400
7 6 5 7 66 5
500 500 7
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
x Axis x Axis
(a) (b)

FIG6 (a) the first and (b) second phases in calibration.

displayed on the screen. Then, cali­

bration is done by repeating the fixa­ 35 cm
Eye
tion process using a set of 17 random
22 cm

points on the screen in two phases.

During the first phase, these points
are displayed on the screen as red­col­ 25°
ored dots. The subject is instructed to (a)
fixate on these points, and whenever
a fixation is detected, the correspond­
Eye
ing pupil position is recorded. The
red­colored dots in Fig. 6 represent
actual screen coordinates, and green
dots represent their corresponding
pupil coordinates. 39°
These coordinate pairs as recorded
in the first phase are used to establish
the polynomial affine transformation
22 cm

(PAT) by computing the coefficients

of the polynomial. These polynomial
coefficients are used to estimate a
35 cm
point of gaze on the screen. The sec­
(b)
ond phase is added to improve estima­
tion accuracy. During this phase, the FIG7 (a) a vertical visual cone and (b) horizontal visual cone.
estimated points of gaze and actual
screen coordinates computed using
the PAT are used in polynomial re­ TABLE 1. The error analysis of the pupil localization.
gression to obtain a function to opti­ PARAMETERS MATRIX INVERSION LEAST SQUARES
mize the estimated point of gaze.
horizontal error 2.8° (59 pixels) 1.81° (38 pixels)
The screen was placed 50 cm
away from the subject’s eye. The Vertical error 3.2° (77 pixels) 2.4° (58 pixels)
overall resolution for displaying the
mean error 3.0° (68 pixels) 2.10° (48 pixels)
scene was 800 × 600 pixels. The
scene was 22 cm high and 35 cm
wide. The angles subtended by the 24 pixels per degree vertically and sion was calculated. The experiment
eye vertically and horizontally with 21 pixels per degree horizontally. was performed using the matrix in­
the scene were measured (Fig. 7). The To test the system, five random version technique and least­squares
vertical angle was 25°, and the hori­ points were displayed, and their cor­ method to deduce the mapping func­
zontal angle was 39°. This provides responding points of gaze were calcu­ tion. The average errors in both cases
a visual cone with a resolution of lated. The average error for each ses­ are presented in Table 1. The overall

IEEE PotEntIals Januar y/Februar y 2019 ■ 41

 TABLE 2. A comparative perspective of various eye-tracking algorithms.
METHODS USED
3d geometrical model (meyer et al.)
3d geometrical model (hennessey,
noureddin, and lawrence)
3d geometrical model
(lai, Shih, and hung)
3d geometrical model
(ohno and mukawa)
2d regression of the current article
average error was found to be 48 pix­
els. This corresponds with 2.105° of
the visual angle. The performance of
the proposed method is compared
with state­of­the­art methods pre­
sented in Table 2.
Conclusion
The presented system uses only one
feature of the eye, i.e., the center of
the subject’s pupil, to obtain the point
of gaze on a screen. This reduces the
computational effort required for de­
tecting other features, such as eye
corners, corneal reflections, and glints,
enabling high­speed gaze tracking.
Also, noise in the detection of these
additional features is eliminated,
thereby increasing the confidence and
reliability of the data acquired from
our system. As the proposed system
uses a simple Web camera, the cost
of the system is lower than many
other eye trackers and, at the same
time, the system still achieves accu­
racy levels similar to those of expen­
sive methods.
There are multiple future direc­
tions of this research, including the
integration of the proposed system
in a smart health­care framework.
The point of gaze on a screen can be
fused with time intervals when the
subject’s electroencephalogram beta
wave is high to ascertain the point of
fix on the screen.
Read more about it
• D. A. Robinson, “A method of
measuring eye movement using
a scleral search coil in a magnetic
42 ■ J a n u a r y / F e b r u ar y 2019
HARDWARE SETUP
two ir lEds and one camera
multiple ir illumination sources, one pupil contour using bright-pupil effect
camera and glints using dark-pupil effect
two cameras and four ir lEds
two cameras for face detection and one 3d facial orientation, glints,
camera on pan-and-tilt mechanism
modified Web camera with built-in ir
lEds and ir blocking filter removed
field,” IEEE Trans. Biomed. Electron.,
vol. 10, no. 4, pp. 137–145, 1963.
• A. Bulling, “Eye movement anal­
ysis for activity recognition using elec­
trooculography,” IEEE Trans. Pattern
Anal. Mach. Intell., vol. 33, no. 4, pp.
741–753, 2011.
• A. T. Duchowski, “A breadth­first
survey of eye­tracking applications,”
Behav. Res. Methods Instruments
Comput., vol. 34, no. 4, pp. 455–470,
2002.
• J. G. Wang and E. Sung, “Study
on eye gaze estimation,” IEEE Trans.
Syst., Man, Cybern. B, vol. 32, no. 3,
pp. 332–350, 2002.
• H. C. Lee, D. T. Luong, C. W.
Cho, E. C. Lee, and K. R. Park, “Gaze
tracking system at a distance for con­
trolling IPTV,” IEEE Trans. Consum.
Electron., vol. 56, no. 4, pp. 2577–
2583, 2010.
• A. Meyer, M. Böhme, T. Marti­
netz, and E. Barth, “A single­cam­
era remote eye tracker,” in Proc. Int.
Tutorial and Research Workshop Per-
ception and Interactive Technologies
for Speech-Based Systems, 2006,
pp. 208–211.
• C. Hennessey, B. Noureddin,
and P. Lawrence, “A single camera
eye­gaze tracking system with free
head motion,” Measurement, vol. 1,
pp. 27–29, Mar. 2006.
• C. C. Lai, S. W. Shih, and Y. P.
Hung, “Hybrid method for 3D gaze
tracking using glint and contour fea­
tures,” IEEE Trans. Circuits Syst. Video
Technol, vol. 25, no. 1, pp. 24–37, 2015.
• T. Ohno and N. Mukawa, “A
free­head, simple calibration, gaze
IEEE PotEntIals
FEATURES REQUIRED ACCURACY
Glints and pupil center roughly 1°
0.90°
pupil contour and glints 1.18–1.43°
roughly 1°
and pupil contour
pupil center only 2.10°
tracking system that enables gaze­
based interaction,” in Proc. 2004
Symp. Eye Tracking Research and
Applications, pp. 115–122.
• A. Kar and P. Corcoran, “A re­
view and analysis of eye­gaze es­
timation systems, algorithms and
performance evaluation methods
in consumer platforms,” IEEE Ac-
cess, vol. 5, pp. 16,495–16,519, Aug.
2017.
about the authors
Narayan Panigrahi (pani@cair
.drdo.in) earned his Ph.D. degree
from the Indian Institute of Technol­
ogy, Bombay, and is a scientist at
the Center for Artificial Intelligence
and Robotics, Bengaluru, India.
Kiran Lavu (kiranlavu9@gmail
.com) earned his B.E. degree in elec­
tronics and communication engi­
neering from Vignan’s University,
Vadlamudi, India, in 2017.
Sai Krishna Gorijala (saikrishna
gorijala@gmail.com) earned his B.E.
degree in electronics and communi­
cation engineering from Vignan’s
University, Vadlamudi, India, in 2017.
Peter Corcoran (dr.peter.corco
ran@ieee.org) is the founding editor­
in­chief of IEEE Consumer Electronics
Magazine and holds a personal chair
in electronic engineering at the Col­
lege of Engineering and Informatics
at the National University of Ireland
Galway. He is a Fellow of the IEEE.
Saraju P. Mohanty (s a r aju
.mohanty@unt.edu) is a professor at
the University of North Texas, Denton.
 Smart Consumer Electronics Systems
©iStockphoto.com/lvcandy

E-marketing via
augmented reality: A case
study in the tourism and
hospitality industry
neda Shabani, arslan munir, and azizul hassan

M
a rket i ng pl ay s a n
extremely significant
role in the success of
any business. In the
hospitality and tour­
ism industr y, marketing is a key
tool to introduce various desti na­
tions to customers and attract them
to visit a place or stay in a particu­
lar hotel. The forms of marketing
have considerably evolved during
the past decades owing to enhance­
ments in technology.
E­marketing is a recent product
of this evolution that has revolu­
tionized the world of marketing. It

©iStockphoto/Bet_noire
has created a huge impact on dif­
ferent businesses and sectors, in­
cluding the hospitality and tourism
industry, in many ways, such as
customer satisfaction and revenue
growth. In recent years, augment­
ed reality (AR) has been used as a
tool to boost e­marketing in many
ways due to its unique features AR in e­marketing for these indus­ from conventional to nonconven­
and capabilities. tries with a few real­life examples tional and then to online composi­
In this article, we discuss how and use cases. tion. All of these transformations
AR has revolutionized e­marketing are aimed at improving effectiveness
in the context of the hospitality What is e-marketing? and efficiency.
and tourism industry. We further Marketing has been in existence Among the different types of mar­
review the successful integration of for centuries, since humans began keting strategies, e­marketing is the
trading to address human­centered one that has recently received the
demands. However, the form of mar­ most attention for a multitude of rea­
Digital Object Identifier 10.1109/MPOT.2018.2850598
keting has transformed over time. sons. However, before we mention the
Date of publication: 28 December 2018 In recent decades, it has revamped reasons for e­marketing’s eminence,

0278-6648/19©2019IEEE IEEE PotEntIals Januar y/Februar y 2019 ■ 43

 E-marketing is entirely technology focused
and has replaced centuries-old marketing
principles and tactics.
let us first define the term. E-market-
ing, in its simplest manner, can be
defined as enacting the techniques
and principles of marketing through
the use of electronic media and, more
often, the Internet. It is the outcome
of modern and developed communi­
cation systems, where the Internet is
the key platform. E­marketing is en­
tirely technology focused and has re­
placed centuries­old marketing prin­
ciples and tactics. Furthermore, it
embraces newer routes for marketing
communications that do not neces­
sarily exist in conventional market­
ing systems.
E­marketing can play a vital role
in the success of a business, prod­
uct, or service. One of the main re­
quirements for running a successful
business is to deeply understand the
customers’ needs, demands, prefer­
ences, responses, and attitudes to­
ward a particular product or service.
In marketing, the measurement of
postpurchase consumer responses
and behavior is extremely significant
for businesses to enhance the qual­
ity of their products and/or servic­
es. These behaviors and responses
can appear in many forms, such as
feedback, replies, reactions, or com­
ments. Because the interaction be­
tween product or service providers
and consumers are fundamental in
creating and maintaining functional
relationships, e­marketing provides
businesses, products, and services
the opportunity to communicate and
interact with both existing and po­
tential customers.
By adopting e­marketing, busi­
nesses can ensure efficiency, better
return on investment, and enhanced
profitability that leads to increased
cost effectiveness. Furthermore,
cam paigns in e­marketing boost
online sales by attracting potential
consumer bases and by minimizing
negative risks.
In the context of tourism and hos­
pitality, the interaction and communi­
44 ■ J a n u a r y / F e b r u ar y 2019
cation with consumers must be more
focused due to the specific nature of
the industry. We would like to em­
phasize that the tourism and hospi­
tality industry is one that is service
based. To market any service in­
dustry, the first thing to consider is
the uniqueness, along with the per­
sonalization and customization, of
all services. Some of the distinctive
characteristics of the tourism and
hospitality industry services that
distinguish its marketing methods
from other industries are
■■perishability
■■inseparability
■■intangibility
■■heterogeneity
■■seasonality
■■agreeability
■■inconsistency
■■imitability
■■inflexibility.
Besides of all of these unique char­
acteristics, the consumer­oriented
nature of the tourism and hospitality
industry plays a significant role in
choosing the most appropriate and
effective marketing strategy. In many
cases, e­marketing has become an
essential ingredient to an existing or
potential business enterprise related
to the tourism and hospitality indus­
try. Selecting an effective technology
for e­marketing and making proper
use of it are critical to helping those in
the tourism and hospitality industry
to achieve their objectives and goals.
aR for e-marketing
Boeing researcher Thomas Caudell
coined the term AR in 1990, when he
invented the technology to help air­
craft workers in assembling long
bundles of wires for the new 777 jet­
liner as an alternative to costly dia­
grams and marking devices. AR has
gained more traction in recent years
due to further advances in sensors,
optics, computer vision, image pro­
cessing, audio processing, and com­
puting. From the computer­science
IEEE PotEntIals
perspective, AR is a direct or indi­
rect real­time (live) view of a physi­
cal/real environment whose elements
through computer­generated percep­
tual information are augmented or
superimposed across a potential
multitude of sensor modalities, in­
cluding but not limited to visual,
auditory, and haptic. Vivian Rosen­
thal, the founder of GoldRun, a New
York City­based AR start­up, has
defined AR as “a digital layer over
the real world that you can’t see
with the naked eye, but you can see
with the camera on your smartphone
or computer.”
AR technology is a shift from the
earlier technology version of virtual
reality (VR). Both VR and AR are
anticipated to be disruptive tech­
nologies with applications not only
in entertainment but also business
and education. AR differs from VR
in that VR is an artificial, computer­
generated simulation that creates
a virtual environment very similar
to real life and makes it difficult for
users to recognize that the environ­
ment/experience is not real, whereas
AR is a mixture of real life and VR.
VR enables users to interact with it
by wearing a VR helmet or goggles.
AR developers create images and
computer­generated enhancements
in applications that enable the us­
ers to combine those with real life,
thus permitting interaction with
VR in real life. In AR, the augmen­
tation techniques are often per­
formed in real time when integrat­
ing the digital information with the
user’s environment.
Characteristically, AR technology
is a reality simulation by computer­
generated visuals where a user can
experience a reality in a virtually
created environment. AR brings two
widely diverse dynamics: personal
exclusivity perception and a sensory
experience that is commonly mul­
tidimensional. AR has numerous
applications in consumer electron­
ics, entertainment, marketing, and
business. Even without realizing it,
almost all consumers are more or
less familiar with the experience that
AR can generate. The revenue gen­
erated from AR was US$2.39 billion
in 2016 and is projected to reach Both VR and AR are anticipated to be disruptive
US$120 billion in 2020.
The use of AR technology in mar­ technologies with applications not only in
keting is seemingly less expensive entertainment but also business and education.
than conventional marketing ap­
proaches. Print­based advertising and
marketing campaigns are relatively sions regarding these destinations. marketing program with an innova­
more expensive and require con­ In particular, tourism and hospitality tive AR technology application.
tinuous investment of resources. AR marketers can leverage AR to miti­ In the tourism and hospitality
as a computer­based technology gate the tourists’ concerns by various industry, which solely relies on of­
is less expensive because it can be means, such as providing real­time fering services, word of mouth has a
used in digital platforms or online. travel assistance and augmented special place. Word of mouth is di­
Although the development cost to set maps with markers and informa­ rectly related to expressing human
up AR­based marketing campaigns tion regarding shops, grocery stores, feelings and emotions toward dif­
might be higher, the lifetime operat­ tourist attractions, gas stations, hos­ ferent experiences. AR applications
ing cost mostly remains stationary. pitals, emergency help, banks, post can bolster the emotional connec­
This claim is evident from Rosen­ offices, cafés, gyms, cinemas, and so tion of tourists to destinations that,
thal’s statement that “It honestly forth, as illustrated in Fig. 1. in turn, lead to repeated sales and
depends on the scope of the project, In the tourism and hospitality business. The establishment of this
but AR campaigns can be as inex­ industry, AR provides significant connection between service provid­
pensive as US$5,000 and as high as advantages over conventional broad­ ers and consumers is also supported
US$100,000. That’s nothing com­ cast or print advertising in terms of by Rosenthal:
pared to print, and in many ways, application and usability. A basic It’s not just a matter of tak­
it’s worth the risk in my eyes.” feature of AR that warrants its inte­ ing photos and sharing, which
gration in tourism destination mar­ is what we’re building. But
aR in tourism and keting is its simplicity. AR technology what really matters to us is the
hospitality e-marketing is viewed as simplistic but with the ability to take photos with vir­
People who travel to different destina­ ability to connect closely with target tual characters, products and
tions are the main drivers behind the consumer bases. The content and environments, which is where
tourism and hospitality industry, features of a tourism destination can it will resonate most and devel­
and they do so for a variety of rea­ be viewed more congenially by super­ op that deep brand connection.
sons, such as visiting a destination imposing a virtual environment with Tourism destination marketing
for recreational, medical, or business reality. The St. Petersburg/Clearwa­ or advertising based on AR applica­
purposes; a pilgrimage; meeting ter, Florida, Convention and Visitors tions can grab the attention of par­
relatives; and many other reasons. Bureau, e.g., has augmented their ticular customers. One of the goals
Thus, the tourism and
hospitality industry pro­
vides substantial reve­
nue for a country.
However, tourism in
particular countries/
places is often restrict­
ed for many reasons,
i ncludi ng consumer
distr ust of ava ilable
tourism services, insuf­
ficient security mea­
sures, and travel scams
and frauds in tourism
spots. It is v it a l for
t o u r i s m a n d hospi­
tality marketers to in­
troduce and market a
destination in the best
possible way to attract
the largest number of
travelers and alleviate
the tourists’ apprehen­ FIG1 an ar application in tourism.

IEEE PotEntIals Januar y/Februar y 2019 ■ 45

 AR brings two widely diverse dynamics: personal
exclusivity perception and a sensory experience
that is commonly multidimensional.
of AR in the hospitality and tourism
industry is to reach millennials. AR
can be purposefully applied to ap­
peal to and attract younger tourists
that often have a common interest
in innovative technology use. Some
of the AR applications that have en­
hanced the experience of consum­
ers include
■■interactive travel guides
■■social travel gamification
■■city tours
■■real­time navigation and direction
■■aug mented hotel environment
and room booking
■■interactive hotel rooms
■■augmented language translation
■■augmented access to dining and
entertainment centers
■■participative destination manage­
ment
■■augmented transportation
■■augmented services in restau­
rants
FIG2 an ar museum.
46 ■ J a n u a r y / F e b r u ar y 2019
■■reliving his toric places and
events
■■AR browsers incorporated into
destinations
■■museum interactivity.
The growth of social media connects
consumers (including millennials) by
creating a network of like­minded
people that allows further capitaliza­
tion of such innovative technologies
as AR.
The incorporation of AR in market­
ing and advertising campaigns can of­
fer a variety of benefits to tourism des­
tination marketers. The AR campaign
offers relatively long­lasting targets to
consumer bases. AR enables greater
involvement of both marketers and
consumers with brands by increasing
their ability to make positive sale and
purchase decisions, respectively. AR
provides memorable interactive con­
sumption experiences in the before­,
during­, and after­purchase stages.
IEEE PotEntIals
AR’s ability to offer personal and
customized feelings to the target con­
sumers plays a central role in the
success of AR in the hospitality and
tourism industry. This customization
allows wider interconnectivity and in­
teractivity in both the sales and pur­
chasing processes. Hence, AR can cre­
ate a bridge between the offered and
purchased products and services.
Although AR is a revolutionizing
technology for e­marketing with a va­
riety of benefits, its integration with
tourism destination marketing is
a relatively complex process that re­
quires expertise and resources. At an
early stage, the demand for an AR ap­
plication needs to be recognized and
established. Hospitality and tourism
marketers must explain the added
benefits that AR provides, such as
combining real and virtual images in
real time and the ability to interact
with virtual content.
It is often better to consider AR as
an addition to the existing market­
ing and advertising tools to make the
most efficient use of AR technology. In
other words, AR needs to be an add­
ed, but not the only, strategy to serve
specific objectives. Because AR tech­
nology is rather innovative and new,
it is advantageous to capitalize the
maximum benefits in unique ways
to pull it off effectively.
Use cases and examples
There are many use cases of AR in
the tourism and hospitality indus­
try. The following are examples that
successfully incorporate AR in busi­
ness marketing and have resulted
in both increased profitability and
customer satisfaction.
AR museum
The AR museum is a strategy em­
ployed by some museums to super­
impose the virtual world over the
real world to provide visitors with a
vivid view and understanding of the
origins, history, and details of vari­
ous exhibits and artifacts. The AR
museum integrates and/or blends
digital information with live video
or the user’s environment in real
time using special three­dimen­
sional AR software programs. Fig­
ure 2 illustrates the use of AR in
an AR museum.
Inamo restaurant in London
Inamo is identified as a restaurant
t hat offers a di n i ng ex per ience
with a unique combination of a Far
Eastern feeling in a Western locale.
The virtual environment furnished
by the E­Table chef’s camera, puz­
zles, Battleship game, information
browsing on the local areas, and
even booking a taxi are all rela­
tively new in the hospitality indus­
t r y not only in L ondon but also
the world.
At Inamo, customers can order
food and drinks by using AR tech­
nology­enabled menus placed on
custom­designed customer serving
tables. These tables work as electron­
ic display boards as well as electronic
point­of­sale terminals. The tables
incorporate a waterproof touch panel
operated by Bluetooth wireless tech­
nology. Each table is equipped with
a high­quality projector and a com­
puter that steers the interactive AR
applications and also communicates
AR can be purposefully applied to appeal
to and attract younger tourists that often
have a common interest in innovative
with the central server system. The
central server incorporates various
tools, including business logic, which
define the restaurant’s operational
aspects. Furthermore, the central
server drives the printers in the bar
and kitchen.
To get a better understanding of
the customer perspective on the use
of AR in the hospitality and tourism
industry, we interviewed a few cus­
tomers of Inamo restaurant. After
asking a couple of semistructured
interview questions, we heard some
interesting opinions from those cus­
tomers regarding AR.
■■According to the first respondent,
“The hospitality industry market­
ers in London mostly retain a set
level of technological standard
that actually makes their con­
sumers to become capable to use
a new technology.”
■■The second respondent reported:
“Augmented reality draws the line
that tends to exist between the real
and the digital by using an ap ­
propriate technology.”
■■The third respondent viewed AR
as entertaining and commented:
“Gastronomy is mostly about ex­
periences where AR technology is
seemingly also very interesting
with the involvement of entertain­
ing features.”
■■According to the fourth respon­
dent, “In a technology­induced hos­
pitality business market, there are
many websites, blogs, or online
media allowing customers to get
involved, get attached, and ex­
press new ideas. The customers in
Inamo have competitive operational
knowledge of a technology. This
customer group normally reads
online reviews of early customers in
Inamo. Also, they follow videos and
reviews on popular online stream­
ing media such as YouTube and
many others.”
IEEE PotEntIals
technology use.
Conclusion
Innovative technologies have been
consistently making significant im­
provements in tourism and hospitality
business services. AR is a booming
innovative technology that has helped
the e­marketers in tourism and hos­
pitality industry to enhance their
service quality and create value for
their consumers.
The use of AR in the hospitality
and tourism industry has exceeded
consumer expectation and satisfac­
tion by creating a memorable experi­
ence. Because AR is a new technol­
ogy and limited research on it has
been done so far, especially in the
context of the tourism and hospital­
ity industry, future research and in­
novation in AR technology have the
potential to make a huge impact in
this industry and generate tremen­
dous revenue.
acknowledgment
We would like to thank Prasanna
Kansakar for his help with the figures.
about the authors
Neda Shabani (nshabani@ksu.edu)
is a graduate student in the College
of Business Administration, De­
partment of Accounting at Kan­
sas State University, Manhattan.
He earned his M.S. degree in hos­
pitality and tourism management
from the University of South Florida,
Sarasota­Manatee.
Arslan Munir (amunir@ksu.edu)
is an assistant professor in the De­
partment of Computer Science at
Kansas State University, Manhattan.
Azizul Hassan (m.hassan15@
outlook.cardiffmet.ac.uk) is a mem­
ber of the Tourism Consultants Net­
work of the U.K. Tourism Society
and is currently working toward a
Ph.D. degree at Cardiff Metropolitan
University, Wales.
Januar y/Februar y 2019 ■ 47
gamesman problems
Problem #1:
Rock the Boat
A lightweight boat is float-
ing in calm water. A man
walks from one end of the
boat to the other. Does the
boat move while he is walk-
ing and then stop, continue
moving, or not move at all?
Problem #2: Break the Chain Numbers—© CaN stoCk Photo/agsaNdrew,
There is a chain that contains 23 paper aNdroid—© CaN stoCk Photo/kirstyPargeter
clips. If you remove two of the paper
clips, you can make a chain of any length,
from one to 23, out of the five pieces. Which two paper
clips should you remove?
Problem #3: Pocket Change
What is the largest amount of change in U.S. coins you
can have in your pocket and still not be able to make
change for a dollar bill?
Digital Object Identifier 10.1109/MPOT.2018.2872279
Date of publication: 28 December 2018
the way ahead (Continued from page 4)
plans and estimate regarding fall 2019. If you change of-
ficers in the fall, they can submit their academic year
plan in October/November 2019, and it will count for the
February 2020 deadline. I know some northern hemi-
sphere Student Branches that elect officers in December/
January to assure a greater transition overlap and re-
lieve graduating seniors from having IEEE duties in their
busy final semester.
If you are a Student Member but not a student officer,
contact your officers to provide input. Don’t wait for an
engraved invitation. Student Branches can always use
more active members, and by being more active, you will
get more out of your membership in both professional de-
48 ■ J a n u a r y / F e b r u ar y 2019 IEEE PotEntIals
Problem #4: Juan For
the Road
Fill in the blank: Pedro is
the son of Juan. Juan is
the ______ of Pedro’s father.
Problem #5: space
Elevator
Thank you to Vaughan Clark-
son and Ian Clarkson for submit-
ting this problem.
Douglas Adams writes in The Hitch-
hiker’s Guide to the Galaxy, “There is an
art to flying, or rather a knack. The knack
lies in learning how to throw yourself at the ground and
miss.” Actually, all you need is a space elevator. At what
height could you step off such a structure at the equator
and miss the ground?
If you have a problem for the Gamesman,
please submit it along with the solution
to potentials@ieee.org.
solutions are on page 5.
velopment and networking opportunities with the profes-
sional IEEE Members of your Section.
By the way, consider attending some of the meetings
held by your local Section. All IEEE Student Members
are members of the local IEEE Section and are always
invited, even if the event notice doesn’t explicitly target
Student Members. What a great way to start a new year!
about the author
Elisa H. Barney Smith (SACchair@ieee.org) is the IEEE
Member and Geographic Activities—Student Activities
Committee chair. She is a professor of electrical and com-
puter engineering at Boise State University, Idaho.
Introducing IEEE Collabratec™
The premier networking and collaboration site for technology
professionals around the world.

IEEE Collabratec is a new, integrated online community where IEEE members,

Network.
researchers, authors, and technology professionals with similar fields of interest
can network and collaborate, as well as create and manage content. Collaborate.
Featuring a suite of powerful online networking and collaboration tools, Create.
IEEE Collabratec allows you to connect according to geographic location,
technical interests, or career pursuits.

You can also create and share a professional identity that showcases key
accomplishments and participate in groups focused around mutual interests,
actively learning from and contributing to knowledgeable communities.
All in one place!

Learn about IEEE Collabratec at

ieee-collabratec.ieee.org
 ILLUMINATE EDUCATE ENGAGE ENERGIZE
the possibilities of the next generation a wider audience innovation by
technology by using of innovators and in appreciating the celebrating
it to address global engineers value and importance technological
challenges of engineering and excellence
technology

The world’s most daunting challenges The IEEE Foundation is leading a special
require innovations in engineering, and campaign to raise awareness, create
IEEE is committed to finding the solutions. partnerships, and generate financial resources
needed to combat these global challenges.

Our goal is to raise $30 million by 2020.

DONATE NOW
ieeefoundation.org