Upload
michelle-tu
View
227
Download
4
Tags:
Embed Size (px)
DESCRIPTION
Bruin Innovation & Technology (BIT) Magazine is the first student-written, edited and managed engineering publication at UCLA. Founded in 2010 by UCLA Engineering graduate students, Jamal A. Madni and Neil Tilley, BIT is written by engineering graduate students on the impact their research has on new & emerging technologies, and is intended for all members of the Bruin family, prospective students and the greater community. The current issue includes cutting edge research in mobile & wireless health, a new paradigm for the Internet, an interview with Dean Dhir on the state of engineering at UCLA through the lens of the non-technocrat, as well as entrepreneurial opportunities for students from all over campus.
Citation preview
V O L U M E 1 | S P R I N G 2 0 1 1 | B R U I N I N N O VAT I O N & T E C H N O LO G Y
B I T
Neil TilleyMS Computer ScienceCo-Founder & Managing EditorBIT Magazine
FIRST PITCHThe Technocrat. Our world is becoming more dependent on the impact that their
innovations have on new and emerging technologies. Yet a great paradox has emerged: the
more dependent our world has become on the technocrat’s innovations, the more myste-
rious the technocrat has become to the masses. Part of that is the inherent complexity of
their creations. Part of that is also the complexity with which they convey how their crea-
tions were actualized. The Bruin campus is a microcosm of this world. Herein lay the need
for Bruin Innovation & Technology Magazine.
How do Bruin engineers and technologists think, and what is their workflow? What
are their specific research projects that will have a significant impact on new and emerg-
ing technologies? How can a “non-expert” be able to interact, assist, and collaborate with
these technocrats on specific projects, initiatives, and ventures? BIT Magazine serves to
answer all of these questions for each customized member of the Bruin Family. BIT is for
you, whether you are in another field, employed at UCLA, considering applying, involved
as an alumnus, or looking to invest in new, developing ideas.
BIT personifies the “True Bruin” values and, more specifically, epitomizes the mis-
sionary pillars of our great university: Access, Affordability, and Opportunity. BIT serves to
provide access to the technological developments happening on our campus in a simple,
comprehensible, and jargon-less manner. BIT has no subscription cost, is available online,
and is free to any reader. Finally, BIT is a specific mechanism providing opportunity for a
student-centric entrepreneurial culture and cross-disciplinary collaboration.
We look forward to BIT being embedded in Bruin culture for many years to come.
Let the dialogue begin.
Jamal A. MadniMS Electrical Engineering, MS Biomedical EngineeringCo-Founder & Managing EditorBIT Magazine
HTTP://ENGINEER.UCLA.EDU
SPONSORS
Vijay Dhir Dean, Henry Samueli School of Engineering & Applied Science
ENGINEERING GRADUATE STUDENTS ASSOCIATION
Asad MadniPresident, Engineering
Alumni Association
SPECIAL THANKS
Frank ChangChair, Electrical Engineering Department
Jiun-Shyan (JS) ChenChair, Civil & Environmental Engineering Department
Karl HolmesManagement Services Officer,
Civil & Environmental Engineering Department
Richard KorfVice Chair, Computer Science Department
Adrienne LavineChair, Mechanical & Aerospace Engineering Department
Janet LinAdministrative Assistant, Electrical Engineering Department
Harold MonbouquetteChair, Chemical & Biomolecular Engineering Department
Jens PalsbergChair, Computer Science Department
STAFF
Managing EditorsJamal Madni
Neil Tilley
Graphic & Layout DesignerMichelle Tu
Contributing WritersJong Hoon Ahnn
Navid AminiRahul Basava
Jeff ChangFang Gong
Derek KulinskiChenni Qian
Tanuj ThapliyalSarah Warren
Wenyao XuZhenkai Zhu
Photography DirectorPeyman Nazarian
Special AdvisorStacey Meeker
Sincerely,
CONTENTSA Brain-Computer Interface to Help the Elderly Reduce the Risk of Falling
The Dean’s Corner
Designing Applications Using Named Data Networking
Entrepreneurial Opportunities from the Other Side of Campus
A Wireless Home Automation System for Childhood Obesity Prevention
TEC: A Training Ground That Brings the Universe to our University
Innovative Design Method Creates New Varieties of Rotary Engines
The Social Network is Everywhere in Engineering—What to Read? With Whom to Work? Where to Publish?
Well...
Bruin Innovation & Technology (ISSN [TBA]), Volume 1, Spring 2011, is published two times a year at the University of California, Los Angeles, UCLA Henry Samueli School of Engineering and Applied Sci-ence, 7256 Boelter Hall, Box 951600, Los Angeles, California, 90095-1600. Annual subscription $10.00 domestic, $15.00 foreign air mail; single copies $5.00. Send subscriptions to UCLA Engineering, Office of External Affairs, 7256 Boelter Hall, Los Angeles, CA 90095. Third class postage paid at Los Angeles, CA. All rights reserved. No part of BIT may be printed without specific permission. Some parts of this publication are available electronically at http://www.engineer.ucla.edu/visitor-links/current-students/bit-magazine. As a publication of UCLA HSSEAS, BIT carries authorized advertisements about offerings of the school. The school assumes no responsibility for opinions of contributors on other subjects. © 2011, University of California, Los Angeles. Published by UCLA Engineering and the Engineering Alumni Association. Telephone 310-206-0678.
Editorial control is retained by the editorial staff of BIT.
Advertising in BIT does not indicate an endorsement by UCLA HSSEAS.
BY WENYAO XU AND FANG GONG
WITH VIJAY DHIR
BY RAHUL RAO BASAVA, DEREK KULINSKI,
CHENNI QIAN, AND ZHENKAI ZHU
BY JEFF CHANG
BY NAVID AMINI
BY TANUJ THAPLIYAL
BY SARAH WARREN
BY JONG HOON AHNN
04
10
14
22
24
34
36
48
56
4 VOLUME 1 / SPRING 2011
Falls are reported by one third of all people
65 and older and have become a leading cause of
death. Recently, the number of elderly people with a
fall-induced injury has increased dramatically. The
corresponding health costs for fall-related medical
treatments such as fractures, open wounds, and
head traumas has risen from $20 billion in 2006
to $23 billion in 2010. As such, injuries due to falls
also account for large health care costs and have
become one of the most significant concerns in
geriatry. Therefore, a smart and low-cost solution
of fall prevention for the elderly is urgently sought.
The most promising approach for fall preven-
tion is a wearable assistive system which can detect
and forecast the risk of falling, given the fact that
over 4 million individuals in the United States pay
for health care support with such wearable assis-
tive devices in their daily lives. However, the use of
A BRAIN-COMPUTER INTERFACE TO HELP THE ELDERLY REDUCE
THE RISK OF FALLINGWenyao Xu, Computer Science, and Fang Gong, Electrical Engineering
MANY ELDER PEOPLE OVER 65 ARE AT HIGH RISK OF FALLING BECAUSE OF THEIR GENERAL FRAILTY AND MULTIPLE PATHOLOGIES.
these assistive devices also introduces a negative
influence of an additional cognitive burden for peo-
ple who suffer from cognitive disability because of
age or due to other causes. Fortunately, these limi-
tations are being overcome with the advancement
of sensing technology, especially with emerging,
wearable sensors. The novel sensing technology
can dramatically reduce the footprint of traditional
sensors and thereby integrate them into plastic, pa-
per, and even fabric materials. With wireless com-
munication and computing technology, these wear-
able assistive systems can be used by individuals
comfortably and non-invasively.
Moreover, recent studies empirically prove
that many mental and physiological signals such as
brain waves are also related to falls. They have dis-
covered surprisingly that brain waves, or brain sig-
nals, can offer a better solution for forecasting falls
5BRUIN INNOVATION & TECHNOLOGY
in terms of promptness and accuracy. As shown
in Figure 1, there is a significant anomaly in EEG
signals about 3 seconds prior to the fall [1]. There-
fore, it is possible to assess “fall risk” and activate
fall prevention measures through pre-warning pat-
terns from EEG signals.
Three UCLA graduate students, Wenyao Xu,
Fang Gong, and Ju-Yueh Lee, led by Professor Majid
Sarrafzadeh from the Wireless Health Institute and
Professor Lei He from the Electrical Engineering
Department, have invented a low-cost, lightweight,
wearable device to record the brain waves (so-
called “brain computer interface” or BCI) and fur-
ther interpret them for fall-prevention purposes.
OVERVIEW OF BRAIN COMPUTER INTERFACE SYSTEM
Brain waves were first scientifically observed
in 1912 through the use of Electroencephalograms,
or EEGs, as a kind of bioelectric potential related
to the thoughts of a human. In the 1970s, scientists
at UCLA began their research on a brain computer
interface (BCI). This project received a grant from
the National Science Foundation (NSF) and DARPA.
Through this grant program, several BCI systems
have been built to capture brain waves and inter-
pret them to meaningful thoughts.
In more recent decades, scientists have tried
various methods to decode these mysterious brain
signals. This is known to be a complicated and ex-
pensive task. Traditional BCI systems include two
parts: a signal acquisition module and a data min-
ing module. Conventional scalp BCI systems are
cumbersome and uncomfortable. Dozens of elec-
trodes corresponding to different channels need
to be deployed on the scalp with a conductive gel
or paste. Moreover, the scalp area needs to be pre-
pared ahead of time to remove dead skin cells, all of
FIGURE 1
BRAIN SIGNAL, INCLUDING:
ORIGINAL OUTPUT MEASURE-
MENTS IN A FALL OCCURRENCE
(TOP), EVOLUTION OF MEAN VAL-
UES (MIDDLE), TIME-FREQUENCY
PLOT SHOWING THE POWER
CHANGES DURING A TRANSITION-
TO-FALL STAGE (COLORED).
(COURTESY S. SLOBOUNOV [1])
THEREFORE, IT IS POSSIBLE TO ASSESS “FALL RISK” AND ACTIVATE FALL PREVENTION MEASURES THROUGH PRE-WARNING PATTERNS FROM EEG SIGNALS.
6 VOLUME 1 / SPRING 2011
which is obtrusive and inapplicable for daily use in
fall prevention. Furthermore, for the sake of signal
integrity, sampling rates of conventional scalp BCI
systems can be up to 20 kHz in research applica-
tions, and the recorded EEG signal data can be as
large as 1 gigabyte per day. As such, it is highly chal-
lenging to process an EEG data stream for fall pre-
vention in real-time applications, and efficient sig-
nal processing algorithms are critical. Additionally,
conventional BCI systems are obtrusive and not
suitable for daily use, especially considering their
unaffordability for most people.
Fortunately, there has been good progress in
EEG signal acquisition methods. At UCLA, research-
ers have developed a new, wearable BCI system
(also called a “smart headset”) as shown in Figure
2 that addresses the risks associated with falling.
This new system looks like a normal Bluetooth
headset, but it can detect human brain waves and
function as well as a traditional BCI system.
Professor Majid Sarrafzadeh believes that the
new BCI system is easy to use and will not require
a complicated interface that would affect a user’s
normal daily life. Moreover its low cost can make it
accessible to more people, thus yielding many ap-
plications for this device.
The smart headset can be worn on the head
but does not restrict a user’s activity because it
uses wireless communication. After a user puts a
headset on, it acquires the user’s brain waves via
a sensor on the forehead, passing the captured
signals to a data analysis module where the brain
waves can be deciphered for the detection of mean-
ingful signals.
Most importantly, the headset is very light-
weight, weighing in at a mere ten grams. Addition-
ally, it is affordable, costing about $19. By contrast,
a conventional BCI system costs thousands of dol-
lars and weighs hundreds of kilograms. Therefore,
the new BCI device, properly marketed, will be
more accessible and affordable to more people.
SYSTEM FRAMEWORKIn general, the new BCI system is comprised of
two parts, a client end and a server end. The client
end goes with the user for bio-signal sensing, com-
puting, and transmitting. The server end is for min-
ing the received data. Figure 3 shows the stacked-
layer architecture of client and server, where Figure
3(a) is the structure of the client, and Figure 3(b) is
the structure of the server.
Client
The client part consists of four layers: the
sensing layer, actuating layer, middleware layer,
and communication layer. Sensing and actuating
layers are at the bottom level to interact with us-
FIGURE 2
ENGAGED WEARABLE BCI SYSTEM
MOREOVER, ITS LOW COST CAN MAKE IT ACCESSIBLE TO MORE PEOPLE, THUS YIELDING MANY APPLICATIONS FOR THIS DEVICE.
7BRUIN INNOVATION & TECHNOLOGY
ers directly. Some sensors are included in the sens-
ing layer to acquire miscellaneous user bio-signals.
With the sensed data, middleware implemented
on a microcontroller will perform preliminary
processing such as signal filtering, sorting, com-
pression, and lightweight data analysis for feature
extraction. Some of the expected results are urgent
to report, and they are sent to the actuating layer
to notify users of a risk of falling. Otherwise, the
data will be sent through the communication layer
to the server end for further data mining. The most
common solution for the communication layer is
through Zigbee, Bluetooth, or WiFi, the choice of
which depends highly on the specific applications.
Commented Wenyao Xu, a third-year Ph.D.
student in UCLA’s Computer Science department
and who serves as the architect designer, “We inte-
grated BCI into a Bluetooth headset. Leveraging the
system-on-chips technology, the headset acquires
the brain wave via a simple but effective way. It is
complementary to existing assistive technologies
in terms of mental control.”
Server
For the server part, there are two layers. The
first layer is the communication layer for receiving
data from the client device, and the second layer
is the signal processing layer. A data mining algo-
rithm is implemented in the signal processing layer
to analyze the user’s risk of falling. After the calcu-
lation, the server end will send feedback to the cli-
ent side for actuating.
It is important to note that the data mining
procedure on the server is time-consuming, and
the response time is critical for fall prevention. To
mitigate this problem, researchers have invented
an innovative method for data processing and have
developed an efficient signal processing algorithm
based on a recently established theory of Com-
pressed Sensing (CS) [2][3].
“Decoding the brain wave is a difficult job and
very time-consuming. Perceptible delay can dra-
matically degrade the performance of our brain
control system. To handle this problem, we came
up with some novel algorithms to decode the brain
waves for real-time control purposes,” said Fang
Gong, who is a third-year Ph.D. student in UCLA’s
Electrical Engineering department and serves as
COMMUNICATION LAYER
MIDDLEWARE LAYER
ACTUATING LAYER SENSING LAYER
CLIENT
COMMUNICATION LAYER
SIGNAL PROCESSING LAYER
SERVER
FIGURE 3
(A) CLIENT STRUCTURE
(B) SERVER STRUCTURE
WE INTEGRATED BCI INTO
A BLUETOOTH HEADSET.
LEVERAGING THE SYSTEM-
ON-CHIPS TECHNOLOGY,
THE HEADSET ACQUIRES
THE BRAIN WAVE VIA A SIMPLE
BUT EFFECTIVE WAY.
8 VOLUME 1 / SPRING 2011
the software designer for this project. “These more
efficient algorithms maximize computing resourc-
es, a chief reason why the unit can be so tiny.”
PROTOTYPEFigure 4(a) illustrates the acquisition system,
called Smart Headset. There are two main parts:
an ADC module (analog-to-digital converter) and a
data processing module. The ADC used in the sys-
tem is a kind of adaptive sampling module from
Neurosky Inc [4]. This ADC has a low-sampling
noise, high accuracy, and configurable data rate.
The other part is a data processing module. The key
chip on the module is an ultra low-power micro-
controller, MSP430 from Texas Instruments (TI).
The proposed computer science-based algorithm is
hosted on this module for compressed data analy-
sis. Additionally, Figure 4(b) shows that the whole
packaged system is very tiny when compared to the
size of a quarter (shown), yet the signal accuracy is
as accurate as those from a conventional scalp BCI
system. Figure 5 shows the researchers involved in
the project, analyzing the recorded brain signals in
their laboratory.
Currently, in order to prove empirically the
FIGURE 4
(A) PROTOTYPE CIRCUIT
(B) PACKAGED SMART HEADSET
IN ORDER TO PROVE
EMPIRICALLY THE FEASIBILITY
OF THE DESIGN FRAMEWORK,
THE GRADUATE RESEARCHERS
AT UCLA CONDUCTED A PILOT
STUDY WITH VOLUNTEERS
TO TEST THE SYSTEM.
9BRUIN INNOVATION & TECHNOLOGY
feasibility of the design framework, the graduate
researchers at UCLA conducted a pilot study to test
the system using volunteers. The preliminary pilot
experiments show that a CS-based signal process-
ing strategy could not only preserve the fidelity of
the data but also dramatically reduce the size of the
raw data. In addition, a paper about the BCI sys-
tem has been accepted by HCMDSS/MDPnP 2011
for publication [5]. Aside from fall prevention, it is
possible to consider a number of other potential
healthcare related applications that can be en-
hanced by this BCI system, including for example
brain-control for the impaired and disabled.
REFERENCES[1] Slobounov S, Cao C, Jaiswal N, Newell KM, “Neural basis of pos-
tural instability identified by VTC and EEG”, Exp Brain Res.
2009 Oct;199(1):1–16.
[2] D. Donoho. “Compressed sensing”. IEEE Transactions on Bio-
medical Engineering, 52:1289–1306, April 2006.
[3] E. Candes, J. Romberg, and T. Tao. “Stable signal recovery from
incomplete and inaccurate measurements”. Communica-
tions on Pure and Applied Mathematics, 59:1207–1223,
August 2006.
[4] Neurosky Inc. http://www.neurosky.com/
[5] Wenyao Xu, Fang Gong, Lei He, Majid Sarrafzadeh, “Wearable
Assistive System Design for Fall Prevention”, HCMDSS/
MDPnP 2011, Chicago, US (accepted for publication).
FIGURE 5
RESEARCHERS
ANALYZING DATA
For further reading: http://bit.ly/kDPm8B (published in HCMDSS workshop 2011)
10 VOLUME 1 / SPRING 2011
11BRUIN INNOVATION & TECHNOLOGY
BIT: Let’s begin with graduate life. In-between go-
ing to classes and having papers to publish, what
do engineers do, and how do they work? In groups?
Alone? How many projects are done simultaneous-
ly? What is the workflow from concept to comple-
tion?
Dean Dhir: Most students are taking classes and
working with faculty members in labs by keeping
current of the latest published papers as well as
carrying out research to compose their own papers.
Research involves building an experimental setup,
testing it to get data, then analyzing the data, and
finally deducting conclusions that advance their
knowledge in a particular area. The above would
summarize the activities of the average engineer-
ing graduate student. A small, exceptional group of
students go beyond that level with more initiative
and pursue research unrelated to their graduate
topic, out of curiosity, and to advance knowledge
within a field. Some of these exceptional students
also find applications that are more entrepreneur-
ial, in that they go one step more and commercial-
ize the idea. In large measure, it depends on the
individual.
On the other hand, faculty is focused in an in-
dividual discipline or area of expertise. They are
seeking to find out the truly innovative problems,
what are the “show-stopper” problems in that par-
ticular discipline. They cultivate ideas by writing
proposals for federal grants, and if their ideas and
the packaging are superior, then they get funding,
in the hopes of taking a project to high social and
financial payoffs.
BIT: What do engineers do that isn’t so obvious, yet
relevant, important, and interesting for a non-tech-
nical reader? What sorts of things involve engineer-
ing from the standpoint of health, entertainment,
scholarship, intellectual development, and social
responsibility?
Dean Dhir: Engineers are problem solvers, crea-
tors, and master builders. They create new ideas
and concepts and steer those efforts in the direc-
tion of fabricating a device based on those ideas,
testing it, troubleshooting it, and adjusting it incre-
mentally toward a novel innovation.
Furthermore, engineers must pay attention to
the consequences on the public of a gadget or de-
BIT CO-FOUNDERS JAMAL MADNI AND NEIL TILLEY SAT DOWN WITH DR. VIJAY DHIR, DEAN OF THE UCLA HENRY SAMUELI SCHOOL OF ENGINEERING & APPLIED SCIENCE (HSSEAS), WHO CONVEYED A PICTURE OF THE WORKFLOW, RELEVANCE, AND IMPACT OF ENGINEERING RESEARCH AT UCLA, AS WELL AS TO DESCRIBE TODAY’S TYPICAL ENGINEER.
THE DEAN’S CORNER
12 VOLUME 1 / SPRING 2011
vice of their design. Foremost this speaks of safety.
For example, take the discussion of cell phones:
there is the issue of microwaves from the antenna
in the phone and their effect on the brain. While
not entirely known, there are efforts exploring
whether they pose a tangible, albeit minimal, risk
on people. Engineers develop new technologies
and systems to the benefit of mankind and derive
a financial reward for these services. At the same
time they have a social and ethical responsibility
to be safe—a principle covered in the engineering
ethics classes offered in our school. There are peo-
ple who have made compromises in areas of safety;
these have come to haunt them. We’ve got to be
very careful at all times as to the consequences.
From an entertainment standpoint, we can
think of electronic arts: computer-aided design,
movie effects, sports arena jumbo screens, video
games—both standalone and within a smart phone,
for example. These are all part of everyday lives,
yet the underpinnings lie in computer science.
BIT: What makes engineers different people, and
what makes them people you might want to write
a movie about, such as The Social Network? If there
is a category called “engineer cool,” what comprises
that?
Dean Dhir: I think “engineer cool” is comprised of
people who go to an engineering school, acquire
an education in engineering principles, and then
use these principles to invent something that helps
the public do something faster and in a more eco-
nomical way than they are used to doing. You get so
much information on a cell phone now, for example,
that you can raise your productivity and have more
informed discussions.
Are engineers wired in a unique way? No, I
don’t think so. Engineering’s impact tends to be
more diffuse and generally affects many people,
more than the work of a medical specialist treating
a specific patient, for example. Broadly speaking,
engineers should go through a formal education—
this applies to 99% of the people in engineering;
perhaps 1% of the people are the exception, and
they may not need much formal training in princi-
ples, having an fundamental feel for it already. But
on the whole, good engineering is not limited to the
naturally gifted; rather it is much more fundamen-
tal than people think, because it is learnable.
BIT: Most of the great engineering pioneers in
recent memory dropped out of school to invent
a revolutionary idea in their garage. What ought
the school of engineering at UCLA offer (socially,
economically, politically, institutionally) to change
this paradigm and foster a fertile, attractive envi-
ronment benefiting the growth of the student and
school alike?
Dean Dhir: It’s the same thing you can find in ath-
letics—like Kareem Abdul-Jabbar, Bill Walton, or
Magic Johnson—relatively speaking, there are only
a few superstars—the outliers, if you will. However,
the majority of people, whether they are athletes or
engineers, need to have a good education so they
can thrive later in life. For example, the athletes
who aren’t capable of superstardom by skipping
college right away and directly going to the pros,
they cultivate their athletic skills in college for four
years in order to make a decent enough living in the
GOOD ENGINEERING IS NOT
LIMITED TO THE NATURALLY
GIFTED; RATHER IT IS MUCH
MORE FUNDAMENTAL THAN
PEOPLE THINK, BECAUSE
IT IS LEARNABLE.
13BRUIN INNOVATION & TECHNOLOGY
pros after that. So I don’t think institutionally there
needs to be a fundamental change in the paradigm.
However, from a cultural and educational
standpoint, we’ve got to provide our engineering
students the fertile settings that give an outlet to
their ideas. If they have ideas, they should be fos-
tered within the school. At the student level, for ex-
ample, there is the Technical Entrepreneurs Com-
munity (TEC), and the organization’s members are
creating an environment where students can learn
about how to market their technology; but at the
same time the university needs to create environ-
ments for students to bring their ideas from their
basic research and grow it where it becomes of in-
terest to venture capitalists. The Institute for Tech-
nology Advancement (ITA) helps with the matu-
ration process of specific technologies developed
in-house at UCLA, where venture capitalists will
take an interest to acquire this technology.
Also, we have been trying to raise funds for a
new Creativity Center, slated to be in place by the
end of this summer. The first function would be to
let young kids, including our students, build some-
thing unique, with their hands. But it is their own
imagination they are exercising. It could be junior
high and high school kids who come in summer. We
provide them with hardware, software, and techni-
cal support, in the form of mentors, and then let
them loose and have their imagination run wild. If
it doesn’t work, they troubleshoot it, fix it, test it,
and try again.
It will cost about $1.5 million; we have raised
about half that at this time. It will be down on the
interior courtyard level of Boelter Hall, and we
hope to have the rest of the funds available by the
end of summer. To operate and administer the
center, we will look to hire someone part-time and
staff graduate students as mentors. This is on the
cutting edge of creating an infrastructure for stu-
dent imagination, and UCLA is the first engineering
school in the nation to provide a facility of this type.
During the academic year, student clubs can access
it, and then this will provide hands-on experience
during the summers, for groups of three to four in
order to stir their creative thinking.
BIT: How can our engineering school outreach to
the rest of the university to have a hand in campus-
related problems?
Dean Dhir: It’s a very good question. I’ve been say-
ing this very thing on the campus committees I’m
part of: Why don’t we utilize the resources we have
in the school? Faculty can be the consultants. At
the moment, we hire from outside, which is so iron-
ic. For example, some of our faculty are the fore-
most authorities in the field of cyber-security, and
yet, we hire “outside” experts to deal with cyber-
security issues. Students can definitely help in the
area of Information Technology. But this mentality
is not institutionalized. I don’t know the reasons
why. I think it should be done, but where possible,
we should advertise we have the resources and
they should be used.
BIT: Why would I want to hire an engineer for a
non-engineering job?
Dean Dhir: Engineers are very good at analyzing.
They know how to solve an open-ended problem
where the solution may not be unique, but they
come up with an optimal solution. Engineers get
training with ill-defined problems and find very
creative, elegant, and efficient solutions to these
“asymmetric” problems. This type of thinking is
needed in many functions and industries. But,
broadly speaking, this isn’t trained in very many
other fields of study.
http://bit.ly/kezzdQhttp://bit.ly/mvEBIU
14 VOLUME 1 / SPRING 2011
You start watching the video. Behind the
scenes, the current Internet creates a link to one
of the YouTube servers, and the server streams the
video on this link. This happens with every single
person who watches the same video around the
globe, leading to a content distribution problem
that is solved today using a patchwork of Content
Distribution Networks.
Now suppose the router in your local area
network knew you were asking for the same video
your friend had requested a few minutes ago and
returned you a cached copy of the video it delivered
to your friend. Wouldn’t that be better? If we im-
plement this router intelligence on a global scale,
it forms an elegant way of solving the content dis-
tribution problem. This is but one of the major fea-
tures of a new Internet architecture called Named
Data Networking, or NDN, being developed here at
UCLA in association with PARC (a Xerox company)
and other partner institutions including the Univer-
sity of Arizona, UC Irvine, UC San Diego, Colorado
State University, the University of Illinois Urbana-
Champaign, the University of Memphis, Washing-
ton University, and Yale University.
DESIGNING APPLICATIONS USING NAMED DATA NETWORKING
Today’s Internet uses IP addresses, which are
number addresses to identify physical computers
at which the data can be found, in order to move
(route) data from one place to another. The vision
of NDN however is a new direction, to give names
to the data itself that applications and users care
about instead of addressing the machine where
they can be found. This allows applications simply
to ask the network for the data they want instead of
specifying where to find the data. It is the network’s
job to locate and deliver data to the requesting ap-
plication.
So how does NDN work? Applications send an
Interest packet, which carries the name of the de-
sired content (a name such as /parc.com/videos/
WidgetA.mpg as shown in Figure 1). The router that
receives the Interest packet remembers the inter-
face from which the request comes in, and forwards
the packet to other nodes (a node is a data source
or a router on the path to the data source) by look-
ing up the name in its Forwarding Information Base
(FIB). Once the Interest reaches a node that has the
requested data, a Data packet is sent back that car-
ries both the name and content of the data, togeth-
Rahul Rao Basava, Derek Kulinski, Chenni Qian, and Zhenkai Zhu, Computer Science
YOUR FRIEND ACROSS THE ROOM IMs YOU TO WATCH THE NEW YOUTUBE VIDEO THAT WENT VIRAL AN HOUR AGO.
15BRUIN INNOVATION & TECHNOLOGY
er with a digital signature by the creator of the data.
The Interest packet leaves a breadcrumb trail
across the network (the breadcrumb is an entry in
the Pending Interest Table, or PIT, of a router) for
the Data packet to retrace the path to the applica-
tion that requested it. Neither Interest nor Data
packets carry IP addresses; Interest packets are
routed towards data producers based on the names
carried in the Interest packets, and Data packets
are returned based on the state information set up
by the Interest packets at each step along the route.
When multiple Interests requesting the same data
are received, the router forwards only the first In-
terest packet towards the data source. When the
Data packet returns along the path from the source,
the router sends out copies of the packet in all di-
rections from which Interest packets had arrived
earlier. Then the router clears the corresponding
PIT entry and caches the data in its Content Store
to satisfy future requests for the same data.
Three driver applications are being developed
at UCLA to drive and test the prototype implemen-
tation of the NDN architecture. These applications
leverage built-in features such as efficient content
distribution and packet authentication. Each of
these projects represents a broad area of applica-
tion design, and each is expected to yield further in-
sight into the requirements of the new architecture.
At the same time they demonstrate performance
and functional advantages of NDN in key areas and
show how NDN’s embedding of application names
in the routing system promotes efficient authoring
of sophisticated distributed applications with re-
duced complexity. An introduction to each of these
applications follows.
FIGURE 1
NAMED DATA
NETWORKING - OVERVIEW
THE INTEREST PACKET
LEAVES A BREADCRUMB
FOR THE DATA PACKET
TO RETRACE THE PATH
TO THE APPLICATION THAT
REQUESTED IT.
16 VOLUME 1 / SPRING 2011
LIGHTING APPLICATION PROJECTBuildings of the future are likely to be con-
structed with digitally controlled and networked
facilities such as lighting, touch-activated displays,
sound systems, and video recording. The current
Internet architecture makes it cumbersome to
build applications for such systems with hetero-
geneous components, especially in the area of ad-
dressing schemes and security.
The Lighting Application project aims at cre-
ating a control system using NDN that will deal
with physical lights and building facilities. NDN’s
intrinsic support for naming data, broadcast, cach-
ing, and fine-grained authentication provide obvi-
ous advantages in this setting. Exploration of this
area expands on UCLA’s experience building in-
teractive spaces of the kinds found in museums,
physical simulation spaces, theme parks, and live
performances. The project will create an applica-
tion library to support naming of various devices;
a discovery mechanism for embedded devices; dis-
tributed, synchronized state management; effective
key management; and distribution for authentica-
tion of control system commands.
Figure 2 shows the various components of
the project. Fixtures are lighting facilities that are
used to illuminate or create lighting in an envi-
ronment. They can be a single luminary unit or a
lighting device that has multiple light units such as
lighting strings. Lighting Interfaces are an inter-
mediary between fixtures and the NDN network,
sending and receiving Interest and Data packets
from Building Services. From the perspective of
the network, the interfaces and their associated
fixtures are indistinguishable. The fixtures used in
our test project are Philips ColorBlast 12 LED wall-
washing lights connected to Gumstix - Overo Air
COM interfaces.
As shown in the diagram, Building Services
are applications that constitute the building con-
trol systems. The Configuration Manager Service
is responsible for assigning the names for fixtures,
maintaining cryptographic certificates and ac-
cess control lists (ACLs) for other services in the
network. Lighting interfaces request this service
for names using an Interest packet. The name is
FIGURE 2
LIGHTING APPLICATION
PROJECT – OVERVIEW
17BRUIN INNOVATION & TECHNOLOGY
returned as data. Subsequently the service sends
Interest packets to the interfaces for a periodic
‘heartbeat’ message, for checking connectivity and
status.
The Architecture Lighting Service is respon-
sible for everyday building lighting control. This
service sends control commands as Interest pack-
ets to the Interfaces (for example, UCLA/boel-
ter/3551/lights/fixture/*/rgb-8bit-hex/FAF87F,
as shown in Figure 3). Here the Interest packet is
used as a signed command, while the returning
data packet carries a simple signed acknowledge-
ment of reception of the command. The command
carries the RBG (red-blue-green) color settings for
the fixture. This service may be extended to provide
real-time status of the lighting infrastructure. The
Fire Life Safety Service includes features like fire
prevention and emergency evacuation. This service
has highest priority, and its commands override all
other service commands. Environment Monitoring
Service allows building administrators to view fix-
ture status and compute statistics regarding energy
consumption of the building.
PERSONAL DATA CLOUD PROJECTIn a world filled with mobile devices where
activity monitoring sportswear (like the Nike+iPod
system*) is becoming more and more common,
personal data is also becoming ever more accessi-
ble. Data collected by applications on such devices
include your location, time of events in your day-to-
day life, pictures (geo-tagged and time stamped),
and other information that the individual is record-
ing (ranging from eating habits to daily memos).
The Personal Data Cloud (PDC) project aims
to build a secure data ecosystem that can be con-
trolled by a single user. The PDC is composed of
many entities such as mobile phones and sensing
devices, cloud-based secure repositories, and per-
sonal computers. The concept is illustrated in Fig-
ure 4. Creating the PDC will require the implemen-
tation of a distributed database on NDN with direct
name-based access of individual records, providing
a widely applicable test case for validating NDN-
based systems like data discovery, content distri-
bution, and diverse models for trusted communica-
tion among mobile users, PDCs, and applications.
FIGURE 3
COMMAND FLOW IN ARCHI-
TECTURE LIGHTING SERVICE
* see article on page 24
18 VOLUME 1 / SPRING 2011
The major features of the PDC are disruption-toler-
ant uploads from sensors to the secure data store;
internal communication between nodes of the
PDC; and filtered, often anonymous data sharing
with authorized third-party applications providing
personal services to the individual. The PDC design
envisions all entities (mobile devices included) as
data repositories with varying capacities for stor-
age and processing.
A new protocol called the Stream Protocol has
been defined for communication between entities
in a PDC. A stream is an abstraction of a sequen-
tial set of data sent from one entity to another. The
Stream Protocol is designed to work over the NDN
and in practice replaces the Hypertext Transfer
Protocol Secure (HTTPS) modules of the applica-
tions on the current Internet (HTTPS being the
most common protocol on the Internet used for
any secure communication, for example, purchas-
ing goods online). Figure 5 shows a high-level com-
parison between the module structure of Stream
Protocol and HTTPS. An entity that sends data is
called a Publisher and the entity receiving the data
is called a Receiver. An entity can be both a Publish-
er and a Receiver. Every Publisher can be the source
of many such streams.
AndWellness is one such personal data collec-
tion system developed in the Center for Embedded
Network Sensing (CENS). It uses mobile phones to
collect and analyze data from active/triggered user
experience samples and passive logging of onboard
environmental sensors, for example the GPS locator
and accelerometer. The system includes an applica-
tion that runs on Android-based mobile phones,
server software that manages deployments and
acts as a central repository for data, and a dash-
board front-end for participants and researchers
alike to visualize incoming data in real time.
AndWellness will be the first framework to run
over the PDC architecture. The mobile application
component has been modified to behave as an in-
dependent entity capable of publishing informa-
tion to the user’s data cloud. The data receiver is
a cloud-based, secure repository that is considered
as a test destination for the data streams published
by the mobile application. While the mobile appli-
FIGURE 4
PERSONAL DATA CLOUD
PROJECT - OVERVIEW
19BRUIN INNOVATION & TECHNOLOGY
cation does not have the functionality to receive
data and does not have filtering components, the
cloud-based repository will have both capabilities.
AUDIO CONFERENCING PROJECTCurrent audio conferencing tools rely on a
centralized system to discover ongoing conferenc-
es, the speakers participating in each conference
and transmission of voice data from individual
speakers. The Audio Conference Tool (ACT) project
explores multi-host real-time media streaming ap-
plications designed over NDN. ACT takes a named-
based approach, and the resulting design is com-
pletely distributed and robust against component
failures. ACT supports both conference data au-
thentication and participant control. The packets
generated by ACT are digitally signed using the
built-in security primitive in NDN, which guaran-
tees data integrity and provenance. An encryption-
based access control scheme is employed in the pri-
vate conference mode where a session key is used
to encrypt all data traffic of the conference and is
only distributed to legitimate users. ACT will test
the effectiveness of NDN in the context of time syn-
chronization, quality-of-service selection, security
of control / media streams, routing scalability, and
trust model.
Figure 6 shows an overview of the ACT design.
A separate module handles conference discovery—
both scheduled and ongoing—and facilitates the
task of joining a conference. The decoupling of con-
ference discovery from voice media delivery gives
the flexibility to extend ACT with other features
such as video, whiteboard, and text message. A user
who initiates a conference creates data that de-
scribes the various aspects of the conference, such
as estimated start time, duration, media type, pur-
pose, etc. Conference discovery requires propaga-
tion of Interest packets from participants in a given
conference for conference information across the
network. This is achieved by choosing the names
for conference information data from a broadcast
namespace, which is a dedicated set of names con-
figured on the routers for network broadcast. Once
conference discovery is completed, the speakers of
each ongoing conference are discovered. Partici-
pants in a conference should answer the Interests
for speaker information sent by listeners.
The Interest packets for voice data streams
need to be generated at frequencies orders of mag-
nitude higher than those needed for conference
setup and speaker management; thus broadcast
is not used for voice streaming. ACT uses speaker-
specific names for voice data, and this task is del-
egated to the Voice Data Distribution module. Voice
data is fetched by sending Interests directly to each
participant speaker in a conference. Pipelining of
Interests and flow balance are enforced to mitigate
the effects caused by the delay of round trips and
loss of packets, thus smoothing out the sound of the
voice transmission.
FIGURE 5
COMPARISON BETWEEN
HTTPS AND STREAMS
20 VOLUME 1 / SPRING 2011
EXPLORATION OF THIS
AREA EXPANDS ON UCLA’S
EXPERIENCE BUILDING
INTERACTIVE SPACES
OF THE KINDS FOUND
IN MUSEUMS, PHYSICAL
SIMULATION SPACES,
THEME PARKS, AND LIVE
PERFORMANCES.
21BRUIN INNOVATION & TECHNOLOGY
Media data processing and user interface de-
sign are necessary components of ACT but largely
decoupled from networking specifics. To focus our
effort on NDN-specific design issues, ACT adopted a
client-server based open-source audio application
package. It embeds the modules for speaker dis-
covery and voice data distribution in the original
server code, leaving the client intact. This modified
server runs on the same machine as the client and
communicates with the client using the standard IP
protocol stack. At the same time it communicates
with other participants (who may be using the
modified server) over NDN.
REFERENCES[1] Named Data Networking (NDN) Project. www.named-data.net.
[2] Named Data Networking (NDN) Project. PARC Technical Re-
port NDN-0001. L. Zhang, D. Estrin, J. Burke, V. Jacobson, J.
D. Thornton, D. K. Smetters, et al. http://www.named-data.
net/ndn-proj.pdf.
[3] AndWellness: An Open Mobile System for Activity and Expe-
rience Sampling. John Hicks, Nithya Ramanathan, Donnie
Kim, Mohamad Monibi, Joshua Selsky, Mark Hansen, De-
borah Estrin. Wireless Health ’10. http://openmhealth.org/
wp-content/uploads/wireless-health.pdf.
FIGURE 6
CONFERENCING
PROJECT - OVERVIEW
Many thanks to Jeff Burke, Lixia Zhang, Deborah Estrin, and Alessandro Marianantoni at UCLA and to the researchers of UCLA partner institutions for their support.
22 VOLUME 1 / SPRING 2011
23BRUIN INNOVATION & TECHNOLOGY
ENTREPRENEURIAL OPPORTUNITIES FROM THE OTHER SIDE OF CAMPUS
Jeff Chang, UCLA Anderson School of Management
WHAT IF YOU WANT TO BE IN A CREATIVE ROOM THAT SPAWNS MORE
IDEAS AND INSPIRES YOU? WHAT IF YOU WANT A CAREER AS A VENTURE
CAPITALIST? WHAT IF YOU HAVE AN IDEA AND WANT TO KNOW WHAT
TO DO NEXT?
Odds are, there is something in North Campus for you.
Enroll in TIP (Technology & Innovation Partners)
TIP is a yearlong course for accelerating research commercialization. It brings together students from vari-
ous UCLA graduate schools. TIP classes are taught by strong professors and guest speakers under the lead-
ership of Al Osborne, Jr., Senior Associate Dean and founder of the Price Center for Entrepreneurial Studies.
Become part of the UCLA Entrepreneur Association
The Anderson EA organizes nearly 150 events per year. Many of those events are open to all UCLA students.
A few examples are the following:
• eLabs – where students pitch their ideas and get feedback and support from the audience.
• Knapp Prep events get you ready to put together a business plan.
• eWeek – a week of seminars and keynotes with successful entrepreneurs.
• The EA Conference in mid-May – an entire day of panels, speakers, the fast-pitch competition, and op-
portunities to make contacts at breakfast, lunch, and happy hour.
You can also compete in a business plan competition any time of year. As a UCLA grad student, you are
encouraged to contact any of a hundred Anderson students through the site www.KnappComp.com if you have
an idea and need to put together a team. A feature on the site is a list of all one hundred worldwide, open-
format business plan competitions for the upcoming year – So stay tuned!
TIPhttp://bit.ly/lScAWQ
Entrepreneur Associationhttp://bit.ly/lbjspT
Knapphttp://bit.ly/ieDMVN
24 VOLUME 1 / SPRING 2011
A WIRELESS HOME AUTOMATION SYSTEM FOR CHILDHOOD OBESITY
PREVENTION
Childhood obesity is recognized as a seri-
ous public health concern due to the rising preva-
lence of obesity in children (Troiano 1995). In the
United States, direct measurements of body mass
and height obtained by the National Health and
Nutrition Examination Survey indicate that about
15% of 6–19 year olds are classified as overweight
(Ogden 2002). As the children spend a significant
amount of their time at home, a sedentary lifestyle
accounts for the leading cause of childhood obes-
ity (Walker 1998). Many children fail to exercise
because they spend time doing stationary activities
such as playing video games or watching TV. Cer-
tain reports provide evidence that television view-
ing is a reason for increased body fatness and that
reducing television viewing is a promising strategy
for preventing childhood obesity (Andersen 1998,
Robinson 1998).
On the other hand, exercise would help chil-
dren control their weight. It also helps to reduce
the risk of some illnesses such as high blood pres-
sure, heart disease, sleep problems, and other simi-
lar disorders (Freedman 1993). Based on the men-
Navid Amini, Computer Science
PEOPLE ARE BECOMING MORE CONCERNED WITH THE PROBLEM OF CHILDHOOD OBE-SITY, WHEN EXCESS BODY FAT NEGATIVELY AFFECTS A CHILD’S HEALTH.
tioned facts, it is widely recommended that one
should moderately exercise for at least 30 minutes
five times a week (Sallis 1994, Pate 1998, Corbin
1994). Furthermore, it is reported that standards
for recommended pedometer-determined steps
per day for 6–12 year olds are 12,000 for girls and
15,000 for boys (Tudor-Locke 2004). This high-
lights the importance of physical activity for chil-
dren. Accordingly, many control units have been
implemented in games and appliances by parents
in an effort to encourage children to perform physi-
cal exercise. Using these units, parents can limit the
time that their children can spend playing compu-
ter or TV games by manually controlling the cor-
responding appliances using the GUI provided by
these units (Thompson 2006, Coshott 2007). How-
ever, the units are not capable of preventing chil-
dren from watching another channel or playing an-
other game that is not equipped with the parental
control option.
In this paper, the system proposed can miti-
gate the previously mentioned problems by con-
trolling power outlets instead of controlling certain
25BRUIN INNOVATION & TECHNOLOGY
TV channels or particular games. It has been sev-
eral years since the introduction of smart homes. In
such homes, RF (radio frequency) signals from spe-
cial controllers are exploited to activate or deacti-
vate power outlets. In some cases, the controller is
able to regulate the intensity of room light as well.
A number of research projects have been carried
out based on these or similar controllers to build
smart homes (Cook 2003, Jiang 2000, Cole 2002).
Likewise, a certain type of RF power controller is
used in this project in order to acquire complete
control over the home’s power outlets. In addition,
a wireless sensor is utilized that can calculate the
travelled distance and average pace for a walk (or
a run) so as to monitor the daily activity level of
children. The recorded data can later be uploaded
to any computer via the USB interface. The goal of
the proposed system is to make a virtual connec-
tion between power outlets and a wireless activity
monitor sensor. This way, the children could be en-
couraged to do more exercise in exchange for more
time of home entertainment (e.g., watching TV).
The more exercise children perform, the more time
they secure to do stationary activities. The pro-
posed system, called No Pain No Game, is intended
to prevent childhood obesity disease and thus im-
prove childhood fitness levels.
The remainder of the paper is organized as
follows. Section 2 provides a brief overview of re-
lated work in the context of home automation and
childhood obesity prevention. Preliminary notions
followed by the structure of the No Pain No Game
system are covered in Section 3. The important fea-
tures of the system are summarized in Section 4. Fi-
nally, Section 5 concludes this study and highlights
future research directions.
RELATED WORKThere are several research projects based
on home automation systems. As an example,
MavHome smart home architecture developed at
the University of Texas at Arlington allows a home
to act as an intelligent agent (Cook 2003). In the
MavHome system, a sensor network including
light, humidity, temperature, smoke, gas, motion,
infrared, and switches is developed to keep track of
the home environment. Based on the data collected
from the sensor network, a server makes a decision
for resource management and executes the decision
through the smart actuators (e.g., X10 ActiveHome
Kit) with which the appliances in the home are
equipped. Generally, the MavHome aims at maxi-
mizing the inhabitant’s comfort and productivity
and minimizing costs. For elderly and disabled peo-
ple, the system can provide a fine health care and a
higher level of security. As another example, to help
people with disabilities, Jiang et al. has designed
a voice-activated environmental control system
which consists of a universal remote control with
X10 home automation capability, a Motorola 6811
microprocessor, and an off-the-shelf voice recogni-
tion circuit (Jiang 2000).
In the context of activity monitoring, in a
number of projects X10 components are integrated
into a toolkit which is intended to monitor the daily
activity of an individual (Cole 2002). In addition to
the home automation systems, certain appliances
are developed to encourage children to exercise.
For example, an interrupter system can be added to
an existing connection between a game console and
its controller (Coshott, 2007). The interrupter sys-
THE PROPOSED SYSTEM
CONTROLS POWER OUTLETS
INSTEAD OF CONTROLLING
CERTAIN TV CHANNELS OR
PARTICULAR GAMES.
26 VOLUME 1 / SPRING 2011
tem is used selectively to enable and interrupt the
modified connection between the game controller
and the game console, dependent on the detected
operation of the exercise machine. Accordingly, the
parents can arrange it in such a way to block nor-
mal playing on the game console unless the child
performs an adequate amount of exercise on the
exercise machine.
In (Annavaram 2008), a wireless body area
network is developed for wearable monitoring ap-
plications, and the intended use of the system is to
avoid pediatric obesity. The network is composed
of heterogeneous sensors (e.g., heart-rate sensor
and accelerometer sensor), and the goal is to rec-
ognize, predict, and reason automatically about
human physical activity and behavior states by the
evaluation of multimodal sensing and interpreta-
tion.
THE PROPOSED SYSTEMNo Pain No Game is a home automation sys-
tem that encourages children’s daily exercise by
rewarding them with more home entertainment
time. In other words, the time children spend doing
stationary activities should be proportional to the
time they spend doing physical exercise. Aside from
the interface software, the system is composed of
three parts: a sport kit that monitors the exercise
records, a controller that commands the power
outlets, and a database that stores the health sta-
tistics and exercise records of children. The current
version of the system exploits Nike + iPod Sport Kit
as a means to monitor the physical exercise. It also
uses X10-based devices in the shape of controlla-
ble power outlet modules in order to acquire full
control over different entertainment appliances.
Microsoft Access databases are utilized to store the
relevant exercise data. The software that connects
these three parts is written in Visual Basic and C
Sharp and runs on a Windows platform.
X10 Standards
X10 is an international and open industry
standard for communication among electronic de-
vices used for home automation. It primarily uses
power line wiring for signaling and control, where
the signals involve brief radio frequency bursts
representing digital information. A wireless radio
protocol is also defined, where the data packets
are very similar to those used for power line wires.
The operating frequency of the wireless protocol
is 433 MHz and 310 MHz in the European systems
and U.S. systems, respectively. It should be noted
that the wireless protocol allows the operation of
keypad remote controllers on top of the underlying
wired X10 modules.
X10 is popular in the home environment, with
millions of units in use worldwide and new compo-
nents inexpensively available. Figure 1 illustrates a
typical configuration for an X10 network.
By using The ActiveHome Pro Scripting In-
terface provided by X10, one can create software,
FIGURE 1
A TYPICAL CONFIGURATION FOR
X10 CONTROLLED APPLIANCES.
27BRUIN INNOVATION & TECHNOLOGY
web pages, and other tools that use the USB In-
terface to control and interact with X10 modules,
sensors, and remote controls (The ActiveHome Pro
SDK from X10). The address of each module is set
by the dials located on it. Accordingly, one can com-
mand the interface to turn on the intended module
by providing a valid address corresponding to the
module. Upon receiving a command (e.g., from a
remote control), the interface reports the action of
receiving the command which enables the user to
benefit from certain options. For the sake of more
reliability, each packet is sequentially sent twice to
make sure the receivers understand it even in the
presence of power line noise.
Nike + iPod Sport Kit
The Nike + iPod Sport Kit (Figure 2) consists
of a wireless sensor and a small wireless receiver
that plugs into an iPod. The sensor is a piezoelec-
tric accelerometer pedometer that fits inside spe-
cial Nike shoes and a wireless receiver that is con-
nected to an iPod. Each sensor has its unique serial
number that is used as an identification number for
each child in the proposed system. The personal
training application on the iPod can provide infor-
mation on distance and speed while a user is listen-
ing to music. By considering the number of steps
taken and the elapsed time, users (children) can
schedule their desired workout in the form a speci-
fied distance or a certain time period that can fit a
plan based on their goals and their previous per-
formance. The exercise records are stored in files
with XML format. The proposed system will upload
those files to the server and extract the useful data
from the files. Therefore, the data can be analyzed
to decide how much entertainment time the child
has earned.
The Proposed No Pain No Game System
In the proposed system, a server is main-
tained by parents while children will use remote
controls sending matching signals to trigger the
appliances. A child must register his or her sports
kit on the server, given that every sensor has a par-
ticular identification number. The proposed sys-
tem simultaneously analyzes the current exercise
records when children update the database with
new exercise records merely by plugging their iPod
into the client computer. This means that the server
automatically calculates the total time budget that
a child is allowed to use his/her intended applianc-
es (e.g., TV). A formula is applied to derive the time
budget based on the calories burned by the child:
where MET (Metabolic Equivalent of Task) is a
physiological concept that shows how intense the
exercise is (Dr Gily’s Health Portal), w is the child’s
FIGURE 2
APPLE NIKE + IPOD SPORT KIT.
THE SENSOR FITS INSIDE
SPECIAL NIKE SHOES AND
A WIRELESS RECEIVER IS
CONNECTED TO AN IPOD.
Calories METs w T=× ×
×3 5
200
.(1)
28 VOLUME 1 / SPRING 2011
weight in kilograms, and T represents the total du-
ration of exercise (e.g., running or walking) in min-
utes.
MET expresses the energy cost of physical ac-
tivities as multiplies of the Resting Metabolic Rate
(RMR) and is defined as the ratio of metabolic rate
(and therefore, the rate of energy consumption)
during a certain physical activity to one’s metabolic
rate at rest (quiet sitting), set by convention to 3.5
ml O2 . kg-1 . min-1 or equivalently 1 kcal . kg-1 . hr-1. By
convention 1 MET is considered the resting meta-
bolic rate obtained during quiet sitting. MET values
of physical activities range from 0.9 (sleeping) to
18 (running at 17.5 km/h). It should be noted that
the MET value for watching television is 1 and for
walking at a speed of 3.2 km/h is about 2 (Meta-
bolic Equivalent 2009).
After the burned calories are derived, a ba-
sic requirement for this value is enforced. If the
amount of burned calories is less than a certain
quantity (called the basic requirement), no time
budget will be granted. However, if it exceeds the
basic requirement, the basic amount of time will be
given, plus extra time depending on the bonus rate
and the overhead. The required amount of calories,
the amounts of the basic time budget and extra
time can all be adjusted by the Account Manage-
ment Interface.
The total time budget for each child is stored
in a database by Microsoft Access. The time budget
gets verified and consumed every time a child tries
to activate an appliance. Moreover, after the time
budget is depleted, the server will automatically
power off the related appliance (e.g., TV). A child
has to perform more exercise and then plug in his/
her iPod into the client computer in order to in-
crease the time budget again. Figure 3 illustrates
the structure of the No Pain No Game system.
The implementation of the system is con-
structed with two ends, the client and the server
(Figure 3). On one end, the program on client com-
puters will parse iPod exercise records by creating
XML readers supported by the .NET Framework.
After validation, the files will be sent to the serv-
er unless they are expired. On the other end, the
server will increase the corresponding time budget
by an appropriate formula based on children’s
BMI and exercise records. The server also verifies
whether the sensor identification number of the
record is valid, and it makes sure that the record
is not outdated. Meanwhile, if one wants to regis-
ter more children (through Account Management)
or verify the status of all children, the server can
update or retrieve the data as required (see system
block diagram illustrated in Figure 4).
Moreover, the server handles the events when
the Power Control Module (ActiveHome Automa-
tion System interface) receives signals from remote
controls. Each received signal will be mapped to
a particular child and the module he/she has re-
quested. Therefore, the server can retrieve the cor-
responding data and determine if it should send a
command (sendrf in this case) to the automation
system in order to trigger the requested module.
In addition, for each appliance, there is a timer
in the server to update and monitor the available
time budget and, correspondingly, to turn off the
appliance when the time budget has exhausted. A
detailed block diagram of the server is depicted in
Figure 5.
IN THE PROPOSED SYSTEM,
A SERVER IS MAINTAINED BY
PARENTS WHILE CHILDREN
WILL USE REMOTE CONTROLS
SENDING MATCHING SIGNALS
TO TRIGGER THE APPLIANCES.
29BRUIN INNOVATION & TECHNOLOGY
SYSTEM FEATURESSince each child has his/her own iPod + Nike
Sport Kit, the system can distinguish between dif-
ferent children by evaluating their identification
numbers. Furthermore, the system takes into ac-
count the date of exercise records to avoid copies
of previous exercise records from being taken into
consideration. This validation process prevents
children from cheating to some extent. For exam-
ple, a child will not gain any extra time budget by
plugging in the iPod twice with the same exercise
record.
As can be seen in Figure 6, a user-friendly GUI
on client computers is provided for children to use
so that they can easily learn how to update the ex-
ercise records without any difficulty. Furthermore,
the system allows children to watch TV or use other
entertainment appliances only if they have per-
formed an adequate amount of exercise throughout
the day. This gives the children a sense of achieve-
ment while the parents need not worry about their
kids being too inactive. Additionally, parents can
record the time and let the system monitor the ac-
tivity level of their children and the status of the
entertainment appliances. They can view earlier
data (e.g., how many calories have been burned)
and observe the progress of their child’s fitness
level. Figure 7 shows an example table containing
a child’s recent exercise data. This exercise table is
accessible from the server.
DISCUSSIONS AND FUTURE WORK In the proposed system’s current form, after
performing exercise, a child connects the iPod to
the client computer. This will update a database on
the server, and eventually a new schedule for the
related appliances will be sent from the server to
the ActiveHome Automation System. However, it is
more desirable to operate the system in real time,
where the updating process is done automatically
and wirelessly. This can be done by using a Blue-
tooth adapter connected to the iPod. This way, there
is no need for the child to connect the iPod to the
computer in order to extend his/her time budget.
Apart from Nike + iPod Sport Kit, other types
of sport kits can be added to the system in future to
FIGURE 3
STRUCTURE AND APPLICATIONS
OF NO PAIN NO GAME.
30 VOLUME 1 / SPRING 2011
FIGURE 4
SYSTEM BLOCK DIAGRAM
FIGURE 5
DETAILED BLOCK
DIAGRAM OF THE SERVER.
31BRUIN INNOVATION & TECHNOLOGY
provide the children with more options to perform
physical exercise. Also it should be noted that the
server side and client side of the system (Figure 3)
can run on two networked computers, or they can
run on a single machine.
A new method of gaming can be designed in
order to make players benefit from doing physi-
cal exercise in the real world. This can be achieved
by extending the system to a server maintained
by game companies. Similar to the design for the
Home Automation System, the system now rewards
children who have performed a sufficient amount
of exercise with the profit in the game. For example,
a child’s character in the game becomes stronger
and thus is able to carry better weapons. It is antic-
ipated that this idea gives encouraging incentives
for children in exchange for performing physical
exercise.
No Pain No Game can also cooperate with
a children’s health organization to advocate for-
mulating an exercise plan, encouraging children
by a competitive display record on the organiza-
tion’s website. A child can upload his/her exercise
progress and get ranked, accordingly. Further, the
exercise data collected from children can be a use-
ful resource for related research institutes. People
in health-related organizations can analyze how
the trend of health data progresses during the ex-
ercise.
CONCLUSIONSChildhood obesity has been a public concern
as people expend money and resources in order
to maintain their children’s physical fitness and
overall health. As such, it is necessary to build an
application that continually encourages children to
perform physical exercise. Several parental control
units have been designed, but few of them can ef-
fectively impact children. With home automation
system technology ever improving and with the
number of related research projects growing, the
proposed system is built on the idea of home au-
tomation systems. In this system, called No Pain No
Game, children have to perform exercise for enough
time in order to be able to activate any entertain-
ment appliance. The activity level of children is
monitored using a Nike + iPod Sport Kit, which is
becoming more popular. It records the number of
steps taken and time elapsed. Therefore, the appli-
cation can calculate calories and other related data
from the records. A database is utilized to store the
exercise records and the time budget for every reg-
istered child.
No Pain No Game lowers the risk of being
cheated by means of its validation process. Also,
its user-friendly GUI easily allows children to learn
how to extend their entertainment time. Children
can obtain a sense of achievement, while parents
can let the system automatically monitor the ac-
FIGURE 6 (TOP)
A TYPICAL SCREENSHOT
OF THE CLIENT GUI IN
NO PAIN NO GAME.
FIGURE 7 (BOTTOM)
AN EXAMPLE TABLE
OF THE EXERCISE DATA.
32 VOLUME 1 / SPRING 2011
NO PAIN NO GAME CAN ALSO
COOPERATE WITH CHILDREN
HEALTH ORGANIZATIONS
TO ADVOCATE FORMULATING
AN EXERCISE PLAN…
A CHILD CAN UPLOAD HIS/HER
EXERCISE PROGRESS AND GET
RANKED ACCORDINGLY.
33BRUIN INNOVATION & TECHNOLOGY
tivity level of their children and the status of the
appliances. It is believed that a complete No Pain
No Game product would enable the parents to su-
pervise their children’s exercise and entertainment
inexpensively and in a completely unobtrusive
fashion.
REFERENCES
[1] Troiano, R.P., Flegal, K.M., Kuczmarski, R.S., 1995. Overweight
prevalence and trends for children and adolescents. Arch
Pediatr Adolesc Med 1995;149:1085– 91.
[2] Ogden, C.L., Flegal, K.M., Carroll, M.D., Johnson, C.L., 2002. Prev-
alence and trends in overweight among US children and
adolescents, 1999 – 2000. JAMA 2002;288(14):1728– 32.
[3] Walker, A.R.P., Walker, B.F., 1998. Rises in schoolchildren’s an-
thropometry: what do they signify in developed and devel-
oping populations? J R Soc Health 1998;118(3):159– 66.
[4] Andersen, R.E., Crespo, C.J., Bartlett, S.J., Cheskin, L.J., Pratt,
M., 1998. Relationship of physical activity and television
watching with body weight and level of fatness among chil-
dren: results from the Third National Health and Nutrition
Examination Survey. JAMA 1998;279:938–42.
[5] Robinson, T.N., 1998. Does television cause childhood obesity?
JAMA 1998; 279: 959–60.
[6] Freedman, D.S., Dietz, W.H., Srinavisian, S.R., Berenson, G.S.,
1993. The relation of overweight to cardiovascular risk fac-
tors among children and adolescents: the Bogalusa heart
study. Pediatrics 1999; 103: 1175–1182.
[7] Sallis, J.F., Patrick, K., 1994. Physical activity guidelines for
adolescents: consensus statement. Pediatr Exerc Sci 1994;
6: 302 –314.
[8] Pate, R., Trost, S., Williams, C., 1998. Critique of existing guide-
lines for physical activity in young people. in Biddle, S., Sal-
lis, J., Cavill, N. (Eds), Young and Active? Young People and
Health-Enhancing Physical Activity-Evidence and Implica-
tions, Health Education Authority, London, pp. 162–76.
[9] Corbin, C.B., Pangrazi, R.P., Welk, G.J., 1994. Toward under-
standing of appropriate physical activity levels for youth.
President’s Council of Physical Activity and Sport. Phys Act
Fit Res Digest 1994;1(8):1– 8.
[10] Thompson, D.I., Cullen, K., Baranowski, T., 2006. Using Com-
puter Games and Other Media To Decrease Child Obesity.
Agricultural Research 2006 March.
[11] Coshott, R.J., 2007. Encouraging exercise whilst playing elec-
tronic games. USPTO Application number: 20090098979.
[12] Tudor-Locke, C., Pangrazi, R.P., Corbin, C.B., Rutherford, W.J.,
Vincent, S.D., Raustorp, A., et al, 2004. BMI-referenced
standards for recommended pedometer-determined
steps/day in children. Preventive Medicine, 2004.
[13] Cook, D.J., Youngblood, M., Heierman, E., Gopalratnam, K.,
Rao, S., Litvin, A., Khawaja, F., 2003. MavHome: An agent-
based smart home, in: Proceedings of the IEEE International
Conference on Pervasive Computing and Communications,
2003, pp. 521–524.
[14] Jiang, H., Han, Z., Scuccess, P., Robidoux, S., Sun, Y., 2000.
Voice-activated environmental control system for persons
with disabilities. Proc. of the IEEE 26th Annual Northeast
Bioengineering Conference, 2000, pp. 167–169.
[15] Cole, A., Tran, B., 2002. Home automation to promote inde-
pendent living in elderly populations. Presented at Proceed-
ings of the 2002 IEEE Engineering in Medicine and Biology
24th Annual Conference and the 2002 Fall Meeting of the
Biomedical Engineering Society (BMES / EMBS), Oct 23–26
2002, Houston, TX, United States, 2002.
[16] The ActiveHome Pro SDK from X10, http://www.x10.com/
activehomepro/sdk/
[17] Dr. Gily’s Health Portal, http://www.drgily.com/exercise-cal-
orie-counter.php
[18] Annavaram, M., Medvidovic, N., Mitra, U., Narayanan, S., Spru-
ijt-Metz, D., Sukhatme, G., Meng, Z., Qiu, S., Kumar, R., and
Thatte, G., 2008. Multimodal sensing for pediatric obesity
applications. In UrbanSense08 Workshop at SenSys Raleigh,
NC, USA, November 2008.
[19] Metabolic Equivalent of Task (MET) from Wikipedia, http://
en.wikipedia.org/wiki/Metabolic_equivalent
For further reading: http://bit.ly/kLTVD3
34 VOLUME 1 / SPRING 2011
TEC: A TRAINING GROUND THAT BRINGS THE UNIVERSE TO OUR UNIVERSITY
At UCLA, the Technical Entrepreneurial Community
(TEC) is a group operating since 2005 with just that
aim. TEC actively fosters a culture of entrepreneur-
ship at UCLA—answering the question, “What will
I innovate or do with what I have learned?” TEC
believes that a business entrepreneur can come
from any background or direction. Its activities
help create environments where successful alumni
share with Bruins to discuss, educate, network, and
start companies.
How would you like to find out if you should start a
business with your degree?
Because TEC is open to all UCLA students, faculty,
alumni, and staff and is free of charge, its events
help connect the dots for people in all disciplines,
whether undergraduates, graduates, or post-docs.
TEC currently offers eight programs for the UCLA
community. Every year since its inception, the
number of programs and attendees has grown.
The organization takes part in several entrepre-
neurial programs and services as follows:
ITA Program
Participants work with the Institute for
Technology Advancement to help develop
their startup ideas.
Technology Assessment Group
Participants learn the process to evaluate
early-stage technology for commercial
feasibility through a seminar series, re-
port writing, and delivering a presenta-
tion on a UCLA patent. Participants who
complete the program obtain a signed
certificate from the Dean of the Engineer-
ing school and from the Vice Provost.
Tech Coast Angels Mentorship (TCA)
Startups receive guidance from angel in-
vestors.
Cross-Campus Entrepreneurship Mixer
A mixer is arranged where engineering
Tanuj Thapliyal, Electrical Engineering
LIFE AT COLLEGE LETS YOU LEARN ABOUT ONE WORLD. LIFE AFTER EARNING YOUR DEGREE REQUIRES AN ENORMOUS READJUSTMENT. IS THERE A WAY TO BRIDGE THE TRANSITION BETWEEN THE TWO?
35BRUIN INNOVATION & TECHNOLOGY
students can network with business
school students. 93 students attended the
inaugural mixer in February.
Dinner with an Entrepreneur
Participants have a private dinner with a
successful alumnus.
In-House Legal Counsel
Participants obtain free legal counsel on
intellectual property.
Seminars
Successful entrepreneurs, investors, and
lawyers speak on a variety of issues.
TCA Screening Sessions
Participants are a “fly on the wall” as they
hear entrepreneurs pitch business plans
to investors.
TEC is entirely student-run and set up to incorpo-
rate new programs with people’s interests. Students
who have ideas for entrepreneurship activities that
are not currently offered are encouraged to become
a part of TEC and submit a proposal. Working in
collaboration with other members of TEC will defi-
nitely foster one’s own development of networking,
communication, and leadership skills—all impera-
tive when Bruins complete their training at UCLA
and begin their transition into industry.
WORKING IN COLLABORATION
WITH OTHER MEMBERS OF TEC
WILL DEFINITELY FOSTER ONE’S
OWN DEVELOPMENT OF NET-
WORKING, COMMUNICATION,
AND LEADERSHIP SKILLS—
ALL IMPERATIVE WHEN BRUINS
COMPLETE THEIR TRAINING AT
UCLA AND BEGIN THEIR TRAN-
SITION INTO INDUSTRY.
HTTP://WWW.TECBRUINS.ORG
36 VOLUME 1 / SPRING 2011
FIGURE 1
MAZDA’S CROSS-SECTIONAL ROTARY ENGINE
DISPLAYED AT THE MAZDA RACEWAY IN MONTEREY, CA.
37BRUIN INNOVATION & TECHNOLOGY
After seeing their first working prototype,
Americans Glenn Curtiss and the Wright broth-
ers recognized the potential of such a small, light-
weight, and powerful engine; they immediately
bought the North American rights and built their
own. Incredibly, the Wankel-type rotary engine
was about half the size, half the weight, half the
manufacturing cost, and had fewer than half the
moving parts, for the same horsepower as the well-
established piston engine. Because of the intrinsic
rotational motion, the rotary is a quieter engine, is
more compact, and has less vibration while being
higher revving. By 1975, nearly every major auto-
motive company in the world was anticipating the
eclipse of the reciprocating piston engine and ei-
ther held a license, or was in negotiations, to build
Wankel’s rotary engine [1].
Fifty years have passed since the debut of the
Wankel engine. NSU was eventually absorbed by
Volkswagen then coalesced into Audi, and the Wan-
kel disappeared from the auto manufacturing land-
INNOVATIVE DESIGN METHOD CREATES NEW VARIETIES OF ROTARY ENGINES
Sarah Warren, Mechanical & Aerospace Engineering
scape with the advent of the Iranian revolution and
the subsequent 1979 crude oil shortage. The rela-
tively new engine had been under limited develop-
ment for about ten years and had not yet achieved
the fuel efficiency that it commands today; it was
not even competitive with its centenarian rival, the
piston engine. This single deficiency nearly erased
the Wankel from automotive history. Now the only
manufacturer building a production rotary engine
is Mazda, for their RX-8 sports coupe.
Even today Wankel’s rotary engine does not
have the fuel efficiency of a comparable piston en-
gine, inhibiting any economical applications and in-
stead reserving it for performance purposes. Mazda
stunned the world in 1991 by winning the 24 Hours
of Le Mans with their rotary race car, prompting an
immediate rule change to make rotary engines in-
eligible against piston engines due to their inherent
and insurmountable advantage. Even so, the light-
weight powerhouse has proved itself repeatedly in
racing competition since then. The Wankel’s lower
THE ONLY ROTARY ENGINE DESIGN THAT HAS ENJOYED ANY MEASURE OF POPULARITY AND SUCCESS IS ATTRIBUTED TO ENGINEER FELIX WANKEL AND THE GERMAN AUTOMOTIVE COMPANY NSU.
38 VOLUME 1 / SPRING 2011
fuel efficiency may be attributed to a few inher-
ent features, including the combustion chamber
and the apex seals. But optimizing the combustion
chamber and engine sealing—in fact, changing any
part of a Wankel engine profile—is not so easy.
One other remarkable and unique character-
istic of the rotary engine is its fuel versatility. The
engine design itself is inherently octane-flexible
and easily adaptable to alternative fuels. Mazda
has demonstrated this with gasoline and hydrogen
hybrid vehicles, designing a duel-fuel rotary engine
that can alternate between the two fuels at the flip
of a switch. It is conceivable that the rotary engine,
with improvements, could reduce fossil-fuel de-
pendency while allowing the defense industry, rac-
ing industry, and vehicle enthusiasts to indulge in
performance.
The scope of this research project includes
applying a relatively new method of conjugate pair
generation in order to design optimal rotary engine
alternatives to the Wankel. This method is called
the deviation-function (DF) method and was origi-
nally developed by Yang, Tong, and Lin and applied
to lobe pumps [9].
WANKEL’S ROTARY ENGINE The photograph in Figure 1 shows a cross-sec-
tion of Mazda’s Wankel engine. The triangular rotor
rotates eccentrically inside the oval-shaped engine
housing. The engine’s crankshaft is at the center of
the housing and rotates three times for every full
revolution of the rotor. During its orbit, the rotor is
always and only in contact with the housing at the
three apexes, never contacting the housing along
its flanks. This creates three chambers between the
rotor flanks and the housing, effectively sealed and
thus thermodynamically isolated from each other.
The chambers are constantly changing shape as the
rotor turns, allowing the four strokes of the Otto
cycle to occur sequentially for each chamber and
simultaneously on all sides of the rotor, similar to
a multi-piston internal combustion engine. The in-
take and exhaust breathing is through ports in the
housing, eliminating the need for valves and a valve
train. This is one reason why the rotary engine has
so many fewer parts than the reciprocating engine.
Thus the rotor and housing are analogous to a pis-
ton and cylinder, but one clear advantage of the
Wankel is the frequency of the power stroke. For a
one-cylinder 4-stroke engine, the power stroke oc-
curs once for two rotations of the crankshaft. For a
one-rotor Wankel engine, the power stroke occurs
on every rotation of the crankshaft, utilizing its dis-
placement twice as frequently.
Although not always credited by name, all pro-
duction rotary engines today are of Wankel’s origi-
nal design proportions, whether the application is
a sports car, compressor, model airplane, or out-
board motor. Since no other rotary engine design
has ever been in production, the name Wankel is
often thought to be synonymous with a rotary and
reciprocating combustion engine. In fact, the Wan-
kel engine has specific proportions for the 3-lobed
rotor and its conjugate housing profile. All design
variations are engendered by tweaking a single ra-
tio of rotor radius R to eccentricity e, called the K
factor, or trochoid constant.
ALL PRODUCTION ROTARY
ENGINES TODAY ARE OF
WANKEL’S ORIGINAL DESIGN
PROPORTIONS, WHETHER
THE APPLICATION IS A SPORTS
CAR, COMPRESSOR, MODEL
AIRPLANE, OR OUTBOARD
MOTOR.
39BRUIN INNOVATION & TECHNOLOGY
The housing of a Wankel engine is very spe-
cifically defined by generating a particular epitro-
choidal curve. An epitrochoid is a type of trochoid
curve, and a trochoid is generated by the path of a
single point within a circle that rolls along the pe-
riphery of another (base) circle. The base circle is
stationary, the rolling circle is called the generating
circle, and the point that draws the path is called
the generating point. The two circles are related
in size because one must have a radius that is a
whole number multiple of the other. To create an
epitrochoid curve, the generating circle rolls on the
outside of the base circle, illustrated in Figure 2(a).
For the housing of a Wankel engine, the generating
circle of radius r2 is equal to half of the base circle
radius r1. In Figure 2(a), the stationary base circle
is large, the generating circle is small, and the gen-
erating point is identified by its radius, eccentricity
e, and is marked along its epitrochoidal path. The
generating circle rolls without slipping on the base
circle, and all of its points describe trochoids, ex-
cept the center whose path is a circle. A family of
epitrochoid curves is defined by choosing different
points (excluding the circumference) on the gener-
ating circle as shown in Figure 2(b). Once an epitro-
choid is defined for a potential engine housing, the
difference between the other curves in the same
family is the eccentricity. Thus, eccentricity is one
of the characteristics used to design Wankel engine
profiles, the other being rotor radius R. Rotor ra-
dius is the distance between the center of the ro-
tor and its apex, as indicated in Figure 3(b). The
profile of the 3-lobed rotor is found by rolling the
2-lobed epitrochoid around the rotor’s eccentric
center. Examining all incremental positions of the
epitrochoid during its orbit at the same time re-
veals two 3-lobed envelopes of the overall motion,
FIGURE 2
EPITROCHOID GEOMETRY.
(A) EPITROCHOID
(B) CHANGING ECCENTRICITY
ECCENTRICITY IS ONE
OF THE CHARACTERISTICS
USED TO DESIGN WANKEL
ENGINE PROFILES; THE OTHER
BEING ROTOR RADIUS.
40 VOLUME 1 / SPRING 2011
shown in Figure 3(a). The silhouettes of the various
positions of the engine housing profile while roll-
ing around describe the conjugate inner envelope
and the conjugate outer envelope. A rotary engine’s
3-lobed rotor is the engine housing’s conjugate in-
ner envelope, correctly positioned together in Fig-
ure 3(b). The ratio of rotor radius R to eccentricity
e (the K factor) determines the engine compression
ratio, the relative volumetric displacement, and the
maximum leaning angle. There is a unique K factor
for all Wankel engine designs, so by deciding on a
single design criterion, such as maximum leaning
angle, the design of the engine profile is completed.
This method of designing a trochoidal mechanism
may recall the child’s toy Spirograph, and indeed
part of the success of Wankel’s design can be attrib-
uted to the easily understood geometry. However,
this simplistic methodology belies the profound
and powerful relations of trochoids and their con-
jugate envelopes [5] [8].
FIGURE 3
WANKEL ENGINE GEOMETRY.
(A) INNER AND OUTER ENVELOPES
(B) HOUSING AND ROTOR PROFILES
http://bit.ly/lFkRRP
41BRUIN INNOVATION & TECHNOLOGY
The main design considerations for the pro-
files of a Wankel engine are size (physical dimen-
sions), relative volumetric displacement, theoreti-
cal compression ratio, and maximum leaning angle.
The size of the profiles relative to the pitch circles
and the eccentricity are both manifest in the gear
set that positions the rotor relative to the housing.
The annular gear (for positioning, not power trans-
fer) is placed inside the rotor, and the eccentricity
of the designed profiles is sustained in the rotor’s
eccentric shaft. Theoretical compression ratio and
maximum leaning angle are indicators of horse-
power and efficiency, and both are limited by prac-
ticality. The leaning angle, also called angle of oscil-
lation or incidence, refers to the angle of contact that
the apex seals have with the housing as the rotor
orbits. The maximum leaning angle indicates how
well each chamber can be sealed from its neighbors
and thus serves as a sealing index. Manufacturers
of the Wankel engine have achieved optimal sealing
by using housings with fatter “waistlines.” This type
of chamber has lower maximum leaning angles, uti-
lizing smaller swaths of contact on their apex seals.
A higher maximum leaning angle, characteristic of
housings with high eccentricity (a sharply defined
waistline), means that the apex seals are in working
contact closer to their edges, which is detrimental
to sealing and will wear the seals out sooner [3].
DEVIATION-FUNCTION METHODIntroducing a new method to rotary engine
design demands adhering to certain industry
standards so that results of this project can be eas-
ily, fairly, and convincingly compared to the Wankel
engine. In accordance with this objective, the same
design and performance criteria of displacement,
compression ratio, leaning angle, and efficiency
are and will be calculated for DF-designed profiles.
The variety of rotary engine designs now available
for optimization is expected to generate interest
among design engineers and manufacturers. Ide-
ally, the complete development of the DF algorithm
as it applies to the rotary engine will appeal as an
accessible methodology for achieving previously
inaccessible results.
FIGURE 4
PITCH PAIR EXAMPLES.
(A) TWO GEARS WITH
THEIR PITCH CURVES
(B) NONCIRCULAR
EXTERNAL PITCH CURVES [7]
http://bit.ly/mBRETN
THE VARIETY OF ROTARY
ENGINE DESIGNS NOW
AVAILABLE FOR OPTIMIZATION
IS EXPECTED TO GENERATE
INTEREST AMONG DESIGN
ENGINEERS AND
MANUFACTURERS.
42 VOLUME 1 / SPRING 2011
The possibilities of the DF method are wide-
ranging but cannot be described without first ap-
preciating conjugate pairs. Two rigid bodies that
are in contact along their profiles and have relative
motion, such as rolling or sliding, are called conju-
gate pairs. Two gears in mesh are an example of a
conjugate pair that have combined rolling and slid-
ing contact. When a conjugate pair is in pure roll-
ing contact, then the profiles are called centrodes,
and when the centrodes each have respective fixed
rotation centers, they are referred to as pitch pro-
files, or pitch curves. In general, the relative motion
of a conjugate pair is specified by its correspond-
ing pitch pair. Figure 4(a) shows two gears in mesh
(the conjugate pair) and the corresponding pitch
circles (the pitch pair) [6].
The design of conjugate pairs involves two steps:
1. Design the conjugate pitch pair.
2. Design a generated pair based on the
pitch pair.
The pitch pair has pure rolling contact and
defines the relative motion of the meshing con-
jugate pair. In generating the conjugate pair from
the pitch pair, the geometry of the pitch profile
is altered such that the required task can be per-
formed; for example, a pitch circle is transformed
to the toothed body of a gear. The generated conju-
gate pair in mesh has sliding contact, so the relative
motion is not pure rolling, but the speed ratio be-
tween them is the same as for the parent pitch pair.
The collective drawbacks to all current methods for
conjugate pair generation are
1. Limitation in the types of pitch pairs
they can be applied to.
2. Use of noncircular pitch curves requir-
ing solutions by high-order nonlinear
equations.
3. Lack of geometrically intuitive design of
profiles of the generated pairs.
4. Inability to assign analytical properties
to the designed conjugate profiles.
FIGURE 5
EXAMPLES OF CIRCULAR AND
NON-CIRCULAR INTERNAL
PITCH CURVES.
THE INEFFICIENCY OF THE
DESIGN PROCESS IMPEDES
THE INVENTION OF NEW
GEARS, ROTORS, AND OTHER
CONJUGATE MECHANISMS.
43BRUIN INNOVATION & TECHNOLOGY
Because of these limitations, the design of
conjugate pairs is a blind trial-and-error process.
This makes it difficult to improve conjugate pair de-
sign and analyze resulting profiles. The inefficiency
of the design process impedes the invention of new
gears, rotors, and other conjugate mechanisms.
The deviation-function (DF) method was de-
veloped to address these problems; it works for all
pitch pairs, circular or noncircular, it enables geo-
metrically intuitive designing of profiles, and ana-
lytical properties can be assigned to the generated
conjugate pairs [6].
In practice, pitch curves are almost always a
pair of perfect circles, or circular solid objects, such
as cones or cylinders. Because of the limitations of
conventional conjugate pair generating methods,
only numerical methods are used to generate non-
circular pitch curves. The only known application in
industry for noncircular pitch curves is noncircular
gears. The large amount of computation required,
combined with poor understanding of their prop-
erties and how to design them, has limited their
use. The most frequent application has been for a
pair of identical elliptical gears [2]. Their potential
has been acknowledged, however, because of their
successful application in quick-return mechanisms,
speed matching on assembly lines, and stop-and-
dwell motion. Noncircular gears have been used to
improve the functionality and simplicity of mecha-
nisms, but their diversity has so far been limited
to ellipses [4]. Figure 4(b) shows a pair of external
noncircular (and nonelliptical) pitch curves that
were designed for use with the DF method as it ap-
plies to lobe pumps [7].
Internal pitch curves, like those in Figure 5,
are the type that apply to the rotary engine. The in-
novation, potential, and main objective of the work
described here is in the development of how the DF
method and noncircular pitch curves can optimize
rotary engine design. The DF method can be used
with pitch circles to design conventional Wankel
FIGURE 6
CONVENTIONAL WANKEL
ENGINE PROFILES WITH THEIR
CIRCULAR PITCH CURVES.
http://bit.ly/lVESVB
THE INNOVATION, POTENTIAL,
AND MAIN OBJECTIVE
OF THE WORK DESCRIBED
HERE IS IN THE DEVELOPMENT
OF HOW THE DF METHOD
AND NON-CIRCULAR PITCH
CURVES CAN OPTIMIZE
ROTARY ENGINE DESIGN.
44 VOLUME 1 / SPRING 2011
engine profiles, and examples of these results are
shown in Figure 6, along with their calculated K fac-
tors.
In practice, noncircular internal pitch curves
are even more unusual than noncircular external
pitch curves. The acknowledged success of non-
circular gears in certain applications suggests that
mechanisms using internal pitch pairs may also be
improved by noncircular pitch curves. The devia-
tion-function method is a new solution to an inhib-
iting problem for noncircular internal pitch curves;
up until the DF method was developed there was
no systematic way of designing, and therefore uti-
lizing, noncircular internal pitch curves.
The DF method involves first and foremost
choosing a so-called deviation function, which in-
evitably contains parameters that are adjustable
for rendering different profiles. This is the crux of
using the DF method to design for particular me-
chanical properties. Even considering one type
of deviation function, say sinusoidal or quadratic
polynomial, the possible combinations of param-
eters are literally endless, encompassing a broad
range of design solutions that are inaccessible by
conventional design methods. Because of the sheer
number of possibilities, compounded by the variety
of monotonically increasing continuous functions,
a selection method to find the best solution is nec-
essary for practicality. The algorithms for applying
the DF method to circular and noncircular pitch
rotary engine design have been developed; all that
remains for the practical application of these equa-
tions is relating them and their parameters to the
standard design criteria for Wankel engines. One of
the ultimate goals of this research is to develop a
DF-based selection method for the rotary engine.
Figure 7 shows some of the possible rotary
engine profiles resulting from a single pair of non-
circular pitch curves, using different deviation
functions. The noncircular pitch pair is drawn in
thin lines, and the rotary engine housing and trian-
gular rotor are drawn in heavy lines. Only a single
parameter in the deviation function is changed to
generate the different profiles. Each deviation func-
tion may have four or more parameters that can
be manipulated to achieve different results. Figure
FIGURE 8 (ABOVE)
DF-DESIGNED ROTARY ENGINE
PROFILES USING DIFFERENT
NON-CIRCULAR PITCHES.
http://bit.ly/kmZxA8
FIGURE 7 (BELOW)
DF-DESIGNED ROTARY ENGINE
PROFILES USING THE SAME
NONCIRCULAR PITCH.
http://bit.ly/mwzl8w
45BRUIN INNOVATION & TECHNOLOGY
8 shows examples of using different noncircular
pitch curves with the same deviation function. The
pitch curves in Figure 8 were generated using the
same function, but increasing the noncircularity
from left to right.
To demonstrate possible design improve-
ments over the conventional Wankel engine solu-
tion, the relative volumetric displacement, maxi-
mum compression ratio, and maximum leaning
angle are calculated for a range of noncircular pitch
rotary engines.
WANKEL AND DF-DESIGNED ROTARY ENGINE COMPARISON
Felix Wankel’s original rotary engine design
was undoubtedly ingenious, but unforgiving and
limited in terms of modification and experimen-
tation. The complexity and calculations required
to manipulate and design other conjugate profiles
made Wankel’s simplicity ironclad. The plot in
Figure 9 shows the three main design criteria for
a Wankel engine on a single line. As mentioned pre-
viously, by selecting any of the three criteria specifi-
cally, or by selecting the K factor, the design of the
engine profile is completed. In the conventional de-
sign process the relative displacement and theoret-
ical compression ratio are limited by a range within
which the maximum leaning angle is minimized.
The reason for this is that the maximum leaning an-
gle is probably the most important criterion since it
serves as a sealing index and has the largest impact
on engine efficiency and longevity. Additionally, the
relative displacement and theoretical compression
ratio can be adjusted by the width of the rotor after
the two-dimensional profile has been determined.
The scattered dots in Figure 9 represent de-
viation-function-designed noncircular-pitch rotary
engine profile results. Only three of eight deviation-
FIGURE 9
ROTARY ENGINE DESIGN
CRITERIA CHART.
46 VOLUME 1 / SPRING 2011
THE ROTARY ENGINES
POTENTIAL HAS NOT BEEN
REALIZED WITH ALTERNA-
TIVE FUELS; AND POSSIBLY
ITS PERFORMANCE COULD BE
IMPROVED FOR GASOLINE
AND DIESEL-POWERED
APPLICATIONS.
47BRUIN INNOVATION & TECHNOLOGY
function parameters were manipulated to generate
this plot: pitch curve noncircularity, eccentricity,
and switch angle. Switch angle represents the con-
jugating range of the rotor and the housing and thus
affects the maximum leaning angle. Besides the DF
design results that are clinging to the conventional
Wankel engine line, clusters appear as certain fami-
lies of profiles emerge with similar results. This
suggests that a more exhaustive survey and smaller
increments of manipulation will produce distinct
blobs of heretofore inaccessible rotary engine so-
lutions. Dramatic changes to the engine profile are
not necessary, however, to improve the maximum
leaning angle. Many DF design results fall close to
the conventional Wankel engine design line, but
not directly on it. So a Wankel engine of certain dis-
placement and compression ratio can be changed
slightly but improved significantly by selecting a
DF-designed rotary engine with a lower maximum
leaning angle.
CONCLUSIONS AND FUTURE WORKThe ultimate goal of this research is to develop
a method of selection in which the DF parameters
can be manipulated to control performance crite-
ria. All results so far have been obtained computa-
tionally, so the next immediate step is to fully real-
ize the equations governing the noncircular-pitch
engine housing and rotor profiles, as well as the
resulting characteristics. Specifically applying the
DF algorithm to rotary engine design requires com-
putational geometry, especially envelope theory,
to define the housing and rotor profile equations.
After all profile equations have been established in
terms of DF parameters, the relationships between
the rotor geometry and the design criteria can be
determined. Performance indices to evaluate an
engine’s power efficiency and volumetric efficiency
will be defined so that the influence of pitch curve
noncircularity, and all other design parameters, is
clear. Then the new DF method of design can be ac-
cessible and utilized in a familiar and practical way.
More than likely, the rotary engine’s potential has
not yet been realized for use with alternative fuels;
and possibly its performance could be improved for
gasoline and diesel-powered applications. Explor-
ing the alternatives now available by using the DF
method and noncircular pitch curves makes opti-
mization and innovation possible.
REFERENCES[1] Cole, David E. “The Wankel Engine.” Scientific American Vol-
ume 227 No. 2 (August 1972) 14–23.
[2] Dooner, D. B. and A. A. Seireg. The Kinematics of Gearing. John
Wiley & Sons, Inc., 1995, 56–63.
[3] Hege, John B. The Wankel Rotary Engine. Jefferson, NC: McFar-
land & Company, Inc., 2001.
[4] Litvin, Faydor L., Alfonso Fuentes-Aznar, Ignacio Gonzalez-Pe-
rez, and Kenichi Hayasaka. Noncircular Gears: Design and
Generation. Cambridge University Press, 2009.
[5] Norbye, Jan. The Wankel Engine. Philadelphia; Chilton Book
Company, 1971.
[6] Tong, Shih-Hsi. New Conjugate Design Pair - Theory and Appli-
cation. Ph.D. thesis. University of California - Los Angeles,
1998.
[7] Tong, Shih-Hsi, and Daniel C. H. Yang. “Generation of Identical
Noncircular Pitch Curves.” Transactions of the ASME Jour-
nal of Mechanical Design 120 (1998) 337–341.
[8] Yamamoto, Kenichi. Rotary Engine. Toyo Kogyo Co., Ltd., 1981.
[9] Yang, Daniel C. H., Shih-Hsi Tong, and J. Lin. “Deviation-Func-
tion Based Pitch Curve Modification for Conjugate Pair
Design.” Transactions of the ASME Journal of Mechanical
Design 121 (1999) 579–586.
To see animated demos: http://bit.ly/j3OUTK
48 VOLUME 1 / SPRING 2011
THE SOCIAL NETWORK ISEVERYWHERE IN ENGINEERING
WHAT TO READ?WITH WHOM TO WORK?WHERE TO PUBLISH?Jong Hoon Ahnn, Computer Science
SCIENTIFIC TECHNIQUES FOR ORGANIZING AND CONDUCTING ENGINEERING RESEARCH
48 VOLUME 1 / SPRING 2011
49BRUIN INNOVATION & TECHNOLOGY
In recent years, the growth of social networks
within engineering research has become wide-
spread. As no surprise, such a network is composed
of more than databases, relationships, and merely
the linking lots of information together. Research-
ers and scientists for example are commonly tied
by one or several specific types of interdepend-
ency, such as friendship, employment, mentorship,
discipleship, similarity of beliefs, and recognized
knowledge or prestige. From the relationships in
these networks, there have even arisen databases
that track academic published papers with authors:
Who has been published and is related to profes-
sors and other authors. Sifting out these patterns
among data—a kind of meta-database linking their
publishing relationships—has grown more impor-
tant as content contained in the World Wide Web
has grown. I have been fascinated with this type
of problem since 2009 and have developed some
harvesting techniques over the last two years, em-
ploying the world of academic citations mentioned
above as my model.
It was an interesting problem in the begin-
ning: my advisor in Computer Science Miodrag
Potkonjak asked me one day, “Can you imagine a
technique people could use to help them locate the
most productive, creative, and prolific new discov-
eries in a blogosphere of contributions?” To answer
FIGURE 1
A CITATION NETWORK FOR
DAVID E. CULLER AT UC BERKELEY
WHO HAS 308 PUBLICATIONS,
20,978 CITATIONS, AND 283 CO-
AUTHORS. THE CLOSER AN EDGE
LINKING A PEER TO SOMEONE,
THE MORE CITATIONS OF THE
PERSON’S WORK FROM THE PEER..
FOR EXAMPLE, THE CLOSEST CO-
AUTHOR, PHILIP ALEXANDER LEVIS
AT STANFORD, HAS CITED DAVID E.
CULLER 351 TIMES IN THE YEARS
1985–2010.
50 VOLUME 1 / SPRING 2011
that, a data set with which to model and test this
seemed already at hand: as a graduate student, my
world revolves around academic papers—learning
others’ work and contributing my own. It’s a siz-
able population.
According to Microsoft Academic Search,
there are 10,869 researchers with at least twenty
citations in highly respected publications. Taking as
an example the social network of David E. Culler at
UC Berkeley, it is interesting to note he has worked
with 308 collaborators and garnered 20,978 cita-
tions, as shown in Figure 1. The number of cita-
tions is drastically different for different authors. It
is tempting to think that co-authoring papers with
most-cited authors or the most prolific authors
would be helpful for one’s own citation records.
Furthermore, there is a question whether it is bet-
ter to publish at highly specialized, smaller confer-
ences dedicated to an emerging topic or to target
the most prestigious and established venues. In
summary, each researcher must make a number of
important decisions that impact not just her count
of citations, but also the quality of her research.
The need for building such an author citation
database has become apparent: our practical objec-
tive is to address some of the essential productiv-
ity questions for persons in engineering, including
such questions as What papers to read? Where to
publish? and With whom to collaborate? In analyz-
ing publication databases, the scientific objective
is to develop statistically sound techniques for de-
scribing what guides the actions of highly success-
ful contributors to engineering research, whom we
define for the purposes of this article as ones who
author the most cited papers. It is evident such
techniques for establishing relationships (a “Who’s
who” of activity) can be applied to any number of
other kinds of data sets: sports, foreign languages,
the spread of epidemics, patents, and many others.
What we crafted over the two years has the
following characteristics. At the core of our ap-
proach is analysis and prediction of the expected
number of citations for a paper starting at the mo-
ment of its publication. Essentially, for the purpose
of our analysis we assume that the goal of a re-
searcher is to publish highly cited papers. Of course
that is not an altogether correct assumption; how-
ever we postulate that there is a strong correlation
between the number of citations and the actual
impact of the paper. We also assume a strong cor-
relation between the authors of highly cited papers
and their impact. One additional line of reasoning is
that one can benefit from reading papers that will
be highly cited following their publication. There-
fore, our goal is to predict the number of citations
a given paper will receive over a specified period
of time, as well as the likelihood that it will receive
that number of citations, by only using data that is
available at the time of publishing. This informa-
tion includes several citation indices, the average
number of citations for papers at a given database,
and the citation data for co-authors.
In developing this analysis method, we had to
wrestle with questions and technical challenges.
We found first of all that the vast majority of papers
have very few citations. Therefore, to take that into
account, any accurate prediction algorithm for the
majority of papers must predict that a given paper
will have very few citations. The resolution be-
tween well performing algorithms is blurred due to
the distribution of citations followed by the power
ONE OF OUR OBJECTIVES
WAS TO EXTRACT A CAUSAL
RELATIONSHIP BETWEEN EACH
PREDICTION PARAMETER AND
THE NUMBER OF CITATIONS.
51BRUIN INNOVATION & TECHNOLOGY
law. We addressed this problem by employing clas-
sification instead of prediction. Another challenge
was how to select a correct classification algorithm.
There exists already a huge number of widely used
classification procedures; it is tempting to add to
that collection and develop yet another classifi-
cation technique. However, it is evident that goes
into an area of diminishing returns. Our analysis
showed that different classification algorithms best
perform on different data sets. Therefore, we de-
cided to look for better predictions elsewhere, i.e.
in identifying best predictors. Of course, identifica-
tion of essential predictors was critical to our work.
Our procedure for finding accurate prediction pa-
rameters followed two dimensions. The first was
looking for predictors that are accurate by them-
selves. The second was to identify principal com-
ponents, i.e. predictors that predict well when they
are used simultaneously in combinations. Finally
(and maybe most significantly), one of our objec-
tives was to extract a causal relationship between
each prediction parameter and the number of cita-
tions. A firm grasp of this knowledge may in fact be
more important than the actual prediction of the
number of citations for a given paper.
Put all together, we generated prediction re-
sults based on a generic framework that we de-
veloped. We call this framework CiteCast. It is fed
with publication network data sets as inputs, and
it terminates when the prediction accuracy cannot
be improved. During the process, it selects one ma-
chine learning (ML) algorithm and its parameters
to obtain the best prediction results as outputs. In
our case, for a given paper, the goal was to predict
citation counts one year following publication. We
used only information available at publication time
(zero citation history) as predictive features for an
individual data set, extracted from the feature ex-
traction scheme. We solved this citation prediction
problem as a multi-classification problem using a
set of thresholds, where thresholds can be obtained
from the distribution of citation counts in the data
FIGURE 2
BLOCK DIAGRAM OF OUR
CITATION PREDICTION
FRAMEWORK.
52 VOLUME 1 / SPRING 2011
set; a training set was a set of pre-classified sam-
ples. Each sample as a paper instance consisted of
a feature sample. The training set was augmented
with a class vector which represents the class to
which each sample belonged. Using a classifier
with multi-classes, we obtained citation prediction
results as to which class the predicted citations be-
longed. Importantly, we defined prediction accura-
cy to be the number of correctly classified instances
divided by the total number of instances.
For the features, we employed three terms: h-
index, g-index, and e-index, which are closely relat-
ed in that h-index is a basis of g-index and e-index.
The g-index is defined as the highest number g of
papers that together have received g2 or more ci-
tations. From this definition it is already clear that
g = h”. In contrast to the h-index, the g-index gives
more weight to highly cited papers. The e-index on
the other hand tries to represent excess citations
that the h-index ignores.
To create this data shifting technique, we
made progress only after clearly understanding
these relationships. The key technique that we
found was to select features based on a linear cor-
relation coefficient (CC) c, while a regression coef-
ficient is heavily utilized in the literature for feature
selection. The reason is that a CC-based approach
is shown to be less sensitive to the year of predic-
tion in our experiments. Let us denote fi and fj to
be an individual feature. The higher |c| we get, the
more correlation between a selected feature and a
predicting target.
When we had gotten about halfway along, I
mentioned at one point to Potkonjak, “Isn’t this
exciting?” and he gave me a knowing look. Further
work began yielding accurate results to the point
that research about published papers warranted it-
self contributing a paper and sharing the progress
we had made.
Now, where things stand today on this work is
that no matter what data sets are used, we can au-
tomatically extract a set of predictive features from
the data set. The framework builds a classifier with
multi-classes using a set of thresholds determined
from the distribution of citations, selects then one
ML algorithm to predict future citations, identifies
the top n features by one of the ranking algorithms,
and finally tunes parameters of the ML algorithm to
obtain the best results at the end. The framework
depicted in Figure 2 runs until it obtains a best-per-
forming ML algorithm with proper parameter set-
tings. Although we used features with correlation
coefficients less than 0.5, a combination of the top
n features performs well with 82.5% of accuracy in
tenth year predictions. We also applied it to the data
set from the database field, thus generating 83.5%
prediction accuracy in the same setting. Interest-
ingly, the e-index turns out to be a very important
factor in the database field, while the h-index plays
an important role in different data sets.
As an interesting application to our methods,
we applied it to the prediction problem of an au-
WE DEVELOPED CITECAST,
A FRAMEWORK THAT IS FED
WITH PUBLICATION NETWORK
DATA SETS AS INPUTS, AND IT
TERMINATES WHEN PREDIC-
TION ACCURACY CANNOT BE
IMPROVED.
c f fn f f f f
n f f n f fi j
i j i j
i i j j
( , ) =− ( )( )
( ) − ( ) ( ) − ( )∑ ∑ ∑
∑ ∑ ∑ ∑22
22
(1)
53BRUIN INNOVATION & TECHNOLOGY
thor’s future ranking, based on the same data set
used in the citation prediction problem. A co-author
graph presents an equally interesting relationship
between authors. The implication illustrated in Fig-
ure 3 is that, for an edge connected more closely
to a targeted person, the more publications the
connected person has co-authored with the target.
For instance, David E. Culler has had 308 publica-
tions in total as of 2010, and among them twenty-
three papers are with his collaborator Eric Brewer
at UC Berkeley. Such a relationship from the graph
has garnered a great deal of attention in academia,
because quantifying the quality of an individual’s
research outcomes is important to evaluate his per-
formance and make choices about supporting fu-
ture directions. The problem to solve in more detail
concerns what to measure and, more importantly,
how to evaluate the predictive nature of the quanti-
tative metrics used.
For this reason, predicting the future rank
of individual authors is important because it can
provide a powerful new method for assessing the
quality of research outcomes in advance. Faster
identification of promising individuals can draw at-
tention in the early stages of one’s career and pro-
vide potential benefits on promotion, tenure, and
new funding opportunities. We introduce RankCast
based on the same framework used in CiteCast.
A typical approach to author rank is to cal-
culate and rank it via the number of citations for
each author. A widely accepted measure, the impact
factor can be derived by the number of citations di-
FIGURE 3
A CO-AUTHOR NETWORK FOR
DAVID E. CULLER AT UC BERKE-
LEY WHO HAS 308 PUBLICATIONS
AND 283 CO-AUTHORS.
54 VOLUME 1 / SPRING 2011
PREDICTING THE FUTURE
RANK OF INDIVIDUAL AUTHORS
IS IMPORTANT BECAUSE IT
CAN PROVIDE A POWERFUL
NEW METHOD FOR ASSESSING
THE QUALITY OF RESEARCH
OUTCOMES IN ADVANCE.
55BRUIN INNOVATION & TECHNOLOGY
vided by the number of publications for a given pe-
riod of time. By contrast, we define an author rank
in three different ways using the h-index, g-index,
or e-index, respectively. Now, we can define a no-
tion of author ranking as an author rank which can
be calculated by one of the bibliometric measures.
We found that the best feature for the author
ranking prediction is the h-index, which can help
achieve over 80 percent prediction accuracy. This
implies that authors who have a higher h-index at
one point in time are more likely to be important
in the future. The results additionally show that
features obtained from authors’ social networks do
not help much in predicting author ranking.
In summary, techniques for predicting future
citations have had a fundamental difficulty ever
since Price introduced the first quantitative stud-
ies in citation networks in 1965. Many researchers
have studied patterns of citation in scientific pub-
lications. Fu and Aliferis studied three prediction
models in their state-of-the-art work. They used bi-
omedical publications within a horizon of ten years
based only on predictive information available at
the time of publication. Their work is different from
our CiteCast in that our approach is to predict fu-
ture citations with bibliometric features based on
correlation coefficients which are more reliable in
prediction than are regression coefficients used by
Fu and Aliferis. The clear benefit of our framework
is its generic applicability to data sets. As an exam-
ple given above, we successfully applied it to an
author-ranking prediction problem. Although the
approach in FutureRank is similar to our RankCast
in that it utilizes the co-authorship network and
the time of article publication to predict future cita-
tions, RankCast has clear benefits in terms of pre-
diction performance as well as in the method itself.
Likewise, no matter what data sets are used as in-
puts and what prediction algorithms are used, the
framework can be easily applied to different types
of problems generating high prediction accuracy.
REFERENCES[1] D. Feitelson, and U. Yovel, “Predictive ranking of computer sci-
entists using CiteSeer data,” Journal of Documentation, vol.
60(1), pp. 44–61, 2004.
[2] M. E. J. Newman, “Power laws, Pareto distributions and Zipf’s
law,” Contemporary Physics, vol. 46, pp. 323–351, 2005.
[3] J. E. Hirsch, “Does the h-index have predictive power?,” PNAS,
vol. 104(49), pp. 19193–19198, 2007.
[4] L. Egghe, “Theory and practice of the g-index,” Scientometrics,
vol. 69(1), pp. 131–152, 2006.
[5] C-T Zhang, “The e-Index, Complementing the h-Index for Ex-
cess Citations,” PLoS ONE, vol. 4(5), 2009.
[6] L. D. Fu, and C. F. Aliferis, “Using content-based and bibliomet-
ric features for machine learning models to predict citation
counts in the biomedical literature,” Scientometrics, vol.
85(1), pp. 257–270, 2010.
[7] M. Hall, E. Frank, G. Holmes, B. Pfahringer, P. Reutemann, and
I. H. Witten, “The WEKA Data Mining Software: An Update;
SIGKDD Explorations,” vol. 11(1), pp. 10–18, 2009.
[8] C. Lokker, and K. A. McKibbon, “Prediction of citation counts
for clinical articles at two years using data available within
three weeks of publication: Retrospective cohort study,”
BMJ, 2008.
[9] N. L. Geller, J. S. de Cani, and R. E. Davis, “Life time-citation
rates: a mathematical model to compare scientists’ work,”
Journal of the American Society for Information Science,
vol. 32(1), pp. 3–15, 1981.
[10] R. Van Noorden, “The Trials of New Carbon,” Nature , vol. 469,
pp. 14–16, 2011.
[11] Microsoft Academic Search. http://academic.research.micro-
soft.com.
For further reading: http://bit.ly/jDa7Te(accepted for publication in the International Conference on Microelectronic Systems Education (MSE), June, 2011)
56 VOLUME 1 / SPRING 2011
WELL...
COURTESY OF TUMBLR:
http://tumblr.com/xjgsp2v2
What the customer described.
What got budgeted.What the
engineer designed.
How manufacturing installed it.
What marketing advertised.
What got documented.
What the customer finally received.
What the customer was billed for.
What the customer actually wanted.
57BRUIN INNOVATION & TECHNOLOGY
Here at the UCLA Henry Samueli School of Engineering and Applied Sci-
ence, we are committed to providing a world-class education to our remarkable
students, so that they may one day become productive engineers and leaders
capable of addressing society’s most pressing needs and continue our tradition
of engineering excellence.
At the same time, our distinguished UCLA Engineering faculty are lead-
ing efforts to realize innovative and potentially paradigm-shifting concepts that
will address some of the grand challenges of the 21st century. These include re-
search in areas that could transform people’s lives including renewable energy
and energy efficiency; clean water; personalized healthcare; wireless network-
ing, sustainability and cyber security, just to name a few.
Together, the school’s exceptional faculty, students, and alumni have
helped place UCLA Engineering amongst the best engineering schools in the
country.
Recent rankings by the National Research Council of graduate engineer-
ing programs placed three UCLA Engineering programs in the top 10, and seven
in the top 20. This indicates the excellence in breadth and depth throughout the
school and in our seven academic departments.
In addition, our Ph.D. production per faculty member continues to be
amongst the very highest in the nation with 164 doctorate degrees awarded in
2010. We continue to enroll some of the brightest undergraduates in the coun-
try, receiving more than 11,400 applications for the fall 2011 freshman class.
The average weighted GPA of those admitted were 4.36, and their average SAT
score was 2142 out of 2400.
Of course, these numbers are just general indicators of the contribu-
tions of many individuals and their accomplishments. This inaugural issue of
the student-produced Bruin Innovation & Technology magazine spotlights just
a few of our talented UCLA Engineering students and their work. I hope read-
ers, especially those throughout the UCLA community, will come away with a
sense of shared pride in the work these engineering students are doing, and be
inspired by them as well.
One final note, congratulations to all those associated with this new mag-
azine for their foresight and perseverance in bringing it to fruition. I wish this
new publication great success and look forward to it becoming a tradition at
UCLA Engineering.
Sincerely,
Vijay K. DhirDean
65 Years of Driving Innovation