|
|
|
|
|
|
|
MERIT Fair 2004
|
|
|
|
|
Overview, Project Abstracts, & Presentation Posters |
|
|
|
|
Friday, August 13, 2004 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MERIT 2004 Student Participants
|
|
|
|
|
The MERIT Fair is held
at the end of the summer to showcase the research conducted by the
undergraduate students during the summer. This forum provides them
with the opportunity to present their work to the wider public.
Their efforts are subjected to scrutiny from a panel of judges selected
from academia, industry, and government laboratories. Winners are
selected in the two main areas of technical focus: INSPIRE/PEER
and RITE. The prize winning projects are listed just below with
a complete listing of all MERIT 2004 projects after that. |
|
|
|
|
|
|
|
|
|
MERIT
Fair 2004 Prize Winning Projects |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
INdustry Sponsored Projects In Research Engineering (INSPIRE) |
|
|
|
|
|
|
|
|
|
Power
and Energy Electronics Research (PEER) |
|
|
|
|
|
|
|
|
|
Research Internships in Telecommunications Engineering (RITE) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
INSPIRE Project |
|
|
|
|
|
|
|
|
|
Fabrication and Characterization of Visible Blind UV Photodectors and Device Protection Using Radiation Hard Windows
Richard Graham III (Rose-Hulman Institute of Technology)
Sathyaprasad Ramaseshan (University of Maryland at College Park)
Faculty Mentor: Prof. R.D. Vispute (University of Maryland at College Park)
The goal of our project is to fabricate MgZnO film based fully functional UV detector. MgZnO thin films have been grown on sapphire substrate using pulse laser deposition. The UV spectroscopy and X Ray diffraction have been performed on the resultant film. The bandgap of the film has been calculated from UV absorption measurements. Metal-semiconductor-metal (MSM) interdigital contacts have been fabricated by standard photolithography process and the thermal evaporator will be used to deposit Chrome and Gold. The devices have been characterized for I-V and UV-optical response. We also propose to develop an efficient way to package the material with radiation hard windows, after which reliability of the device shall be determined.
Poster Presentation
|
|
|
|
|
|
|
|
|
|
PEER Projects |
|
|
|
|
|
|
|
|
|
Characterization
of Interface Traps in SiC Semiconductor Devices
Michael Bolen (Michigan
Technological University)
Faculty Mentor: Dr. Aivars Lelis
(Army Reasearch Laboratory) and Dan Habersat
(Army Research Laboratory)
Silicon carbide (SiC) semiconductor devices have the potential
to revolutionize high power and high temperature electronics because
of material advantages vis-à-vis silicon. These advantages include
a much wider bandgap for high-temperature operation, a greater
critical electric-field value for high-voltage operation, and
a greater thermal conductivity for high-power operation. However,
SiC metal oxide semiconductor field effect transistors (MOSFETs)
have much greater charge trapping at the silicon dioxide (SiO2)
insulating gate/SiC interface than there is in the Si/SiO2 system.
These charges cause variations in the threshold voltage, lower
the effective inversion-channel mobility, and create concerns
regarding the integrity of the insulating film. To quantify these
effects, we characterized MOS test structures for interface-trap
density versus energy in the band gap and leakage current through
the insulating gate-oxide film. Interface-trap characterization
techniques included C-V and conductance measurements on MOS capacitors,
and charge pumping on fully processed MOSFETs.
Poster
Presentation
|
|
|
|
|
High Temperature Dielectrics for Wide Band Gap Power Devices
Joseph Hernandez (Virginia Tech)
Faculty Mentors: Prof. R. D. Vispute
(UMCP), Dr. Shiva Hullavard
(UMCP), and Aivars Lelis (Army Research Laboratory)
The goal of our project is to develop dielectrics for wide band
gap SiC devices. The dielectric yttrium oxide (Y2O3) was deposited
on sapphire, silicone, and silicone carbide substrates. The UV
spectroscopy and X-Ray diffraction have been performed on the
films on the substrates for different sputtering conditions in
order to optimize the dielectric films being sputtered. A Metal-Insulator-Semiconductor
(MIS) structure has been fabricated by standard photolithography
and metallization process in order to study the dielectric properties.
Finally, electrical characterization will be done to measure leakage
current as a function of temperature to demonstrate usefulness
of newly grown dielectric layers for high-temperature semiconductor
devices.
|
|
|
|
|
Electro-Mechanical Analysis of AlN Micro-Mechanical Beam Resonators
Christopher Lombardo(University of Maryland, College Park)
Faculty Mentors: Prof. R. D. Vispute
(UMCP), Dr. Alma Wickenden (Army Research Laboratory),
Jennifer Grim (Army Research Laboratory), and
Dr. Shiva Hullavarad (UMCP)
AlN-based piezoelectric microelectromechanical (MEMS) resonators
have been observed to exhibit an extremely high mechanical quality
factor, but have shown poor electrical performance, indicating
significant insertion loss. The electrical and mechanical loss
mechanisms were researched, and found to include beam motional
resistance, parasitic losses due to device design, and losses
present within the component materials. An electromechanical model
was used to determine the extent of these losses and recommend
improvements in resonator design as well as future loss analysis
approaches. MEMS resonators and test devices were fabricated using
sputtered AlN layers, to compare material losses relative to devices
previously fabricated using pulsed laser deposited AlN.
Poster Presentation
|
|
|
|
|
High Temperature Capping Layers for SIC Based Devices
Pamela Lee (Columbia University)
and Eri Nishiyama (University of
San Diego)
Faculty Mentors:
Dr. Ken Jones (ARL),
Prof. R. D. Vispute(UMCP),
Dr. Matthew Ervin (ARL),
Dr. Pankaj Shah (ARL),
Michael Derenge (ARL),
Dr. Shiva Hullavarad(UMCP), and
Samantha Lei (ARL)
Silicon Carbide (SiC) is a wide band-gap semiconductor appropriate
for high power and high temperature devices. Annealing the devices
to high temperatures enable implanted atoms to diffuse to their
equilibrium sites. This requires the use of a cap to limit the
surface damage due to silicon evaporation, which damages the device
structure. Experiments in the past have revealed the prevention
of the preferential evaporation through the use of BN/AlN annealing
caps, but sometimes particulates form on the BN film that lead
to reduction of its effectiveness. In addition, stacking faults
sometimes arise in the implanted SiC during the activation anneal.
Tantalum Carbide (TaC) and Tungsten Carbide (WC) deposited using
pulsed laser deposition (PLD) have drawn interest as caps because
they are stable at high temperatures and do not react with the
SiC. In addition, they offer the future possibility of providing
a mask for selective area growth that could result in replacing
the more defective ion implanted material with material that is
grown epitaxially. The morphology of the TaC, WC, and SiC surfaces
were studied using the scanning electron microscope (SEM); The
crystalline structure of the TaC and WC films and the implanted
SiC were examined using x-ray diffraction (XRD), Rutherford backscattering
spectroscopy (RBS), and transmission electron microscopy (TEM);
and the electrical properties of the SiC diodes were studied using
electron beam induced current (EBIC) measurements.
Poster Presentation
|
|
|
|
|
|
|
|
|
|
RITE Projects |
|
|
|
|
|
|
|
|
|
Testing and Refinement of a Low-Power, Lightweight Neural Amplifier and Spike Sorter
Kevin Boyle (University of Notre Dame)
Dorielle Tucker (Clark Atlanta University)
Faculty Mentors: Pamela Abshire (UMCP) and Timothy Horiuchi (UMCP)
To understand the complex dynamics of a brain, it is necessary
to retrieve and analyze neural data from animals exhibiting natural
behaviors. There are many technical barriers in obtaining neural
signals from a freely-moving animal, including the need to restrain
the subject. The purpose of our project is to simplify data acquisition
by testing and refining a lightweight, low-power integrated circuit
containing a neural amplifier and spike sorter. This summer we
analyzed the circuit and adjusted the biasing to obtain proper
functionality. In addition, we refined methods for obtaining neural
signals from live blowflies and tested the chip with live signals.
Developing this integrated circuit is an important step towards
enabling the wireless retrieval of data simultaneously from large
populations of spiking neurons.
Poster Presentation
|
|
|
|
|
Packaging Bio-amplifiers to Monitor Extra-cellular Activity
Tony Li (University of Maryland at College Park)
Faculty Mentor: Prof. Pamela Abshire
(UMCP)
Integrated bio-amplifiers are able to directly measure extra-cellular
voltage signals from biological cells and amplify those weak signals.
A major technical barrier to obtaining signals directly from cells
is the contact between electrodes on the chip surface and biological
samples, including cells and aqueous growth medium. The technical
challenge is to ensure that cells are not harmed by non-biocompatible
materials in the chip package, and that the chip package is isolated
from corrosive materials in the cellular environment. A packaging
assembly was designed, built, and tested for making injected silicone
molds to act as an isolation barrier. In addition, a biocompatibility
study was performed to determine the biocompatibility of various
silicone materials with bovine aortic smooth muscle cells in order
to find a suitable mold material. Further, a graphical user interface
and data acquisition program was designed and coded in Matlab
for the acquisition of the voltage signals from the cells through
the bio-amplifier.
Poster Presentation
|
|
|
|
|
Wavelet Analysis of Magnetoencephalographic Auditory Activity
Chris Rodgers (Purdue University)
Faculty Mentors: Prof Jonathan Z. Simon (UMCP) and Dr. Yadong Wang (UMCP)
The technique of magnetoencephalography (MEG) holds exciting promise
for the study of electrical impulses in the brain. Based on the
detection of minute magnetic fields caused by those impulses,
MEG measurements offer high temporal resolution, well-suited for
understanding how the brain processes stimuli. Tones evoke neural
responses in the gamma band (around 35 Hz) which are phase-locked
to the stimulus and are therefore easy to detect by averaging.
Previous work has shown that an `induced' gamma response, which
is not phase-locked to the stimulus, may also occur. Because of
the inconsistent phase and because of the high noise associated
with MEG data, the induced response can be hard to detect, but
it is important because it may be an indication of higher-level
brain activity. In this project we will use wavelets to analyze
MEG data, which was collected using modulated short tones as stimuli,
and attempt to detect an induced gamma response.
Poster Presentation
|
|
|
|
|
Cell Clinics: Design of an Integrated Potentiostat and Testing of Capacitance Sensors
Joel Van Sickel (Grove City College)
Faculty Mentor: Prof. Pamela Abshire
(University of Maryland, College Park)
Cell clinics are biolabs-on-a-chip for studying individual cells
comprising lidded microvials in which the cells are directly interfaced
with bio-amplifiers and other sensors. Isolation of cells requires
the integration of CMOS electronics with MEMS actuators to control
lids on microvials. CMOS potentiostats will be designed to enable
control of the MEMS actuators and allow them to be integrated
directly into the substrate on which they are fabricated. Lids
will be opened and closed on individual cells by controlling the
electrochemical potential with the potentiostats. Previously fabricated
capacitance sensors will also be tested. Anticipated challenges
include positioning an electrode with nanometer accuracy and using
an electrode that is electrically floating instead of grounded.
Poster Presentation
|
|
|
|
|
Human Movement Categorization Through MOCAP Data and Body Angles Analysis
John Chaparro (Texas Tech University)
Faculty Mentor: Prof. Rama Chellappa (University of Maryland, College Park
Though the use of MOCAP data, we will be investigating the detection
and categorization of human movement to eventually classify abnormalities
in human gait. Periodic movement detection is achieved though
the use of similarity matrices. Through this detection and from
parameters extracted from the MOCAP data, movement will be classified
into four different categories: 1. Walking, jogging: 2. sitting:
3. sweeping: 4. crawling. Once preliminary classification is complete,
more categories will be added to allow movement abnormalities
to be detected and categorized.
Poster Presentation
|
|
|
|
|
Normalization of Facial Expressions
Divya Jhalani (University of Maryland at College Park)
Faculty Mentor: Prof. Rama Chellappa
(University of Maryland at College Park)
Robust face recognition involves many issues such as pose, illumination,
aging and expressions. To improve the performance of the existing
recognition algorithms, and make them invariant towards the issues
mentioned above, the effects of all these factors need to be compensated
for as much as possible. In this work, the focus is on making
the face recognition algorithms insensitive to the facial expressions
of a person, particularly the expressions where the mouth is involved
as the mouth moves the most in the face. Therefore, to make a
face expression-free, or in other words neutral, we change the
mouth area to get a face which is closer to being a neutral one.
This is achieved by using a triangular mesh that initially represents
a generic face. This mesh is modified to make it as accurate a
representation as possible of a face with expression by changing
its various parameters. Once the mesh is created, it can be easily
used to go back to the neutral face. This neutral face can then
be used instead of the original face in the face recognition algorithms
to yield better results.
|
|
|
|
|
Automatic Speech Recognition: Identifying Nasalization in Vowels
Shuo Chen (Virginia Tech)
Faculty Mentor: Prof. Carol Espy-Wilson
(University of Maryland at College Park)
The Speech Communication Lab (SCL) is pursuing a radically new
approach to automatic speech recognition that centers on the study
of acoustic-phonetics and its usage in capturing phonetic features
from speech waveforms. In this project, the SCL is investigating
acoustic parameters for the phonetic feature nasal, particularly
when nasals are co-articulated with preceding vowels. This co-articulation
is called "nasalization". The goal of this project is to develop
acoustic parameters, specific to nasalization, that can be automatically
and effectively extracted from a continuous speech signal.
Poster Presentation
|
|
|
|
|
Investigation of Acoustic Features in Text-Independent Speaker Verification
Thomas Plummer (University of Miami)
Faculty Mentor: Prof. Carol Espy-Wilson, (University of Maryland at College Park)
Text-independent Speaker Verification has received considerable
attention due to its extensive application in security. However,
the current front-end of the speaker verification system is affected
by the acoustical channel and various environments. This sensitivity
motivates the present research on the processing of the speech
derived acoustical features. In particular, the proposed research
explores the application of source information in speaker verification
and the processing of source information with conventional Mel-
Frequency Cepstral Coefficients (MFCC). Principal Component Analysis
(PCA) will also be implemented to reduce data dimensions by extracting
the top few principal component feature sets from the MFCC processed
source information, which will in-turn reduce data and calculation
overhead while improving system accuracy.
Poster Presentation
|
|
|
|
|
Acoustic Parameter Based Speech Recognition in Noisy Environment
Qin Zou (University of Maryland at College Park
Faculty Mentor: Prof. Carol Espy-Wilson
(University of Maryland at College Park)
One of the main reasons current automatic speech recognizers perform
poorly in practical applications is the mismatch in the training
and the testing environments. We have developed a set of Acoustic
Parameters (APs) that capture the salient features of various
broad classes of speech. The aim of this project is to study the
effect of different types of background noise on these APs and
to develop compensation algorithms that minimize the distortion
of these APs in order to improve the accuracy of the recognition
system.
Poster Presentation
|
|
|
|
|
Multi-Echo Integration for Sonar-based Vehicle Naviagation
Marshall Miller (University of Maryland at College Park)
Gaurav Singal (Columbia University)
Faculty Mentor: Prof. Timothy Horiuchi (University of Maryland at College Park)
Based on principles of bat echolocation, we have designed a robotic
vehicle capable of navigating through a field of obstacles using
sonar. The robot, equipped with a speaker and two microphones,
emulates the bat's mouth and ears. We developed an algorithm to
convert sonar echoes into positional information about the obstacles
in the field and implemented it on a PIC microcontroller. We used
a second PIC to implement a motion-control algorithm, "Openspace",
which combines obstacle information with target direction to choose
the best heading. This heading is converted into motor commands
and sent to the robot, which executes the command.
Poster Presentation
|
|
|
|
|
GPS-Based Location Determination
Kun Lin (University of Maryland at
College Park)
Faculty Mentor: Dr. P. S. Krishnaprasad
(University of Maryland at College Park)
Systems that utilize feedback are called closed-loop control systems.
Such systems continuously adjust their control signals according
to the received feedback signals (e.g. cruise control). Currently,
we are constructing a closed-loop control system modeled after
the classical pursuit problem. The system uses two robots, programmed
using Motion Description Language Extended, named Fox and Rabbit.
Fox will attempt to intercept course with Rabbit using both its
GPS (Global Positioning System) location data and that of the
Rabbit's. Challenges encountered are robot hardware configuration
and GPS data accuracy. Although, technical difficulties still
persist, we remain optimistic.
Poster
Presentation
|
|
|
|
|
Ad Hoc Network Security
Wui Cheong Wong (Binghampton University)
Faculty Mentor: Prof. Ray Liu (UMCP)
A mobile ad hoc network is a group of mobile nodes that do not
require centralized administration, in which nodes can communicate
with each other even when they are not in direct communication
ranges. However, before ad hoc network can be successfully deployed,
the security issues must be resolved first. In this project, we
investigate the secure routing protocols for mobile ad hoc networks,
and implement a testbed using C++.
Poster Presentation
|
|
|
|
|
Adjustable Time Delays for Optical Clock Recovery Systems
Amir Ali Ahmadi (University of Maryland at College Park)
Elric Von Eden (University of Maryland at College Park)
Faculty Mentor: Prof. Tom Murphy
(University of Maryland at College Park)
We built an all-optical clock recovery system based on two-photon
absorption (TPA) in a silicon avalanche photodiode. The system
is polarization insensitive, compact, broadband, and capable of
supporting future high-speed optical networks. One key element
in this new system is an adjustable time delay that allows the
clock timing to be modulated. This was accomplished by using a
phase modulator and polarization maintaining fiber (PMF) to periodically
dither the timing of the clock signal. Finally, a phase detector
circuit followed by a phase-locked loop (PLL) was implemented
to synchronize the clock and data signals.
Poster
Presentation
|
|
|
|
|
Evaluation of Polarization Dependence in Nonlinear Optical Detectors
Audrey Strunc (Washington University)
Faculty Mentor: Prof. Tom Murphy (University of Maryland, College Park)
This project investigates the nonlinear process of two-photon
absorption in the photomultiplier tube, a nonlinear optical detector.
We theoretically predict and experimentally measure the polarization
dependence in the photomultiplier tube. In addition to this, we
examine the photomultiplier's sensitivity to the input continuous
wavelength light. We also measure how the output signal changes
as a function of the detector position.
Poster Presentation
|
|
|
|
|
An Examination of Security Algorithm Flaws in Wireless Networks
Hirsh Goldberg (University of Maryland at College Park)
Erica Simcoe (Pennsylvania State University)
Faculty Mentor: Prof. Sennur Ulukus (University of Maryland at College Park)
The area of interest in this research is Wireless Local Area Networks
(WLANs) specifically based on the IEEE 802.11 protocol. Although
there are numerous measures in place that have been designed specifically
to protect sensitive user information, research has shown that
there are significant flaws in current security mechanisms implemented
in this standard. Emphasis is placed on examining these security
flaws and understanding what measures have been taken to eliminate
them. Finally, several simulations of an attack on a WLAN using
the IEEE 802.11 standard are studied in order to prove the overall
insecurity of this standard.
Poster
Presentation
|
|
|
|
|
Improving the Robustness and Imperceptibility of Curve Fingerprinting
Daniel Cardy (Washington University)
Faculty Mentor: Prof. Min Wu (University of Maryland, College Park)
Due to the proliferation of scanning and pen-based devices, images
consisting primarily of curves, such as maps, drawings and signatures,
are becoming increasingly common in the digital domain. These
digital documents can be easily stored and stolen, used and misused.
Consequently, it is important to be able to trace the unauthorized
distribution and malicious tampering of such documents. This can
be accomplished by hiding information in them, where the hidden
data provides tell-tale signs, for example, from whom a classified
document is leaked. B-splines, which are piecewise polynomial
functions, can be used to approximate a curve from a small set
of characterizing data called control points. Information can
be embedded in the curve by making slight modifications to these
control points. My algorithm optimizes the placement of control
points to facilitate robust and invisible data embedding. Instead
of simple uniform placement, fewer control points are placed in
linear areas and more in areas with large curvature and drastic
changes. With my algorithm, the overall changes introduced by
the embedding step can be substantially reduced. It allows more
information to be embedded in the curve, or the same amount of
information be embedded in a more robust way.
Poster Presentation
|
|
|
|
|
|
|
|