Graduate Research - Undergraduate Research - Rice @ Affiliates
Thursday, September 13, 2007
1:00pm - Martel Hall, Duncan Hall
Demo: Multi-hop MIMO Network
Student Leads: Patrick Murphy and Chris Hunter
Faculty: Ashutosh Sabharwal, Behnaam Aazhang, Joe Cavallaro, Ed
Knightly
Abstract: In this demo, we will showcase a clean-slate
design of multi-hop MIMO network using WARP nodes. The key
ingredients include adaptive MIMO Physical layer (Alamouti coding
and/or spatial multiplexing), opportunistic MAC protocols and packet
forwarding to avoid self-contention. The demo serves as the first
major step toward a deployed clean-slate backhaul link for the Rice-TFA
network.
Posters:
- Real-time low-complexity soft-output MIMO detectors - (pdf)
Kiarash Amiri, Predrag Radosavljevic, Joseph Cavallaro
- Architectures for LDPC-coded Relay Networks - (pdf)
Marjan Karkooti, Yang Sun, Joseph R. Cavallaro
- Finite Alphabet Beamforming Codebooks in MIMO-OFDM
Systems - (pdf)
Melissa Duarte and Ashutosh Sabharwal
Demo: On the Hardware Implementation of Analog to
Information Systems
Student Lead: Tamer Ragheb
Faculty: Yehia Massoud, Richard Baraniuk
Posters:
- Efficient Modeling and Implementation Techniques of Analog
to Information systems
- Design of Fully Integrated Analog and Mixed-Signal
Systems
- Interconnect-Centric Network-on-Chip Analysis and
Thermally-Aware Design Methodologies
- Modeling of Carbon Nanotube-Based and Nanophotonic-Based
Interconnects for Future VLSI Applications
- Analog-to-Information Conversion Theory - (pdf)
Demo:
Micro Power Management of High-Performance Wireless
Interfaces - (
pdf)
Student Lead: Jiayang Liu
Faculty: Lin Zhong
Abstract: We have developed a novel power management method,
called micro power management (micron-PM), to reduce the power
consumption of busy-time in 802.11 network interfaces. micron-PM has
the following features:
- micron-PM reduces energy consumption in network busy time without
sacrificing communication quality.
- micron-PM takes advantage of multiple sleep modes on wireless
network interface and dynamically selects the optimal one based on
traffic characteristic.
- micron-PM does not need any modification to the network
infrastructure.
We have developed a prototype of micron-PM based on WARP. In the demo,
we will show the communication between two WARP boards with micron-PM
applied under various applications.
Demo: Portable Trace Chemical Sensing via Laser
Spectroscopy
Student Leads: Steve So, Daniel Chang
Faculty: Frank Tittel, Gerard Wysocki, Anatoliy Kosterev
Posters:
- Embedded High-Efficiency Mixed-Signal Systems for
Laser-Based Spectroscopic Trace Gas Sensing
- Precise Laser Temperature Controller Embedded in Miniature
Infrared Spectroscopic Trace Gas Sensor
Demo:
Multi-tier Mesh Urban Access Research Platform,
Technology For All (TFA)
Student Lead: Joe Camp
Faculty: Ed Knightly
Abstract: Rice University's Network Group has partnered with
Technology for All (TFA) to deploy a multi-tier wireless network in an
under-resourced community in Houston's East End. The project began in
late 2003 and the first node was deployed in September 2004. The
network serves over 3,000 users as of Summer 2007. The deployment
employs fully programmable network nodes and serves as a
first-of-its-kind research testbed for large-scale urban wireless
networks. In this demo, we present the TFA network environment,
visiting various antenna locations within the network and highlighting
the research being performed on the network including network
management, rate control, mobility, and routing.
Demo:
Single Pixel Camera
Faculty: Kevin Kelly, Richard Baraniuk
Poster:
- Compressive Image Processing
Author: Chinmay Hegde
Abstract: The theory of Compressive Sensing (CS) enables stable
reconstruction of a compressible signal or image from a small number of
random proections, such as the measurements obtained by the
single-pixel compressive camera. We demonstrate using new image
processing theory that the utility of the camera can be extended to
several novel applications, including pattern classification and target
recognition.
Demo:
Atomic Microscope
Faculty: Kevin Kelly
Demo: Network Server Architectures and Virtual Machines
Group: Jeff Shafer, Mike Foss, Scott Rixner, Cox
Abstract: An emerging trend in computing today is the
deployment of virtual machines, allowing the consolidation of multiple
systems onto one physical machine in the data center. We introduce a
new network architecture, Concurrent Direct Network Access (CDNA), that
substantialy accelerates network performance in these virtualized
systems.
Poster: A Unified
Framework for Non-parametric Shape Constrained Regression
- (pdf)
Authors: Mehrdad Majzoobi, Farinaz
Koushanfar
Abstract: We introduced a unified framework for
nonparametric shape constrained regression. Addition of the shape
constrraint which often accounts for the existence of a physical
phenomena or a specific behavior, improves the predictability of the
model. Furthermore, certain shape constraints, e.g. convexity,
facilitate subsequent optimizations. In this work, modeling is
accomplished by transforming the problem into combination domain;
meanwhile, shape constraints are included by bounding the cominatorial
search space. First, we address isotonic shape constraint by a
dynamic programming algorithm in combinatorial domain and explain its
dual solution in graph combinatorics domain. Next, we show how a
number of other important shape constraints including unimodal, convex,
and limited slope can be modeled within the same framework. Besides
the framework allows setting a limit on the number of model parameters
which performs in polynomial time. The beauty of combinational
framework is that it can be applied to any arbitrary user-defined
objective function. Such flexibility is harnessed to add robustness
against outliers to the model. The complexity of each algorithm is
calculated and their performance and effectiveness have been evaluated
on traces of temperature measurements from a deployed sensor
network.
Poster: Challenging Benchmark for Location
Discovery in Sensor Networks: Foundations and
Applications - (pdf)
Authors: Davood Shamsi, Farinaz
Koushanfar
Abstract: The overall goal of this project is to create
challenging and relevant benchmark data
for sensor network localization problem. While a large body of research
effort has focused on addressing the problem, the overall progress is
limited to disparate methods that cannot be integrated or compared. Our
benchmark generation method is built upon studying the complexity of
the localization problem with respect to both its graph theoretical
(discrete combinatorial) properties as well as its continuous
optimization properties. We first discuss combinatorial complexity of
network localization; we show the transition point between localizable
and non-localizable networks based on the density. We build hard
instances (i.e., instances with exponential combinatorial complexity)
of the problem by generating localization data that is on the
transition point. Next, we inspect the properties of the localization
problem from a continuous optimization point of view. Both the effect
of the optimization objective functions and the constraints are studied
and measured with respect to the data set. Specifically, we select a
number of parameters affecting the localization and introduce a few
metrics to evaluate the hardness of the optimization problem on each
dataset. Plackett and Burman (PB) simulation methodology is used for
rapid characterization of the effect of each parameter on the
localization data sets. We have used the methodology on real sensor
data to generate a suit of benchmark instances for localization
problem.
Poster:
Connexions
Presenter: Raymond S. Wagner
Abstract: Connexions (
cnx.org) is an open educational resource
which enables authors to create new instructional content, adapt other
authors' content for new purposes, and assemble texts using their own
and others' materials, free of charge. In this poster, we overview the
Connexions system and introduce two new features: print-on-demand
publication of select Connexions courses and content filtering using
Connexions Lenses.
Poster: State Space Neighborhood Structure under Image
Representation Using Magnitude and Phase Representation - (pdf)
Authors: Gang Hua, Mike Orchard
Abstract: New Image representations using magnitude and phase
signficantly improve the processing performance over traditional
orthogonal representations in applications like image denoising and
prediction. In this poster, we investigate the state space
neighborhood structure under the new representations to explain its
advantage over traditional ones. The neighborhood structure is
determined by the Jacobian of the signal to magnitude and phase
transform. First, we relate the rank of the Jacobian (i.e. the
dimension of the neighborhood tangent space) to the spectral property
of the signal. Then, the conditions on the uniqueness of the
representation of a signal can be deduced easily. Second, we examine
the distribution of the singular values of the Jacobian for typical
signals to show the advantages of the new representations.
Poster: Comparison
of model reduction methods for VLSI circuit simulation -
(pdf)
Authors: Roxana Ionutiu, Sanda
Lefteriu, and A.C. Antoulas
Abstract: We compare different model reduction methods
such as balanced truncation, truncation by balancing one gramian (or
poor man's truncated balanced reduction), positive real balanced
truncation, PRIMA (which is the industry standard), and the newly
developed spectral zero and optimal H2 methods. These methods are
applied to a simple system composed of two coupled transmission lines.
The performance of these algorithms is analyzed in terms of several
criteria such as: error magnitude, computational cost and preservation
of passivity.
Demo: Finite Difference Time Domain Simulations of Complex
Nanostructures
Student Leads: Fei Le, Feng Hao, Nathaniel Grady
Faculty: Naomi J. Halas, Peter Nordlander
Poster: Polarization-Dependent Plasmonic Coupling Between
Silver Nanowires and Nanoparticles - (
pdf)
Authors: Mark W. Knight, Nathaniel K. Grady, Rizia Bardhan, Feng Hao,
Peter Nordlander, Naomi J. Halas
Abstract: Metallic nanoparticles can serve as efficient antenna
for coupling visible light into propagating plasmons of an Ag nanowire.
For long wires, the coupling is maximal for incident light polarized
perpendicular to the nanowire. For sub-10-micron m nanowires, the
polarization corresponding to maximum emission from the ends of the
nanowire was found to be strongly dependent on the nanowire geometry
and position of the vicinal nanoparticle. This nanoparticle
antenna-based approach offers a potential strategy for optimizing
plasmon coupling into nanoscale metallic waveguides.
Poster: Plasmonic Nanoshell Arrays: Combining Enhanced
Vibrational Spectroscopies on a Single Substrate - (
pdf)
Authors: Janardan Kundu, Hui Wang, Fei Le, Carly Levin, Peter
Nordlander, and Naomi J. Halas
Abstract: We have designed and fabricated a
subwavelength-structured metallic substrate, nanoshell arrays, to
simultaneously enhance two complementary vibrational spectroscopies,
surface enhanced Raman spectroscopy (SERS) and surface enhanced
Infrared absorption spectroscopy (SEIRA), by introducing plasmon
resonances and their associated high intensity electromagnetic fields
into the precise frequency regions required for both spectroscopies.
The resulting "analytical chemistry lab-on-a-chip" provides a valuable
new tool for chemical fingerprinting and molecular
identification.
Poster: Plasmons in the metallic nanoparticle-film system: a
tunable impurity problem - (
pdf)
Authors: Fei Le, Zaw Lwin, Jennifer Steele, Naomi J. Halas, and Peter
Nordlander
Abstract: We show that the discrete, localized resonances of a
metallic nanoparticle interacting with the continuum of surface
plasmons of a metallic film is an electromagnetic analog of the
spinless Anderson-Fano model. The characteristics of this interaction
are changed by decreasing film thickness from a regime with a localized
state below the continuum, to that with resonances in the continuum, to
that with a localized state above and an associated virtual state in
the continuum. Verifications from both experiments and numerical
studies using FDTD are presented.
Poster: Plasmon resonances of a Gold nanostar - (
pdf)
Authors: Feng Hao, Colleen L. Nehl, Jason H. Hafner, and Peter
Nordlander
Abstract: Using the Finite-Difference Time-Domain method we show
that the plasmons of a nanostar result from hybridization of plasmons
of the core and tips of the nanoparticle. The nanostar core serves as a
nanoscale antenna, dramatically increasing the excitation cross section
and the electromagnetic field enhancements of the tip plasmons. Our
analysis demonstrates that the plasmon hybridization picture can be
combined with numerical approaches to interpret the physical origin of
the plasmons of highly complex nanostructures. Further numerical
experiments show the huge electromagnetic field enhancement at the tip
of the nanostar is possible for the application of detecting single
molecule.
Demo:
OrbitECG: Ambulatory Wearable ECG
Monitoring Based on the Rice Orbit Platform - (
pdf)
Group: Bryan Grandy (Schlumberger Senior Design Project Winner,
ViaSat Prize Winner), Charlie Ice (ViaSat Prize Winner), Sara Joiner
(ViaSat Prize Winner), Ahmad Rahmati, Lin Zhong
Abstract: We have developed a three-tier solution for real-time
ambulatory ECG monitoring to address its challenges regarding energy
efficiency and security, while remaining scalable, cost-effective, and
configurable. The OrbitECG is based on the Rice Orbit Platform, and
consists of:
- An Orbit Sensor, as a low-power wearable Bluetooth ECG sensor
- A mobile phone with cellular and Wi-Fi connectivity as a personal
server
- An Internet health server for data storage and access
control.
We employ Context-for-Wireless to address the energy challenge of
wireless data transfer, and provide secure multi-layer access control
for the data stored on our server. While we have developed an ECG
monitoring system, our work is general and can be applied to other
health monitoring systems as well. Furthermore, our personal server
supports multiple health sensors simultaneously.
Demo: WARP User I/O Board
Group: Barron Stone, Patrick Murphy (Xilinx Fellow, TI Fellow), Charles
Camp, Ashu Sabharwal
Abstract: We will demo a user I/O daughtercard design for WARP
FPGA boards, which provides designers with
a LCD screen, trackball and array of LEDs for displaying information
during design and operation phases of WARP boards.
INNOVATE/NanoJapan
Posters
A Study on the Nucleation of
Ferromagnetic MnSb Crystallites on a GaAs (001) Surface
Authors: Yimang Wang (NanoJapan 2007 participant); Yusuke Ogawa,
Hiro Munekata (Imaging and Science Engineering Laboratory, Tokyo
Institute of Technology, Tokyo, Japan)
Abstract: III-V compound semiconductors are key materials for
semiconductor lasers, while ferromagnetic compounds can be used for
controlling the propogation of a laser beam. Thus, both types of
materials are essential for optoelectronic applications. But,
integrating the two is difficult due to their different crystal
structures and chemical propoerties. If they can be grown together in
a single form, one can expect a better efficiency of optical
communicatins and lower production costs. Here, we study the growth
of ferromagnetic MnSb on GaAs. Knothat that nucleation is essential
in the crystal growth process and the ultimate way of crystal growth is
to control its nucleation process, we are trying to control the
nucleation of MnSb for producing high-quality ferrmagnetic thin
films. We first grow MnSb thin films on GaAs substrates using
molecular beam epitaxy under ultra-high vacuum and proper temperatures
and beam fluxes; and then utlize tools such as in-situ reflection high
energy electron diffraction (RHEED), ex-situ scanning, electron
microscopy, and ex-situ X-ray diffraction of extract information on the
forms of nucleation, crystal growth orientation, and lattice parameters
of MnSb. So far, the orientation of MnSb crystallites is determined
to be <10-11> normal to a GaAs (001) surface. Moreover,
analysis of a RHEED pattern along the [1-10] GaAs direction seems to
show a dynamic change of the lattice constant of MnSb as a function of
growth time. Further analysis and conclusions will be made by the
time of this presentation.
Reduction of Single-Walled Carbon
Nanotube Bundling Via Density Gradient Ultracentrifugation
Separation
Authors: Tianhe Yang (NanoJapan 2006 participant & Kono Research
Group Member), W.D. Rice, E.H. Haroz, Jun Kono
Abstract: Single-walled carbon nanotubes (SWNT) have the
potential to exhibit extraordinary mechanical, electrical, and thermal
properties. However, the SWNT synthesis process inherently produces
many different chiralities, which makes understanding their underlying
physical nature challenging. Furthermore, the hydrophobic nature of the
SWNTs causes individual SWNTs to aggregate together, creating nanotube
bundles, which also alters the physical response of SWNT samples.
Through the use of surfactants, such as sodium deoxycholate and sodium
cholate, SWNT bundling can be reduced by isopycnically separating the
nanotubes in a mass density gradient, a process first demonstrated by
Arnold et al. This technique takes advantage of the fact that different
SWNT chiralities have slightly different densities.
Using optical absorption spectroscopy, a noticeable trend is observed
in the curvature of the baseline when aliquots are taken at different
positions in the force gradient. Our work shows that the inherent
optical absorption baseline, which is thought to come from the presence
of bundles, can be reduced via this technique. Complementary
characterization via photoluminescence and Resonate Raman
spectroscopies agree with this observation. in addition, we have
performed experiments demonstrating that SWNT isopycnic separation is
due almost solely to kinematic considerations, in contradiction with
results of Fagan et al.
The Economic Digital Divide in Contemporary China
Author: Raj Bandyopadhyay, Department of Computer Science (INNOVATE
2007 participant)
Critical Issues India Must Tackle to Maintain Position in the
Global Market
Author: Marina Boleda, Department of Bioengineering (INNOVATE 2007
participant)
Challenges to China: Maintaining a Position in the Global
Market
Author: Stacy Cheng, Department of Bioengineering & Department of
Asian Studies (INNOVATE 2007 participant)
Rice@DoCoMo USA Labs
Poster: Spatial Sparsity Induced Temporal Prediction for Hybrid
Video Compression - (
pdf)
Authors: Gang Hua, Onur Guleryuz (DoCoMo USA Labs), Mike
Orchard
Abstract: In this poster, we propose a new motion compensated
prediction technique that enables successful predictive encoding during
fades, blended scenes, temporally decorrelated noise, and many other
temporal evolutions that force traditional predictors to fail. By
performing prediction in an aliasing-free transform domain where the
video frames are spatially sparse and exploring the continuous location
change of image features, our work enables better prediction than would
be possible otherwise. Better coding performance is achieved with
standard hybrid video codecs equipped with the proposed prediction
technique.
Rice@Nokia
Poster and Demo: Efficient Reuse Allocation for Cellular
Systems - (
pdf)
Authors: Gareth Middleton, Brett Kaufman, Behnaam Aazhang, Jorma
Lilleberg (Nokia Technology Platforms, Oulu, Finland)
Abstract: Spectrum management issues resulted in major roadblocks
for the deployment of 3G cellular radio systems. As the cellular
technology continues to evolve, similar issues will continue to setback
the deployment of 4G systems, WIMAX systems, and other 802.16
systems. There is a need for flexible management techniques to manage
both users and available spectrum. We will present three practical
protocols for spectrum management in a cellular radio system.
1. Resource allocation at the operator level: the benefits of allowing
inter-operator spectrum sharing are demonstrated.
2. Leveraging geographic and wireless environment: exploiting the
natural attenuation in the wireless environment together with a more
realistic model of traffic patterns can yield significant throughput
gains and power savings.
3. Efficient resource allocation in tiered networks: deployment of a
secondary wireless ad-hoc network that opportunistically uses available
spectrum from a cellular network.
We will demo a multi-hop routing protocol to connect two users with the
unused portions of the spectrum in a cellular radio system.
Rice@TI
Poster: Multi-MAP Multi-Memory Turbo Decoder Design for 3GPP LTE
and WiMax - (
pdf)
Authors: Yang Sun, Yuming Zhu (TI), Manish Goel (TI), Joseph
Cavallaro
Abstract: Efficient hardware turbo decoder implementations are
mandatory for high throughput wireless communication systems; e.g.
WiMax and 3GPP LTE. To apply turbo codes in 4G wireless systems, high
throughput decoding of turbo codes becomes a critical issue. To
achieve the target throughput requirement (>100 Mbps), we employ
parallel decoder architecture to increase the throughput and reduce the
latency by instantiating multiple MAP-decoder and multiple memory
banks.
Rice@TI
Poster: Wireless Class D Amplifier
Authors: Charlie Ice, David Figoli (TI)
Abstract: Class D amplifiers with their 90% efficiency and small
footprint hold the future of digital audio. However, they require the
audio signal to be pulse width modulated(PWM) instead of the
traditional pulse coded (PCM) modulation scheme. The TI C2000 DSPs have
an on chip high resolution PWM peripheral, making them the perfect
choice for controlling a Class D amplifier. Furthermore the C2000 is
designed as a digital power controller, allowing it to control both
the power stage and the amplifier at the same time. By adding a
Chipcon wireless chip into the mix, the system becomes a highly
compact and
efficient wireless speaker.
Rice@ViaSat
Poster: Long-delay Satellite
Channel Emulator
Authors: Barron Stone, Brent Stephens, and C.J. Steuernagel
Abstract: The enormous popularity of modern consumer handheld
devices like Apple's iPhone has demonstrated the market for high-speed
and wide-area Internet connectivity. Among the many new standards
developed to satisfy this burgeoning market is 802.16, more commonly
known as WiMAX. We have created a prototype of a multi-user WiMAX
system for Rice University's Wireless Open-Access Research Platform
(WARP). As well as serving as a demonstration of a WiMAX system for
WARP, the prototype is designed to serve as a WiMAX testbed. This
prototype will allow WARP users to experiment with physical (PHY) layer
and medium-access (MAC) layer techniques for scheduling, power control,
subcarrier allocation, synchronization, and other algorithms to combat
problems that need to be addressed in future WiMAX applications.
Rice@Xilinx
Poster: Enabling WiMAX on WARP:
PHY and MAC Implementation Framework
Authors: Chris Hunter, Raghu Rao (Xilinx), Chris Dick (Xilinx),
Ashutosh Sabharwal
Abstract: The enormous popularity of modern consumer handheld
devices like Apple's iPhone has demonstrated the market for high-speed
and wide-area Internet connectivity. Among the many new standards
developed to satisfy this burgeoning market is 802.16, more commonly
known as WiMAX. We have created a prototype of a multi-user WiMAX
system for Rice University's Wireless Open-Access Research Platform
(WARP). As well as serving as a demonstration of a WiMAX system for
WARP, the prototype is designed to serve as a WiMAX testbed. This
prototype will allow WARP users to experiment with physical (PHY) layer
and medium-access (MAC) layer techniques for scheduling, power control,
subcarrier allocation, synchronization, and other algorithms to combat
problems that need to be addressed in future WiMAX applications.