ECE Currents is an opportunity to see recent research by faculty, graduate students, and undergraduate students in UA ECE. Talks are about 5 minutes long, with the same time for questions. Currents is held on odd Wednesdays from 3:00-4:00 in ECE 530.
For more information about ECE Currents (or to volunteer to give a talk), contact Jonathan Sprinkle (sprinkle@ECE.Arizona.Edu)
|Prof. David Cox||Radiation Effects on Space Electronics|
Model-Based Design Lab (Rozenblit)
|Technical Realization of the Optimal Motion Planning Method for Minimally Invasive Surgery|
This talks describes a mechatronic (mechanical and electronic)
realization of the optimal trajectory planning and guidance
algorithms for minimally invasive surgical training. Specifically,
the realization implements optimal navigation paths for surgical
instruments in laparoscopic exercises. The underlying system platform
is the Computer-Aided Surgical Trainer (CAST) that consist of
mechanical fixtures equipped with encoders and servo motors.
This hardware provides a means to accurately track the tip
movements of laparoscopic instruments used in minimally invasive
surgery. Furthermore it provides feedback to a PID controller which
implements the optimal instrument trajectories. Supporting software
provides all calibration procedures necessary to maintain the desired
system's accuracy. Details of the mechanical, hardware, and
software components are presented, along with their limitations
and preliminary results.
Future ECE Currents Talks
Previous talks in the ECE Currents Series
|4/04||Dr. Peter Hansen||the Tricorder project|
the Tricorder project aims to develop accessible real-world handheld
devices capable of sensing a variety of atmospheric, electromagnetic,
and spatial phenomena for applications from science education to
|An X-Band Luneburg Lens Antenna Fabricated by Rapid Prototyping Technology|
I'll talk about the design, fabrication and performance of a 3-D
Luneburg lens in this talk. Dielectric constant control of the lens
is based on the mixing ratio of air voids and polymer. The effective
permittivity of the unit cell is estimated by effective medium
theory and extracted by full-wave finite-element simulations.
Fabrication is implemented by a polymer jetting rapid prototyping
method. In the measurement, the lens antenna is fed by an X-band
waveguide. The measured gain of the antenna at X band is from 17.3 dB
to 20.3 dB. The measured half-power beam width is from 19 degrees
to 12.7 degrees while the side lobes are about 25 dB below the main
peak. Good agreement between simulation and experimental results is obtained.
|Microwave Direction of Arrival Estimation Using Luneburg Lens|
A broadband passive direction finding system utilizing Luneburg
lens has been investigated. The simulated Luneburg lens model
demonstrates the advantage of broadband behavior and a high gain.
Numbers of detectors are placed in a circle on the surface of the
Luneburg lens as receivers. With the received power information at
the receivers, the Cramér–Rao bound (CRB) expression for 360° 2-D
direction of arrival (DoA) estimation is derived and plotted with
comparison to root mean square error (RMS) of the correlation estimation
method under the assumption of an additive Gaussian white noise.
With the error estimation information, one is able to decide how many
detectors are needed under certain noise environment and frequency
range of interest.
|Thermoacoustic Imaging for Breast Cancer Detection|
|Yoon Kah Leow|
Akoglu, S. Lysecky Labs
|Energy and Performance Tradeoffs in Designing FPGA based Visual Sensor Network in Greenhouse Crop Monitoring Systems|
|3/7||Prof. David Egolf|
Visiting Professor, U. Idaho
|The Acoustic Orifice as a Nonlinear Resistor: Simulated and Experimental Results|
Given in this presentation is a numerical-iteration scheme for simulating the propagation of
broadband sound through an orifice (small hole) in a thin plate. Such devices behave according
to a fluid-mechanics phenomenon known as vena contracta, whereby orifice impedance is linear
at low sound pressures (analogous to voltage) and highly nonlinear at high sound pressures. The
orifice was modeled with an electrical-analog circuit comprised of both linear and nonlinear
elements. The numerical method was a modified version of harmonic balance, where linear
elements are treated in the frequency domain and nonlinear elements in the time domain.
Mathematical convergence was improved by initializing the iteration with orifice volume-
velocity (analogous to current) data collected via the describing function method by earlier
authors. Experimental validation was demonstrated by recording sound-pressure signals both
upstream and downstream of the orifice plate.
|2/22||Dr. Dathon Golish||AWARE-Wide Field|
The AWARE-Wide Field project, formerly MOSAIC, is a DARPA-funded program
to build a high resolution, wide field of view, near real-time camera.
The camera relies on a multi-scale imaging design that partitions the
global field of view among an array of many "microcameras". We have
previously discussed the optical, electronic, and image formation
architectures of the camera. Here, we will provide an update on the
status of the project, tracking the evolution of the image quality over
the past 18 months. AWARE-Wide Field is a multi-institutional effort,
led by Duke University. Prof. Michael Gehm's lab is leading the image
formation and data processing effort.
|Rate-Adaptive Four-Dimensional Nonbinary LDPC-Coded Modulation for Long-Haul Optical Transport Networks|
Modern optical communication networks have an elevated demand for
high-speed transmission over long distances. However, high-speed
optical transmission at data rates beyond 40 Gb/s is severely affected
by fiber impairments including chromatic dispersion (CD),
polarization-mode dispersion (PMD) and nonlinearities. Coded
modulation when combined with capacity-approaching low-density
parity-check (LDPC) codes has been shown to be an effective technique
to attain aggregate data rates beyond 100 Gb/s at low operating symbol
rates. In this project, we combine exceptional error correction
performances of nonbinary LDPC codes with power-efficient 4D
constellations to enable a rate-adaptive four-dimensional
nonbinary-LDPC-coded modulation scheme for high-speed long-haul
optical transport networks. Transmission distances range from 5,800 km
to 8,900 km at flexible information bit rates beyond 100 Gb/s.
|Synthesis of Android Services from Models|
|2/08||Bané Vasic||Postmodern coding theory|
The talk is about a new class of iterative error-correcting decoders
for low-density parity check codes surpassing the state-of-the-art
belief propagation algorithms with a fraction of their complexity.
|InfoMax Control for Acoustic Exploration of Objects by a Mobile Robot|
We may want an agent such as a robot to gather information about its
environment, even without a specific task to be performed. Knowledge
of the surrounding area can help the agent improve its task performance
or identify new useful tasks. InfoMax control, in which negative Shannon
entropy over objects in the environment is the reward function for
reinforcement learning in a POMDP, is a recently proposed framework for
motivating such exploration. We apply InfoMax control to a physical robot
that gathers acoustic signatures of objects to determine the optimal
sequence of actions to maximize the robot's information about objects in
its environment. We also discuss the implications of InfoMax control and
situations in which it may or may not be appropriate.
|Interrogating the Convergent Recombination Hypothesis for Immune Systems of Jawed Vertebrates with GPU Based DNA Recombination Process|
We are exploring the process of DNA recombination, known as V(D)J
recombination. It is responsible for the development of the adaptive
immune system. A greater understanding of V(D)J recombination can lead
to the development of life saving drugs. This process, which occurs
naturally in our bodies, can be simulated using computer hardware.
However it is extremely computationally demanding. In this study we
mapped the recombination process to a massively parallel GPU architecture,
and for the first time have been able to exhaustively perform
recombination on mouse genes. Using the information obtained from this
study we have been able to evaluate the Convergent Recombination Hypothesis.
|11/30||Prof. Rick Ziolkowski||Passive and Active Metastructures: RF and Optical Applications|
Metamaterials are artificial materials whose electromagnetic
properties can be tailored to specific applications. We are
using passive and active metamaterial-inspired constructs to
miniaturize a variety of radiating and scattering systems while
improving their performance characteristics. Several
metastructure-engineered examples, including high performance
electrically small UHF and optical nano-antennas, will be reviewed briefly.
|Cyclic codes exhibiting the gene regulatory network of the cell cycle|
We present a Boolean gene regulatory network based on cyclic codes,
which can correct, at every given time, one gene expression error.
The topology of a network is highly symmetric and requires only
simple Boolean functions that can be synthesized using genes of
various organisms. The methodology allows construction of artificial
gene regulatory networks with the number of phases larger than in
natural cell cycle.
|Iterative Reconstruction Algorithms in Compressed Sensing|
Compressed Sensing (CS) is a relatively new field in signal processing
which concerns the recovery of a sparse signal from a small set of
measurements. The recovery the sparsest data is known to be NP-hard,
however the L1-minimization problem can be solved by Linear
Programming methods. However these methods have a polynomial
complexity in the length of the signal, which is impractical in some
applications. We give then an overview of current results in iterative
reconstruction of sparse signals using low-density parity check (LDPC)
matrices as measurement matrices as the links between Compressed
Sensing and Channel Coding has be done recently. We will describe two
iterative reconstruction algorithms, Interval Passing (IP) algorithm
and verification algorithm. We then compare their performance to the
L1-minimization performance using quasi-cyclic low-density parity
check (QC-LDPC) matrices in order to show the trade-off between the
performance and the complexity in the reconstruction of sparse data.
|"Two-Bit Bit Flipping Decoding of Binary LDPC Codes"|
We propose a new class of bit flipping algorithms for low-density
parity-check (LDPC) codes over the binary symmetric channel (BSC).
Compared to the regular (parallel or serial) bit flipping algorithms,
the proposed algorithms employ one additional bit at a variable node
to represent its "strength." The introduction of this additional bit
increases the guaranteed error correction capability by a factor of
at least 2. An additional bit can also be employed at a check node to
capture information which is beneficial to decoding. These algorithms
outperform the Gallager A/B algorithm and the min-sum algorithm at much
lower complexity. Concatenation of two-bit bit flipping algorithms show
a potential to approach the performance of belief propagation (BP)
decoding in the error floor region, also at lower complexity.
|"Compressive Sensing of Frequency-Hopping Spread Spectrum Signals"|
Rapid switching of the carrier among many frequency channels
using a pseudorandom sequence makes Frequency-Hopping Spread
Spectrum (FHSS) signals difficult to capture if the hopping
sequence is unknown. The conventional approach to capture such
FHSS signals without the knowledge of the hopping sequence
necessitates capturing all frequency channels and, thus,
requires the analog-to-digital converters (ADCs) to sample
at very high rates. The recent compressive sensing theory
demonstrates that sparse signals can be reconstructed from
much fewer samples than suggested by the Nyquist theorem.
Using the fact that the FHSS signals have sparse instantaneous
spectra, we propose compressive sensing strategies for capturing
them without the knowledge of the hopping sequence.
|11/2||Prof. Loukas Lazos|
Assistant Professor, ECE
|"Detecting and Isolating Packet Droppers in Wireless Ad Hoc Networks"|
Wireless ad hoc networks realize end-to-end communications in a
cooperative manner. In this paradigm, multiple nodes collaboratively
form a multi-hop route, when communication needs to take place between
a source and a destination that are not within communication range.
However, deployed environments may have selfish and nodes that do not
comply with the protocol. This talk addresses the problem of identifying
and isolating misbehaving nodes that refuse to forward packets to the destination.
UECS Lab (S. Lysecky)
|"Dynamic Profiling and Fuzzy Logic Based Optimization of Sensor Network Platforms"|
Sensor network platforms are becoming increasingly complex to design
and optimize due to the multitude of interdependent parameters that
must be considered. To assist platform developers and application experts
of diverse technical backgrounds, we present a dynamic profiling and
optimization platform for sensor-based systems that enables users to
rapidly optimize a sensor network for a particular application without
requiring extensive knowledge of the underlying physical hardware platform.
|"Rectennas and Power Harvesting"|
Rectennas, short for rectifying antennas, are an important component
of a variety of power harvesting and wireless power transmission (WPT)
systems. Rectifying efficiency for specified input power levels and size
are two significant figures of merit for a rectenna system. I will
illustrate our approach emphasizing an electrically small, metamaterial-
inspired near field resonant parasitic (NFRP), planar protractor antenna
that is designed for a GPS L1 rectenna (1.5754 GHz). When the input power
is 0 dBm, the measured rectifying efficiency of the rectenna (ka = 0.808)
is 78% at the GPS L1 frequency
|"Spectral Characterization of the Temperature Performance of Silicon Solar cells"|
Silicon photovoltaic cells are characterized with modified incident
spectrum to reduce the cell operating temperature. The incident
spectrum was modified using filters achieving cell temperature
reduction with the power output marginally affected in some cases.
|"Secure Channel Access Protocol for Multi-channel Wireless Networks"|
The IEEE 802.11 standard for wireless LAN allows for the use of
multiple non-overlapping channels at the physical layer. In this
paradigm, multiple transmissions can take place simultaneously
without interference, and thus a higher throughput can be achieved.
In such multi-channel wireless networks, coordinating multiple
users to access a common set of channels becomes a critical issue, and
selfish or malicious nodes may manipulate with protocol parameters
in order to obtain unfair share of network resources. This talk
addresses the problem of identifying such misbehaving nodes, as
well as mitigating their impact on network performance (e.g. throughput).
UECS Lab (S. Lysecky)
|"Application-Specific Customization of Dynamic Profiling Mechanisms for Sensor Networks"|
To reduce the complexity associated with application-specific tuning
of sensor-based systems, dynamic profiling enables an accurate view
of the application behavior, such that the network can be re-optimized
at runtime in response to changing application behavior or environmental
conditions. However, dynamic profiling must be able to accurately
capture application behavior without incurring significant runtime
overheads. We evaluate a variety of profiling methodologies suitable
for dynamically monitoring sensor-based platforms and analyze the
network traffic, power, and code impacts associated with each method.
|10/05||Dr. Liana Napalkova|
Visiting Research Assistant Professor
Model-Based Design Lab (Rozenblit's Lab), ECE
|"Optimization for Minimally Invasive Surgical Training"|
The efficiency of a surgical training system plays a significant role
in the reduction of operative risks and stress associated with
insufficient experience of a trainee. The primary goal of such
systems is to get a trainee to a higher level of proficiency
without putting patients at risk in the operating room.
This talk will present a novel prototype of a computer-assisted
surgical training system, which provides continuous visual
guidance on optimal navigation and enables the performance
assessment based on predefined proficiency levels for the tasks.
|Alejandro Proaño||Resource-Efficient Location Privacy Protection in Wireless Sensor Networks|
Open wireless sensor networks permit adversaries to intercept
packet communications. In this research, we propose a resource-
efficient scheme for privacy protection in a wireless sensor
network, based on the generation of bogus traffic. Our scheme
uses node scheduling techniques, which find connected dominating
sets (CDS) of nodes in the network. The CDSs are used for the
generation of bogus traffic as well as the transmission of real traffic.
|"A Neurorobotic Model of Walking"|
Walking in animals is a fluid process in which the dynamics of the body,
interactions with the environment, and the center nervous system
interact to produce a stable gait. This walking is fluid, resistant
to perturbations, and incorporates passive swings of the limbs. By
comparison, most walking robots are controlled in a top-down fashion
where the body is carefully balanced and controlled to conform to a
desired trajectory. We developed a bipedal robot that is designed
based on principles of mammalian leg muscle architecture, and
controlled using a simulated neural network that models the central
pattern generator in the spinal column. The robot incorporates load
sensors that model the role of Golgi tendon organs in regulating the
gait during locomotion, as well as other sources of afferent feedback.
|Microwave characterization of carbon nanotube with a FET transistor configuration|
Associate Professor, ECE
|"High-efficiency Rectenna for Wireless Power Harvesting"|
|10/27||Dr. Dathon Golish||"The (other) MOSAIC Project"|
MOSAIC is a DARPA-funded program created to build a large field of view,
high resolution, near real-time camera. In traditional camera design
there is a common trade-off between field of view and angular
resolution. Generally, cameras that achieve both require extremely
expensive and complex lens systems. Alternatively, "scan and pan"
systems raster-scan a high resolution, narrow field of view camera over
a large field of view to stitch together a full image. These systems can
be much cheaper; however, they are typically very slow. MOSAIC
leverages multi-scale design at the optical, electronic, and data
processing levels to parallelize many high resolution cameras, which
together look over a large field of view at near real time speeds.
MOSAIC will consist of approximately 3600 small "microcameras" arrayed
behind a single objective lens. Each microcamera will have dedicated
electronics that flow together in a heirarchical way to combine the
images from each camera into one large view. MOSAIC is a
multi-institutional effort, led by Duke University. Prof. Michael Gehm's
lab is leading the image formation and data processing effort for
|Dr. Amit Ashok||"Sampling - Nyquist to Random: What is Next?"|
It is well known that band-limited signals can be sampled without loss of
information according to the Nyquist-Shannon sampling theorem. Papoulis
generalized this sampling theorem to include sampling at 1/mth the Nyquist
rate when using m distinct samplers. Recently, the emergence of compressed
sensing theory has shown that random sampling at a sub-Nyquist rate can
achieve perfect reconstruction for sparse signals. This progressive development
of sampling theorems from Nyquist to random compressive sampling represents
incorporation of more prior information about signal of interest in devising
novel and increasingly efficient sampling schemes. However, by exploiting
prior information more carefully a still more efficient sampling can be
achieved. I will talk about one example of such a sampling approach.
|"Iterative decoding beyond belief ... propagation"|
Error correcting codes are an integral part of modern day communications,
computer and data storage systems and play a vital role in ensuring the
integrity of data. At the heart of modern coding theory is the fact that
the low-density parity check codes can be efficiently decoded by the algorithm
known as belief propagation (BP). The BP is an iterative algorithm which
operates on a graphical representation of a code by sending coded bit
likelihoods - beliefs. In this talk we describe new paradigm for the design
and analysis of decoding algorithms which are much simpler yet better than
belief propagation. Unlike BP decoders, these decoders do not propagate
beliefs but a rather different kind of messages that reflect the local
structure of the code graph.
|"Queral Networks-a novel model of computation"|
The physics of transistors set practical limits on the maximum speed of serial
computation; parallelism provides the only clear path forward. Artificial Neural
Networks (ANNs) are one alternative, deriving their power from their highly-connected
structure and asynchronous behavior. ANNs develop software automatically (by training)
but only in a write-only language. In this talk I will present the Queron, a
generalization of and improvement upon the Neuron, which can be connected in Queral
Networks (QNs). QNs are a novel Model of Computation leveraging the structure of ANNs
while addressing the inherent write-only problem, allowing the developer to validate
and learn from evolved QNs.
|"Crowd Sourcing and Citizen Science to Aid Global Climate Research"|
Currently, expensive satellite and aerial imagery have been the only source
of image data for vegetation index and phenology research. In this project,
we are developing an end-to-end system for collecting land surface images for
vegetation index and phenology research. The citizen science portion of this
project is an iPhone application that uses geo-positioning, accelerometer,
date/time, and other metadata are then embedded with the images in a
transmission to a centralized processing server where they are validated,
post-processed, and made available in an interactive visualization system. Joint
work with Prof. Kamel Didan and students.
|4/21||Prof. Janet Wang|
Associate Professor, ECE
|"Predicting Yeast Growth Using Hierarchical Design Automation Models"|
Current experimental approaches such as chromatin immunoprecipitation (chip)
and microarray analysis of strains by deleting a single network component,
provide only a limited view of gene network structure and function. In
addition, because the nature of the interaction could vary from one target
gene to another, models and algorithms derived from such data are incomplete
and inaccurate. Hence, biologists are skeptical about what exactly these
models and algorithms can help in real experiments and labs. Unknown factors,
changing setups and varying environment introduce uncertainties into the
biology system and thus lead to different outcomes. At the same time, large
volume lab data also demands approaches that can handle high computational
complexity. This project targets a new hierarchical approach based on
Curvelet to build an executable biology framework for budding yeast cell
cycle. We choose budding yeast cell cycle because yeast cell cycle is stable
and has been regarded as the path to understand cancer cell and other cell cycles.
|"Three-dimensional LDPC-coded modulation with hybrid direct/coherent detection."|
We present a three-dimensional (3-D) low-density parity check
(LDPC)-coded modulation scheme that enables optical transmission beyond 320
Gb/s in aggregate rate using currently available commercial components
operating at 40 GS/s. This scheme introduces significant performance
improvement of up to 4.1 dB at a bit-error ratio of 10^-9 over the
corresponding two-dimensional scheme. In addition, by using LDPC-coded
1024-3D-constellation it is possible, at least in theory, to achieve beyond
a total of 1Tb/s transmission using transmission equipment operating at
100GS/s, once it reaches the maturity of 40-Gb/s systems.
SPACL (Prof. Ryan)
|"Leveraging the Parallels Between Error-Correction Coding and Compressive Sensing"|
This talk will provide a brief introduction to compressive sensing (CS)
and error control coding (ECC). The parallels between the two technologies
will be discussed. The topic will be supported by an overview of relevant papers.
|4/7||Prof. Salim Hariri|
|"a-CMF: Autonomic Cyberspace Management Framework"|
The autonomic paradigm is inspired by the human autonomic nervous system
that handles complexity and uncertainties, and aims at realizing computing
systems and applications capable of managing themselves with minimum human
intervention. To achieve the project goal, we propose to use autonomic agents,
along with statistical and data mining techniques, anomaly behavior analysis,
and intelligent decision fusion to: (i) develop an Autonomic Cyberspace
Management framework (a-CMF); (ii) use the a-CMF to manage security and faults
of critical cyperspace applications; and (iii) develop an a-CMF testbed to
develop, demonstrate and evaluate the next generation of trustworthy management
techniques and tools.
SPACL (Prof. Bilgin)
|"Compressed Sensing using a Gaussian Scale Mixture Model in Wavelet Domain"|
Compressed Sensing (CS) theory has gained attention recently as an alternative
to the current paradigm of sampling followed by compression. While initial
Compressed Sensing (CS) techniques assumed that sparsity transform coefficients
are independently distributed, recent results indicate that dependencies between
transform coefficients can be exploited for improved performance. In this talk,
we explain a method of using a Gaussian Scale Mixture (GSM) model for exploiting
the dependencies between wavelet coefficients in CS.
SPACL (Prof. Marcellin)
|"Visually Lossless JPEG2000"|
In this talk, we explain a method of encoding images in a visually lossless
manner using adaptive visibility thresholds and visual masking effects within
the framework of JPEG2000. Our adaptive visibility thresholds (VTs) are determined
through psychophysical experiments for measuring quantization distortion models.
These VTs are further adjusted using visual masking effects present in the
background image where distortion appears, and are used to determine the quantization
level in the embedded bit-plane coder for visually lossless coding. Codestreams
encoded with this method are fully compatible with Part-I of the JPEG2000. Compared
with numerically lossless compression of JPEG2000, this method achieves a significant
reduction in bitrate without visual quality degradation.
|3/24||Prof. Susan Lysecky|
Asst. Professor, ECE
|"SEED Pod Testbed and Future Research Directions"|
Beginning in the fall of 2007, a multidisciplinary team from
the University of Arizona began to design, experiment, and fabricate an
800-square foot home to demonstrate feasibility of solar powered homes
to compete in the Department of Energy’s 2009 Solar Decathlon. In this
talk, we will provide a short overview of the existing solar home testbed
as well as the future research, specifically a management framework to
optimize energy usage of appliances based on existing and predicted
SPACL (Prof. Ryan)
|"Serial Turbo Coding Performance for Rectangular-Grain TDMR Models"|
This paper studies the performance of a serial turbo code on two
simplified rectangular-grain models for two dimensional magnetic
recording at a density of more than 0.5 bits/grain. We derive one-
dimensional (1D) and two-dimensional (2D) rectangular models and from
these present finite-state-machine (FSM) representations. From
the FSM for the 1D model we computed achievable information rates
assuming independent and uniformly distributed (i.u.d.) binary inputs.
From the (approximate) FSM for the 2D model, we present a detector. We
then present a serial turbo code architecture with constituent
convolutional codes that is capable of achieving 80% of i.u.d. capacity
for the 1D model and 65% of the average of upper and lower bounds on
capacity for the 2D model (from an earlier paper). Joint work with
Roger Wood, Alek Kavcic, and Bane Vasic.
Ping-Pong Control system is a learning tool designed to teach the
basic concepts of system identification, modeling and control.
Developed with an undergraduate electrical engineer’s background in
mind, Ping-Pong Control guides a student through typical system design
steps such as construction, linearization, programming and
implementation. The students design constant gain, PI and PID
controllers and evaluate their ability to cope with disturbances and
overall robustness. The Ping-Pong Control system is low cost so the
students are encouraged to keep the system and continue development
upon completion of the course.
|3/10||Dr. Amit Ashok|
Sr. Research Scientist, OCPL
|"Compressive Imaging: Beyond Random Projections"|
Imaging devices such as cameras, scanners, microscopes typically employ a
measurement basis composed of impulse-like vectors resulting in an isomorphic
measurement referred to an "image." However, such a traditional measurement basis
ignores the fact that nearly all natural scenes are inherently redundant
(sparse/compressible) and therefore, the resulting requirement on the number of
measurements (e.g. pixels) is overly strict. Compressed sensing theory shows that
sparse/compressible signals can reconstructed exactly using a random measurement
basis even when the number of measurements is much smaller than the signal/image
dimensionality (hence the name compressive). This talk with focus on a hybrid
measurement basis that incorporates scene prior information beyond sparsity to
further improve the fidelity of the reconstructed signal.
|"Hybrid Control of a Micro Aerial Vehicle"|
The project work focuses on designing a controller for the pitching up of a
micro air vehicle with vertical take-off and landing capabilities. Because
the dynamic model of the vehicle is highly nonlinear it readily lends itself
to a hybrid controller model to provide a robust transition. The designed hybrid
controller uses multiple PID controllers as well as state feedback. It performs
significantly better than a single PID controller.
Prof. Krunz, Prof. Lazos
|"Spectrum Opportunity-Based Control Channel Assignment in Cognitive Radio Networks"|
We address the problem of dynamic assignment of coordination (control)
channels in cognitive radio networks (CRNs) by exploiting time- and space-
varying spectrum opportunities. Motivated by the inherent grouping of Cognitive
Radio (CR) users according to channel availability, we propose a cluster-based
architecture for control-channel assignment in a CRN. CRs are grouped in the
same cluster if they roughly sense similar idle channels and are within
communication range, either directly or via a clusterhead. We formulate the
clustering design as a maximum edge biclique problem. A distributed cluster
agreement algorithm called Spectrum-Opportunity Clustering (SOC) is proposed
to solve this problem. SOC provides a desirable balance between two competing
factors: the set of common idle channels within each cluster and the cluster size.
A large set of common idle channels within each cluster allows graceful migration
from the current control channel should primary radio (PR) activity appear on
RNSL (Prof. Lewis)
|"Computational, Artificial Personality" |
A computational framework for artificial personality in cognitive robots is
introduced. While every robot has some form of personality, our framework is
flexible and enables the exploration of different behaviors on the same robotic
platform. The framework described here maintains a probabilistic representation of
an internal state that includes emotion, motivation, sensing, and previous action.
The next action is computed by using a massive number of rules implemented using
Bayes Rule. This flexible Bayesian representation of personality allows the robot's
personality to be designed by a personality generator algorithm.
Assistant Professor, ECE
|"A New ECE Testbed for Unmanned Ground Vehicle Applications"|
Starting 2/25, our lab will have a fully actuated Ford Escape Hybrid available
as a research platform. This talk will serve as a motivator for collaborative
research among faculty developing sensing, control, and navigation algorithms that
involve unmanned ground vehicles. Topics include available data and control inputs
enabled by actuation of the vehicle.
SPACL (Prof. Vasic)
|"Construction of LDPC Codes from Latin Squares"|
Low-density parity-check (LDPC) codes are error correcting codes that can
approach the theoretical limit of the channels, called Shannon capacity.
LDPC codes can be classified into two categories: random and structured
codes. Random codes possess excellent error-correcting properties but have
many shortcomings, one of which is the lack of mathematical structure in
the parity-check matrices of such codes. Structured LDPC codes do not have
this disadvantage, but at the cost of some performance loss, due to a large
number of harmful subgraphs, usually called trapping sets. We are currently
studying methods to reduce the number of such harmful subgraphs while
keeping the structure of the codes intact.
Associate Professor, ECE
Network tomography refers to identification of link and node characteristics
in networks using end-to-end measurements. Network tomography may be employed
in communication networks to identify bit-error rates on links, delay at nodes,
detect intrusion at links and nodes, by observing the characteristics of
multi-hop paths. In addition, network tomography may be employed to evaluate
the characteristics of wires (connections) in nanoelectronic circuits. In
this talk, we will discuss the fundamental network tomography problem to
identify additive link metrics and its interesting characteristics.
|2/10||Prof. Ricardo Sanfelice|
Asst. Professor, AME
|"Robust Stability Analysis of Hybrid Control Systems"|
Driven by recent technological advances, hybrid systems have become prevalent
when describing complex systems with both continuous and discrete dynamics.
This ongoing trend has been the thrust for research on modeling, stability
analysis, control design, validation, verification, and simulation of hybrid
systems. These research efforts are also relevant in the design of advanced
control algorithms with mixed continuous/discrete dynamics, that is, hybrid
control systems. Even though hybrid control systems have been shown to enhance
the robustness properties of the closed-loop system in most applications, the
analysis and design for robustness of these systems have not been pursued in a
systematic manner due to the lack of formal tools for that purpose. Instead,
this has been carried out with tools suitable for the particular application of
study. In this short talk, we take a dynamical systems approach to modeling and
robust stability analysis of general hybrid systems. We introduce a mathematical
framework for robust stability of compact sets and pinpoint several engineering
SPACL (Prof. Vasic)
|"Iterative decoding beyond belief propagation"|
At the heart of modern coding theory lies the fact that
low-density parity-check (LDPC) codes can be efficiently
decoded by belief propagation (BP). The BP is an inference algorithm
which operates on a graphical model of a code, and
lends itself to low-complexity and high-speed implementations, making
it the algorithm of choice in many applications. LDPC codes approach
theoretical limits of channel capacity underr BP decoding. However,
this capacity approaching property holds only in the asymptotic limit
of code length, while codes of practical lengths suffer abrupt
performance degradation in the low noise regime known as the error
floor phenomenon. Our study of error floor has led to an interesting
and surprising finding that it is possible to design iterative decoders
which are much simpler yet better than belief propagation! These decoders
do not propagate beliefs but a rather different kind of messages that
reflect the local structure of the code graph. This talk will briefly
introduce this new paradigm.
|"Efficient and truthful mechanisms for opportunistic spectrum access: an economic perspective"|
Opportunistic spectrum access (OSA) is a new technology that can significantly
improve the utilization of the spectrum. Under OSA, a secondary radio (SR) is
allowed access to a channel that is not currently being used by the primary radios
(PRs) of the channel. In this talk, I will give an economic view on the operation of
an OSA network, in which the activities of PRs and SRs are driven by profit. I will
identify various research issues that stem from the for-profit nature of the network.
I will also briefly introduce our research achievements on addressing some of these issues.
|W Nov 18||Hyun-jin Park|
HF Packaging Lab
|"Automatic RF Impedance Correction Circuits for ATE Test"|
This is an Automatic Reconfigurable Frequency Matched Tuner design for
an automatic match control (AMC) system to be used for Device Under Test (DUT)
unit. The tuners using Microstrip Rogers Duroid 4003 and 6010 substrate
materials were designed and verified. The tuners are able to
tune its resonant frequency of 1750MHz and 2170MHz to match with frequency
dependent loads using varactor diodes that are loaded at each stub. A parameter
study was performed to minimize the tuner size from 3.17 inch
by 2.7 inch to 0.74 inch by 0.5 inch.
|"An improved two-antenna direction of arrival (DOA) technique inspired by human ears"|
Microwave passive direction finding is a very important technology that has
many military and commercial applications including electronic warfare, mobile
communications, etc. In this presentation, we propose a novel DOA estimation
technique using only two antennas, which is inspired by the human auditory
system. The improved DOA estimation accuracy is domonstrated in both simulation
|"A PLL/DLL-regulated DC-DC Power Converter for VLSI applications"|
Abstract: A delay-locked loop (DLL) regulated single-inductor multiple-output (SIMO)
power converter is presented. It takes advantage of the fast phase locking
property of a DLL to identify the SIMO converter's regulation errors between
the output voltages and their corresponding references. In response to these
errors, an adaptive peak current modulation technique is proposed to adjust
the instantaneous duty ratios, and thus minimize the regulation errors in
the converter. The DLL acquires locking within 350 ns, allowing it to
respond to load dynamics promptly, which is very desirable for DVS-enabled
power-efficient VLSI systems.
|W Nov 4||Murat Arabaci|
|"Non-binary-LDPC-Coded Modulation Schemes Enabling|
Ultra High-Speed Optical Communication"
Abstract: We present coded modulation schemes based on component
non-binary LDPC codes and show how they can be used to enable ultra
high-speed communication in optical fiber links. We compare the
proposed schemes against the corresponding bit-interleaved coded-
modulation (BICM) schemes from both complexity and performance
standpoints, and demonstrate the dramatic improvement that proposed
schemes have to offer for future forward error correction modules.
|"Biologically inspired models of visual speed estimation"|
Many insects use estimates of the angular speed of the visual image for a
variety of behaviors, including flight speed control, visual odometry and depth
estimation. While these behaviors have been studied extensively, we have yet to
identify the underlying neuronal circuit. In this talk I will present data
supporting a non-directionally sensitive, correlation-type mechanism as
potentially underlying the brain's computation of the speed of the visual
image. This data will come from behavioral experiments using live bumblebees
and behavioral simulations using non-directionally selective motion detectors.
|"Design and Fabrication of a Holographic Solar Concentrator"|
Abstract: Most commercial solar cells with moderate series resistance will have
an efficiency gain by concentration of incident solar flux. We have designed
a concentrator system using transmission holographic gratings. This concentrator
system can be tailored knowing the properties of the photovoltaic material.
The system can operate at low to medium concentration (2x-20x) without having
active cooling and tracking of the sun. Holographic grating is fabricated with
spin-coated Dichromated Gelatin(DCG) on glass substrate.
Modeling and prototypes showed at least 20% increase in overall energy yield.
|W Oct 21||Amit Ashok|
Senior Research Scientist
|"When Less is More!"|
Abstract: Compressive sensing is an emerging area of research in the signal
processing community with potential for widespread impact in the general areas
of sensing and imaging. I will brifely talk about the basic theory of
compressive sensing and our group's research to advance and exploit this theory
to design new sensors/imagers with reduced size, complexity, power and weight
|"Medical Imaging with a Portable Ultrasound System"|
Abstract: An adaptive data prediction based ultrasound receiver is proposed
for power-efficient high-resolution ultrasound imaging processing. The system
efficiently comprises of only a LNA and a 10-bits ADC with a 8th-order linear
predictor. The adaptive data prediction based ADC enhance the overall dynamic
range up to 96 dB. Meanwhile, by eliminating the TGC and its controller in the
conventional receiver design, the noise, distortion as well as the power
consumption coming from these circuits is completely removed. Hence, a low-power
high-resolution design is achieved.
|"A Visual Small Target Tracking System Using Live Dragonflies"|
Abstract: The remarkable ability of dragonflies to pursue small targets under high
speed is due to their rapid processing and relaying of visual information from its
brain through the ventral nerve cord (VNC) to the thoracic motor centers. Open-loop
studies have revealed that among the eight target selective descending neurons (TSDNs)
identified in each of the two parallel connectives of the VNC, at least one was
selective for target movement in the leftward, rightward, upward, and downward
directions. This project will investigate the role of these neurons in closed-loop
small target tracking in a simulated environment.
|"Asymmetric Threat Response and Analysis Project (ATRAP) II"|
Abstract: Asymmetric Threat Response and Analysis Project (ATRAP), is a
"cognitive amplifier," that can ingest and process more information than
a human could ever read (e.g. terabytes of emails, web-pages, and reports).
Visual text processing and correlation capabilities allow ATRAP to fuzzy
match and extract patterns from the data. Reoccurring patterns of activity
can even be used to help predict future events. Future versions will include
behavioral analysis of organizations and groups designed to increase confidence
in the software’s threat assessments.
|W Oct 7||Haoting Luo|
Prof. Louri and Prof. Wang
|"Self-Organization and Self-Healing Techniques for Reliable On-Chip Networks"|
Abstract: Fault-tolerant approaches typically require expensive additional
resources and external control, which may consume a large portion of the area
and power budget for the on-chip networks. This talk introduces a
self-organization and self-healing technique by using reconfigurable
multi-functional blocks. This embedded fine-grained hardware can help
repair the system swiftly and flexibly, thus reaching high-reliability
with very little area overhead.
|"Dual-DLL-Based CMOS All-Digital Temperature Sensor for |
Microprocessor Thermal Monitoring"
Abstract: A low-power high-resolution CMOS all-digital temperature sensor
is proposed for microprocessor thermal monitoring. Dual delay-locked loops
(DLLs) are employed to sense the temperature-induced delay variations. The
temperature information is converted into digital domain with 8-bit
resolution. An analog interpolator is also used to further improve
the sensing accuracy.
|Ziran Wu||"3-D Electromagnetic Crystal Based Components|
for THz Applications"
Abstract: Research involving Terahertz (THz) spectrum has enjoyed dramatic growth in
terms of technical achievement as well as commercial implementation
recently. Together with a perspective of THz backgrounds, this
presentation will highlight our group's research on 3-D electromagnetic
crystal based THz components, including design, realization and
characterization of various components, and potential applications for them.
|W Sep 23||Chuan Feng|
Research Asst. Professor, UA ECE,
Model-Based Design Lab
|"Turn-by-turn navigation for minimally invasive surgery"|
Abstract: To optimize the benefits of minimally invasive surgery (MIS),
it is important to minimize the dangers of its requisite technology.
A GPS like turn-by-turn navigation system will give surgeons additional
guidance within the constrained working space. A prototype system is
under development and a "smart" instrument will be introduced in the future.
S. Lysecky and J. Rozenbilt
|"Engineering Education Outreach Work in TUSD"|
Abstract: Students in middle school are at a critical age where exposure to
engineering can greatly impact their future education goals. Yet, the challenge
remains on adapting pedagogical methods commonly used at the college level to
the middle school classrooms. We are working in the development of an educational
platform to enable users with no formal knowledge in programming and electronics
to design and implement a variety of systems in an intuitive manner. Our work
with two local middle schools shows promising results for the advancement of
precollege engineering education.
|"Effects of Gamma-Radiation on Rare-Earth-Doped Optical Fibers"|
Abstract: Factors such as reliability and high efficiencies drive an
ever-increasing demand to utilize optical fiber technology in adverse,
ionizing-radiation environments. In this presentation the effect of
gamma-radiation on rare-earth-doped fibers will be investigated and
the underlying physical process, known as radiation-induced
|W Sep 09||Bane Vasic|
Professor, UA ECE, Math
|"What is Error Correction?" |
Abstract: We give intuition behind iterative decoding of low-density
parity check codes. These codes have revolutionized digital communications
and are becoming part of numerous standards.
Asst. Professor, UA
|"Avoiding Selective Jammers"|
Abstract: We explore selective jamming attacks in wireless networks,
which attempt to maximize the impact on the network performance while
minimizing the energy expended for jamming and the probability of
detection. We pose several challenges in the realization and mitigation
of selective jamming.
We will explore a project to interface a robot with a dragon fly, and the
spike sorting technique for analyzing nerve cord data.
|W Aug 26||Roman Lysecky|
Asst. Professor, UA ECE
|"Dynamic and Autonomous Software-to-Hardware Translation for|
High-Performance and Low-Power Embedded Computing"
Abstract: This talk will provide an overview of some of the current
and future challenges of dynamic software-to-hardware translation
for embedded computing systems, specifically presenting recent results
in efficient non-intrusive profiling of multithreaded applications.
Sprinkle Lab, UA ECE
|"Transition Maneuver Simulation of VTOL MAV based on Wind-Tunnel Data"|
Abstract: Based on wind-tunnel data, the VTOL MAV simulator is developed,
which is used to search the steady state flight conditions of the MAV.
Simulations are performed to investigate MAV design tradeoffs. Different
control algorithms are discussed and flight test data are presented.
ECE Currents, 2008-2009
|W May 6||Dr. Kamel Didan||Research Professor, UA ECE||"NASA Earth Science Data in Support of Global Change Research"|
|Anantha Krishnan||Graduate Student Researcher, Vasic Lab||"Two-Dimensional Magnetic Recording"|
|Hussain Al-Helal||Undergraduate Researchers, Sprinkle Lab||"UAV Search: Maximising Target Acquisition"|
Abstract: This talk describes analysis of optimal control techniques for simulated quad-rotor unmanned aerial vehicles (UAVs) performing search and rescue missions. Analysis of the controller and guidance laws governing the UAV are described in detail culminating in a closed form expression describing the probability of detection by camera over a certain field.
|W Apr 22||Prof. Chuck Higgins||Associate Professor, UA ECE||"Hybrid Bio-Robotics"|
Abstract: Discussion of the lab's recent work in using living insects as sensory systems for mobile robots, including our visual experiments with hawkmoths and concluding with future plans for dragonflies.
|Gregory M. Striemer||Undergraduate Researcher, Akoglu Lab||"Sequence Alignment with GPU: Performance and Design Challenges"|
Abstract: In bioinformatics, alignments are commonly performed in genome and protein sequence analysis for gene identification and evolutionary similarities. However, execution time of these algorithms on general purpose processor based systems makes it impractical for use by life scientists. In this talk we explore the architectural features of Graphics Processing Units (GPUs) and evaluate the challenges the hardware architecture poses, as well as the software modifications needed to map the program architecture on to the GPU.
|Andreas Schuster||Visiting Scholar, Sprinkle Lab, TU Berlin||"Synthesizing Experiments from Structural Models of Component-Based Systems"|
Abstract: In order to build complex robotics systems, software and robotics experts agree that a component-based approach is a good way to allow the reuse of working software. However, most robotics experts are not familiar with middleware programming, and middleware programmers often lack knowledge of the robotics domain domain.This talk discusses a domain-specific modeling environment that provides the features to set up robotics experiments or simulations without having to be either a robotics or middleware expert.
Computer Engineering Seminar, 2007-2008
In 2007-2008, the Computer Engineering Seminar was held. More information on those talks (including PDF versions of the slides) is available at CE Seminar.