Treasures @ UT Dallas
Not a member yet
7697 research outputs found
Sort by
Fast Inference and Learning on Hybrid Relational Probabilistic Graphical Models
No full textProbabilistic Relational Models (PRMs) combine the power of Probabilistic Graphical Mod-
els modeling structural data and the capability of First-order Logic representing relationship
on class level. This type of model has shown success on modeling large data, especially en-
terprise data that are stored in relational databases. Efficient reasoning and model learning
of PRMs are critical problems for applying them to large-scale tasks. With recent advances
in lifted inference, symmetries in the PRMs can exploited to perform efficient reasoning and
learning. However, although most real-world applications involve both discrete and continu-
ous (hybrid) features, most existing lifted inference methods have been restricted to discrete
models or continuous models with systematic assumptions, limiting their applications and
ease of use. To extend the applicability of PRMs, we need to design lifted inference methods
and model learning algorithms that are suitable for hybrid data.
In this work, we develop approximate lifted inference schemes based on particle sampling and
variational inference, which have the ability to perform inference on arbitrary hybrid domain
Markov Random Field (MRF) models. We also introduce Relational Neural Markov Random
Field (RN-MRF) models that allow handling of complex relational features with the help of
both neural potential functions and expert defined relational rules. Finally, we propose a
maximum pseudo-likelihood estimation-based learning algorithm with importance sampling
for training the RN-MRF models. The key advantage of our inference and learning approach
is that they make minimal data distributional assumptions and have the flexibility to be
applied to various real-world application. We demonstrate empirically that our inference
methods and proposed learning model are efficient and outperforming existing approaches
in a variety of settings
Analyzing Tidal Circularization In Exoplanet Systems To Determine The Tidal Dissipation Efficiency Of Giant Planets
No full textA planet in the gravitational field of its parent star experiences a tidal force due to the
variation of the gravitation at different points on it at different distances from the star. The
planet gets distorted, being stretched by this difference of gravity. This distortion raises
two bulges (called “tidal bulges”) along the star-planet joining line on two opposite sides
of the planet (p). When it (p) revolves in an eccentric orbit, the tidal distortion varies
since the tidal force varies with the distance from the star. The repetitive tidal distortion
causes a periodic variation of the amplitude of the tidal bulges. The difference between the
planet’s rotational angular speed and the system’s orbital angular speed also varies with
the planet-star distance. It causes the tidal bulges to move around the planet, creating a
tidal wave. The friction and viscous force within the different layers of the planet resist
the motion of the tidal wave and the variation of its amplitude, resulting in the generation
of heat. Ultimately, some portion of the system’s orbital energy converts into heat. The
gradual loss of the system’s orbital energy reduces the orbital eccentricity and semimajor axis.
A ubiquitously used term that parameterizes tidal dissipation is the modified tidal quality
factor (Q′
pl). Q′
pl is inversely proportional to the tidal dissipation rate. In this project, we
determined a possible range of Q′
pl of short-period gas giants. The periodically varying tide
acting on different parts of the planet, sometimes coupling with other forces (like Coriolis
force), generates multiple components of the tidal wave that depend on the time-dependent
tidal frequency. So we prescribe an empirical model where Q′
pl may depend on the frequency
to consider different possible tidal wave components. We applied our analysis to 78 exoplanet
systems consisting of a single planet orbiting a single host star. We worked out an allowed
range of the frequency-dependent Q′
pl for each system and combined them to find general
constraints on Q′
pl. We determined the upper limit of Q′
pl by requiring that if the system
starts evolution with a sufficiently high initial eccentricity, then the eccentricity simulated at
the present age for which the simulated orbital period matches the measured value of the
orbital period, should be lower than the envelope observed in the ‘eccentricity vs. semimajor
axis to planetary radius scatter plot’ of a collection of exoplanet systems. We determined the
lower limit of the same parameter by requiring that it should not be lower than the measured
orbital eccentricity at the present age. We find that the value of log10 Q′
pl for HJs is 5.0 ± 0.5
for the range of tidal period from 0.8 to 7 days. We do not see any clear sign of frequency
dependence of Q′
pl within the mentioned uncertainties
Application of Microporous Coating in Passive Thermal Management Device
No full textDevelopments in electronic applications present numerous thermal management challenges for
the dissipation of waste heat. The work presented in this dissertation is applicable to the
dissipation of persistent heat from large areas and the spreading of waste heat from highly
concentrated heat sources.
Persistent uniform heat generation from a relatively large area is encountered in the use of
batteries for electric vehicles among many other applications. In these applications, heat must be
dissipated from a relatively large surface area, relative to the spacing between two adjacent heat
sources. Highly concentrated heat sources are increasingly encountered due to the rapid
development of electronic packaging techniques. Specifically, the overall system power densities
continue to increase due to the size minimization trend of the electronic devices.
These applications result in significant thermal management challenges. If not managed
properly, increased temperatures can cause significant deterioration in a device’s performance
and greatly reduce the product reliability. Traditional active cooling strategies, such as singlephase and two-phase active cooling systems, using pumps or compressors as auxiliary
components, dissipate reasonably high heat fluxes and can be applied over large areas. However,
the associated auxiliary devices increase the system’s complexity and decrease its reliability.
Alternatively, passive cooling devices utilizing liquid-vapor phase change thermal systems, such
as thermal ground plane (TGP), effectively dissipate or spread the heat. These can be wick type
or wickless and rely on evaporation or boiling phase change. The wick type TGP or vapor
chamber utilizes capillary forces to recirculate the evaporating liquid, the wicking structure plays
an essential role in the overall heat dissipation effectiveness. For the wickless type TGP, which
utilizes a bubble pumping mechanism to circulate the liquid instead of wicking, boiling heat
transfer performance is dominant.
In the current dissertation, firstly, the performance of an aluminum high-temperature, highconductive microporous coating, which can be used in wick-type thermal ground plane as the
wicking material, is characterized through mass and heat transfer experiments utilizing water and
highly wetting fluids for dissipating persistent heat from large areas.
Secondly, the wettability effect on the nucleate boiling heat transfer performance of a copper
high-temperature, thermally conductive, microporous coating is experimentally investigated.
Thirdly, a wickless and orientation independent ultra-thin thermal ground plane is developed
using this copper high-temperature, thermally conductive, microporous coating for spreading
highly concentrated waste heat
Fault Handling for Medium-voltage (MV) Grids
No full textThis thesis provides an overview of the fault detection and protection methods for medium-
voltage grids. First, it discusses and review the evolving direct-current medium-voltage
(MVDC) grids and their application for various on-shore and off-shore cases. It then explores provided techniques and solutions in the literature to study challenges related to the
short-circuit faults that a grid might be prone to them. Advantages and disadvantages of
each technique are investigated. This is done with the ultimate goal to propose a new and
fast fault detection, classification and location control method to be implemented for any
given medium-voltage grid for prompt fault analysis. The so called proposed Grid Transient
Classifier-Active Impedance Estimation (GTC-AGIE) provides a two-step fault detection,
classification and location method based on the artificial neural network (ANN), wavelet
transform (WT) and active high-frequency signal injection. The GTC part decomposes
voltage and current signals using WT to extract feature vectors. Then, by the aid of two
separate ANN, fault type and an estimation of its location (zone and side where fault has
occurred) can be identified. The AGIE plays a complementary role to calculate fault resistance and its distance in a particular zone and side, which are identified by GTC. The
AGIE performs its function by injecting a small duration high-frequency signal into the
grid and then calculates corresponding impedance to retrieve fault distance and resistance.
Shipboard MVDC system is considered as the case study to investigate applicability of the
proposed method. Shipboard grid includes several power and voltage stages with various
interconnections and load zones in a compact structure with small distances. Compared to
alternative current (AC) system, fault current rises quickly in DC ones and a very fast fault
analysis is required. Hence, shipboard MVDC is considered as a good case study to examine
the effectiveness of the proposed GTC-AGIE. In the following, some solid-state fault current
limiter (FCL) topologies for grid protection are reviewed and their advantages and disadvantages are assessed to identify potential areas for improvements. Finally, a novel intelligent
multi-functional fault current limiter (IMFCL) topology is proposed to provide protection
over short circuit faults and also address any voltage sag/swell by operating as dynamic
voltage restorer (DVR) in a hybrid medium voltage alternative current (MVAC) and MVDC
grid. A simple fault disturbance detector is proposed to quickly identify voltage sag/swell
or fault current conditions. Furthermore, in case of any fault occurrence, control system
in IMFCL injects a short duration high-frequency signal into the grid to quickly calculate
system impedance in new condition. By knowing the impedance, it is possible to calculate fault resistance and estimate fault location. Both simulation and hardware-in-the-loop
(HIL) results are presented throughout the thesis to evaluate performance of the proposed
GTC-AGIE and IMFCL
Coupling Physical Measurement with Machine Learning for Holistic Environmental Sensing
No full textThe interest in characterizing the abundance and nature of airborne particulates has been
increasing over the last decade, driven in large part by the rising awareness of the manifold
health impacts of airborne particulates. Since regulatory observations of airborne particulates
are usually made with expensive instruments, the number of sensors that can be deployed is
naturally limited by the costs involved. This dissertation describes the substantial progress we
have made in the physical sensing of airborne particulates by providing low-cost, high-quality
observations of airborne particulates by utilizing advances in low-cost laser-based sensors,
that can be deployed at scale, coupled with machine learning used for accurate calibration of
these low-cost sensors. The abundance of airborne particulates is usually quantified by an
integrated mass density in µg/m3 over the airborne aerosol size distribution (e.g. PM2.5, the
integrated mass density of all airborne particulates with a diameter of up to 2.5 microns).
A persistent feature of all airborne observations of particulates is the variability over small
temporal and spatial scales. This persistent and ubiquitous variability underscores the value of
being able to deploy a large number of low-cost sensors that can make accurate measurements
every few seconds, 24/7. Taking this into account, I have built, calibrated, and deployed a
large number of sensors across the Dallas-Fort Worth (DFW) Metroplex in Texas as a part
of my dissertation work.
Other physical measurements can also be utilized in accurate assessment of airborne particulates. Just as weather RADARs are used to examine the spatial and temporal distribution of
atmospheric precipitation, we show that if we use machine learning, we can also employ the
weather RADARs to examine the spatial distribution of airborne particulates.
CO2 has gained a lot of attention in recent years due to global warming. It is considered the
principal anthropogenic greenhouse gas driving global warming. As a result, CO2 levels must
be monitored and controlled. The present study describes how machine learning can be used
to calibrate a low-cost CO2 sensor which is already part of the sensor systems that I have
built and deployed.
This dissertation provides an overview of how low-cost physical sensing can be combined
with machine learning to provide environmental sensing systems at scale, thus using physics
in service of society
Millimeter-wave Wideband MSK Receiver and Transmitter in CMOS
No full textThe sub-terahertz portion of the electromagnetic spectrum can provide a large bandwidth for both
wireless communication and wireline communication using dielectric waveguides. To fully exploit
the bandwidth, the communication systems inevitably require frequency division multiplexing.
Since integrating a highly frequency-selective multiplexer and a de-multiplexer is challenging at
these frequencies, use of MSK (Minimum Shift Keying) modulation with reduced out-of-band
emission is a potential approach to alleviate this technical challenge. Furthermore, MSK is a
constant envelope modulation and allows more power efficient operation of transmitters. This is
particularly important at sub-terahertz frequencies, where the power efficiency of circuits is low.
Lastly, MSK signals can be demodulated using a phase locked loop (PLL) based receiver that
tracks the carrier frequency of signals incident to a receiver, which greatly relaxes the frequency
synchronization requirements in both transmitter and receiver. PLL-based receivers are also simple
to implement. Although MSK signals have such merits for sub-THz communication, the
previously reported carrier frequency of Gilbert-mixer-based MSK transmitters is lower than 60
GHz and data rate lower than 2 Gbps. The maximum data rate of PLL-based receivers is 10’s of
Mbps. Increasing the data rate of PLL-based receiver and generation of high-data rate MSK signals
are the main topics of this dissertation.
First, a 180-GHz MSK receiver using a phase-locked loop (PLL), which self-synchronizes carrier
frequency is demonstrated. The mixer first receiver is fabricated in a 65-nm CMOS process. A
double balanced anti-parallel-diode-pair sub-harmonic mixer performs the phase detection,
reducing the frequency of LO by half. Tunable zeros realized by series inductors are used to
improve the stability and to increase the data rate handling capability. Without external LO
synchronization, the receiver demodulates MSK signals at 10 Gbps with a bit error rate (BER) of
< 10-12 and at the maximum data rate of 12.5 Gbps with a BER of 3.8×10-5
. The BER at 10 Gbps
is the lowest and the data rate of 12.5 Gbps is the highest for PLL receivers.
Second, high data rate 180-GHz MSK modulated signals for dielectric waveguide communication
are generated using a transmitter fabricated in 65-nm CMOS. To accomplish this, techniques for
controlling the relative phases of half-sine shaping signal and data, “Misaligned-to-Aligned” are
proposed and demonstrated. Limited by the instrumentation for MSK signal analyses, the eyes of
transmitted MSK signals have been verified for a data rate up to 10 Gbps. The MSK signal
generator provides a 5X higher data rate among all the previously reported MSK transmitters at a
3X higher carrier frequency.
Thirdly, a dual-band minimum shift keying (MSK) transmitter operating at 180 GHz and 315 GHz
is demonstrated in 65-nm CMOS. The transmitter incorporates the data encoder and wideband I/Q
phase alignment for MSK signal generation. Limited by the instrumentation for the MSK signal
analyses, the 315-GHz channel is used to form a 10-Gbps link at BER= 5×10-5 with an on-chip
PLL-based receiver. It has increased the highest carrier frequency of MSK signal generation from
180 GHz to 315 GHz. This work also demonstrates the first single-chip transmitter in CMOS that
supports frequency division multiple access (FDMA) communication above 150 GHz
From Single Component to System Level Approximate Computing
No full textMost Integrated Circuits (ICs) are now heterogeneous Systems-on-Chip (SoC) that contain
a variety of hardware accelerators. These dedicated accelerators execute applications that
have large amounts of parallelism, e.g., Digital Signal Processing (DSP) and image processing
applications. This approach can substantially increase the performance, while decreasing
the energy consumption of the SoC. One orthogonal approach, approximate computing, has
emerged as a powerful alternative to further reduce the power of SoCs. In approximate
computing the error at the output is relaxed to simplify the hardware or the software that
run on the processor and thus, achieve lower power. It has proven to be an effective method
to achieve low power for error tolerant applications (e.g., digital signal processing and image
processing) by trading off the accuracy of the circuit vs. area/power/energy.
Most work in approximate computing focuses on basic approximation primitives that fails
to reduce area/power/energy beyond a certain level. Also, most work on approximate com-
puting has focused on specific components within a system. This severely limits the approx-
imation potential as most Integrated Circuits (ICs) are now complex heterogenous systems.
Another additional limitation of current work in this domain is they assume that the training
data matches the actual workload. This is nevertheless not always true as these complex
Systems-on-Chip (SoCs) are used for a variety of different applications.
To address these limitations, this dissertation, presents new approximation primitives to
further approximate SoC components. In addition, it proposes approximation methodology
to perform system level approximation that is to approximate the entire SoC. Finally, to
address the issue of input data distribution change, this dissertation proposes method to ad-
just approximation based on input data distribution type for both hardware accelerators and
entire SoC. The effectiveness of these proposed methods has been verified from experimental
results and comparison with similar state of the art research
Using Public and Private Blockchains for Secure Data Sharing and Analytics
No full textData Sharing has become a prominent issue in today’s world as more and more data is col-
lected for various reasons. Due to privacy, and security concerns and regulatory compliance
issues, the conditions under which the sharing occurs needs to be carefully specified and
managed. Voter registration, financial compliance, healthcare management systems, insur-
ance companies, monetary banks, etc. are just few examples among many that embrace
data sharing at the heart of their ecosystem. These systems require to share data as per
legal agreements and with the right entities. In some use cases, instead of sharing the actual
data, we may need to share data analytics results based on the shared data, (e.g., computing
auction results based on private bids) and make sure those results are secure. In another use
case, we may need to run Machine Learning(ML) algorithms to generate ML models on the
shared data.
In each of these use cases, along with the compliance of regulatory standards, we need to
ensure that data privacy is preserved and finally, offenders that do not comply with the
requirements are automatically penalized. In this dissertation, we consider each of these
scenarios and address the privacy, regulatory compliance and security challenges by incen-
tivizing honest behavior by using Blockchains and Trusted Hardware. First, we present an alternative for tracking, managing and especially adjudicating data
sharing agreements using smart contracts and blockchain technology. Next, we show how to
leverage both public and private blockchain infrastructures to enable efficient, privacy en-
hancing and accountable digital auctions. Furthermore, we delve into the field of Federated
Learning (FL) and show how to secure machine learning models using a hybrid blockchain
architecture that discourages backdoor attacks by detecting and punishing the attackers.
To further secure shared data, we show how we can use the Trusted Execution Environ-
ments(TEEs) to enable efficient, privacy enhancing and secure applications by implementing
oblivious execution while running the smart contracts inside trusted enclaves again in the
context of auctions. Finally, to secure Federated Learning models against privacy leakage
attacks over Blockchain based smart contracts, we show how to take advantage of TEEs to
run those smart contracts inside trusted enclaves without any significant impact on efficiency
El Nuevo Mestizaje/the New Mestizaje
No full textWhen it comes to the concept of mestizaje, both as a general concept and as it has been
articulated specifically by José Vasconcelos and Gloria Anzaldúa. some recent scholars totally
reject the concept and others partially agree with it an in all cases the issue of eugenics as
articulated by Vasconcelos is rejected. I also reject any kind of mestizo eugenics. With this in
mind, the purpose of my dissertation is, drawing from José Vasconcelos (1882–1959) and Gloria
Anzaldúa (1942–2004), to rearticulate the concept of mestizaje as an ideal of diversity, equality,
solidarity, decoloniality, and postcoloniality 1 that has inspired Mexican, Mexican American, and
Latin American culture and to deracialize the concept to understand it as a cultural phenomenon
Routing in Solar-powered UAV Delivery System
No full textAs interest grows in Unmanned Aerial Vehicles (UAVs) systems, UAVs are proposed to take on
increasingly more tasks that were previously assigned to humans. One such task is the delivery of
goods within urban cities using UAVs, which would otherwise be delivered by terrestrial means.
However, the limited endurance of UAVs due to limited onboard energy storage makes it
challenging to practically employ the UAV technology for deliveries across long routes.
Furthermore, the relatively high costs of building UAV charging stations prevent dense
deployment of charging facilities. Solar-powered UAVs can ease this problem as they do not
require charging stations and can harvest solar power in the daytime. This paper introduces a solar-
powered UAV goods delivery system to plan delivery missions by Solar-Powered UAVs (SPUs).
In this study, when the SPUs run out of power, they will charge themselves on landing places
provided by customers instead of charging stations. Some advanced path planning algorithms are
proposed to minimize the overall mission time in the statical charging efficiency environment. We
further consider routing in the dynamical charging efficiency environment and propose some
mission arrangement protocols to manage different missions in the system. The simulation results
demonstrate that the algorithms proposed in our work perform significantly better than existing
UAV path planning algorithms in solar-powered UAV systems