137091 research outputs found
Sort by
A Multidisciplinary Approach to Evaluating Indoor Air Quality in University Classrooms
No full textIndoor air quality (IAQ) can significantly impact our health, cognition, and well-being. As people spend a substantial portion of their time indoors in settings such as a university classroom, it is essential to create building design and usage guidelines informed by data-driven assessments of IAQ. Towards this end, in-situ air quality monitoring (2020-2023) across the Georgia Institute of Technology was performed using a network of low-cost sensors (LCS) (QuantAQ MODULAIR and MODULAIR-PM) that measured particulate matter (PM1, PM2.5, PM10), CO2, temperature, and relative humidity. Complimentary research-grade instrumentation, including a Vocus Proton-Transfer-Reactor Mass Spectrometer to measure volatile organic compounds (VOC), was deployed during intensive field campaigns. This research is one of the first to evaluate MODULAIR-PM sensor performance, and introduces a novel and easily reproducible method for evaluating IAQ by estimating ventilation rates using LCS PM measurements. This work also implements an innovative use of machine learning to quantitatively compare the impact of building design features, outdoor air, and indoor aerosol sources on IAQ. Finally, VOC measurements during lectures and exams revealed increased levels of human emitted VOC in the air during exams, suggesting that psychological stress impacts indoor air composition. This thesis provides actionable insights into both evaluating and mitigating air pollutant exposure in university settings, with broader implications for public health.Ph.D.Chemical and Biomolecular Engineerin
Differential Games of Mixed Strategies in a Variational Inference Framework: An Application to the Perimeter Defense Problem
No full textThis dissertation formulates a differential game of mixed strategies as an
adversarial Variational Inference (VI) problem so that it can be solved through
the lenses of inference. To achieve the aforementioned goal, the dissertation
extends two existing tools for solving non-adversarial VI problems to the adver-
sarial setting where they are used to solve a non-cooperative differential game
of mixed strategies.
The contributions of this thesis are as follows: a task variable that relates the
likelihood of the success variable conditioned on state and control variables of a
stochastic control problem to the objective function is a general tool that links
control to Bayesian inference; however, to the best of our knowledge, nowhere
in the literature can be found a situation in which it was used such that it
is dependent on states and controls that are adversarial. By formulating the
problem such that the task variable is conditioned on adversarial state and
control variables, we are able to extend results that apply in non-adversarial
settings to adversarial settings: we extends Stein Variational Model Predictive
Control (SV-MPC) to a Min-Max SV-MPC on one hand while we extends the
Cross-Entropy optimization method to a Min-Max Cross-Entropy optimization
method.
Moreover, we demonstrate the approach using robots as players in the perime-
ter defense problem in which multiple defenders are tasked to protect a high-
value target from a team of intruders.Ph.D.Electrical and Computer Engineerin
Metal-Organic Framework Structure-Property Relationships for Selective Sulfur Dioxide Adsorption
No full textAdsorption-based separations offer a relatively energy-efficient approach to flue gas desulfurization, but the development of materials that are selective, stable, and regenerable under realistic conditions remains a challenge. Metal–organic frameworks (MOFs) are promising candidates due to their high surface area, high porosity, and uniquely tunable structures. However, many MOFs are susceptible to degradation or deactivation when exposed to acidic gases, which limits their widespread adoption. This dissertation investigates three MOF design strategies to improve SO2 adsorption performance: mixed-metal incorporation, linker functionalization, and post-synthetic metalation. Each approach is evaluated in terms of its effect on SO2 uptake, selectivity against CO2, and framework stability, with the goal to understand underlying structure–property relationships that can inform better material design.
Chapter 3 explores bimetallic MOF-74 frameworks, incorporating Ni2+ or Cu2+ into MOF-74(Mg) to tune SO2 adsorption behavior. Each of the mixed-metal frameworks exhibited a linear relationship between the two parent frameworks for CO2 adsorption capacity. Increasing Cu content correlated with decreasing SO2 capacity and SO2/CO2 selectivity, but also showed better retention of accessible surface area post-SO2 exposure, likely attributed to stabilization due to distortions at the Cu metal centers. In contrast, Mg–Ni materials displayed irregular behavior with no clear relationship between Ni content and performance. A combination of PXRD,
BET, and XPS showed that across all materials there were significant reductions in porosity with a maintenance of crystallinity, attributed to irreversible binding of SO2. While both Mg and Mg-Ni MOFs irreversibly bound SO2, increasing Cu content was aligned with a smaller amount of SO2 found in the post-desorption samples. These results suggest that while metal substitution can be a strategy to alter adsorptive performance, its results are highly metal-specific.
Chapter 4 investigates the effect of linker functionalization on SO2 and CO2 adsorption in UiO-66 frameworks. A series of derivatives with –CH3, –CH32, –NH2, –NO2, and –COOH groups were synthesized to examine how modifications to surface chemistry and pore size via functional group influence adsorptive performance. Functional groups –NO2,–COOH, and NH2 exhibited the most improved SO2 uptake and SO2/CO2 selectivity in both unary and binary adsorption experiments. The relatively neutral methyl group led to capacity and selectivity improvements as well, though more modest. while –NH2 showed intermediate behavior. Additionally, installation of dual methyl or carboxylic groups led to a larger increase in adsorptive performance. All samples retained crystallinity after SO2 exposure, along with small surface area loss. Combined with the fact that none of the functionalized UiO-66 materials retained detectable SO2 following desorption, SO2 binding in this series appears to be reversible under traditional desorption conditions. These results suggest that functionalization can be used to significantly tune selectivity and uptake and offers potential advantages for regenerability.
Chapter 5 examines the impact of post-synthetic Cu2+ coordination on MIL- 101 frameworks functionalized with –COOH and –SO3H groups. The goal was to introduce alternative adsorption sites and assess whether metalation could improve SO2 uptake or selectivity. Cu coordination to the –COOH group led to a moderate increase in SO2 capacity, while coordination to –SO3H groups resulted in decreased capacity. Both COOCu and SO3Cu materials showed clear evidence of degradation
upon SO2 exposure - PXRD patterns revealed loss of crystallinity, surface areas declined post-exposure, and breakthrough capacity dropped over multiple cycles. Though SO2 was detected in the non-metalated frameworks, it was not detected in either metalated framework after desorption. These results indicate can enable reversible binding in place of MIL-101’s usual irreversible binding sites, but only if the chosen anchoring group does not compromise framework stability. Overall, post- synthetic metalation can alter adsorption behavior, but both the choice of functional group and the local coordination environment must be considered when designing alternative metal binding sites.
Chapter 6 discusses the overall conclusions of this work and outlines future directions to advance MOF design. Overall, this dissertation demonstrates that tuning MOFs for SO2/CO2 separations, like any separations process, requires balancing the competing priorities of capacity, selectivity, and stability. Each design strategy explored (metal substitution, linker functionalization, and post-synthetic metalation) offered distinct advantages, but also introduced tradeoffs. Taken together, the findings highlight how specific structural features influence acid gas adsorption, offering insight into more effective MOF design.Ph.D.Chemical and Biomolecular Engineerin
Leveraging optical clocks and relativistic geodesy for multi-satellite missions and deep space navigation
No full textThe study of gravity throughout history has shed light on a wide range of scientific disciplines, from fundamental physics like Einstein's theory of gravity to planetary geodesy, that influence many aspects of human life today. Despite its rich history, the study of gravity could further benefit from new opportunities enabled by technological advancements. Multi-satellite missions, which provide distributed, robust, and flexible sensor systems, are becoming increasingly feasible due to the advancement in small satellite technologies and reduced launch costs. Additionally, with the advancement of optical clocks, relativistic geodesy, a concept that utilizes gravitational redshift described in Einstein's theory of gravity for gravity modeling, is envisioned to offer a novel approach to observing gravity.
This dissertation aims to develop methodologies for designing and operating a multi-satellite geodesy mission that utilizes relativistic geodesy, as well as for navigating a satellite in deep space by reversing the process of relativistic geodesy. The first contribution is the new assessment of the optimized design of multi-satellite geodesy missions, leveraging genetic algorithms and relativistic geodesy observables. Non-symmetric satellite constellations optimized by genetic algorithms are shown to generate accurate short-term, large-scale gravity products that are complementary to other dedicated satellite geodesy missions. The second contribution is the development of a deep reinforcement learning framework for autonomous operation of a reconfigurable multi-satellite planetary gravity mission. The trained satellites are able to collaborate to achieve the mission objective, highlighting the potential of deep reinforcement learning for more efficient and adaptive operations of future multi-satellite planetary gravity missions. The third contribution is a comprehensive analysis on the feasibility of navigating a deep space mission by reversing the process of relativistic geodesy, in which the state is estimated from the known gravitational potential. While there are some limitations, the results show that the proposed approach could potentially complement the deep space network for future deep space missions.Ph.D.Aerospace Engineerin
Multi-modal data driven approaches for disaster damage assessment and prediction
No full textAs climate change accelerates, disasters pose an increasing threat to human lives and infrastructure. Disaster management is evolving from a reactive approach—addressing damage only after it occurs—to a proactive stage, where potential disasters are anticipated and preparations are made, and ultimately to a predictive stage, where data is used to forecast disaster impacts. While current disaster response remains largely reactive, with growing efforts towards proactive measures, this work addresses the gaps in reactive post-disaster damage assessments and advances the field towards proactive and predictive disaster management, with the goal of improving overall preparedness.
This thesis employs multi-modal data-driven methods to enhance both post-disaster damage assessment and pre-disaster damage prediction. By integrating Geographic Information Systems (GIS), data analytics, and machine learning with diverse data modalities, such as tabular data, social media imagery, satellite imagery, and nighttime light data, the study provides critical insights into disasters like hurricanes, tornadoes, earthquakes, and landslides. These approaches equip stakeholders with valuable information, reinforcing disaster preparedness and response strategies to mitigate future risks and enhance community resilience.
In the field of post-disaster damage assessment, this thesis addresses critical gaps by integrating alternative data sources and objective methodologies to assist prompt and accurate disaster response. Post-disaster damage assessments have largely relied on optical imagery, whether from satellites, drones, or social media, due to the intuitive nature of visual evidence in assessing damage. However, this focus on optical data can overlook other valuable sources of information. Technologies like Light Detection and Ranging (LIDAR) and Interferometric Synthetic Aperture Radar (InSAR), which capture 3D structural data and ground deformation, respectively, play critical roles in areas where optical imagery falls short. In the work "Black Marble Nighttime Light (NTL) Data for Disaster Damage Assessment," the application of NTL data was explored in assessing damage from hurricanes, tornadoes, and earthquakes. The findings revealed that NTL data is particularly effective in identifying hurricane-affected areas needing assistance, thereby enhancing the relief efforts. To address the subjectivity in traditional, human-interpreted damage degree classification, the work "From Pixels to Impact: Estimating Earthquake Damage Severity via Semantic Segmentation of Social Media Images" reframed damage assessment as a semantic segmentation problem. A pixel-level evaluation method with Segformer was developed, providing a more objective and standardized framework for consistent post-disaster reconnaissance.
In the field of pre-disaster damage prediction, this thesis focuses on leveraging advanced machine learning techniques to enhance disaster preparedness. In the work “Enhancing Landslide Susceptibility Mapping (LSM) Using a Positive-Unlabeled (PU) Machine Learning Approach”, instead of considering LSM as a binary classification problem, this work utilized PU learning, which treated areas with no historical landslides as unlabeled rather than negative instances. This approach improves the performance of susceptibility maps, aiding local governments in landslides preparedness. In ‘Predicting Hurricane-Induced Building Damage Using Multimodal Machine Learning: Insights from the StEER Dataset’ and ‘Predicting Tornado-Induced Building Damage: A Comparative Study of Tree-Based Models and Graph Neural Networks’, building damage data from the Structural Extreme Events Reconnaissance (StEER) dataset was explored. One study used multimodal machine learning to integrate pre-disaster data sources, while the other employed graph neural networks to model building interconnections. Both works provide building-level damage predictions, enhancing proactive disaster preparedness in hurricane and tornado scenarios.
This thesis also involves machine learning-aided sensor optimization. In ‘A Data-driven Approach to Optimize the Design Configuration of Multi-Sleeve Cone Penetrometer Probe Attachments’, multi-sleeve Cone Penetration Test (CPT) devices were optimized using a data-driven approach. Sensor configurations were refined to improve soil classification performance while minimizing the complexity, enhancing device efficiency and offering unique data for better pre- and post-disaster soil analysis.Ph.D.Computational Science and Engineerin
Approach for Coupled MBSE and MDAO for Military Transports
No full textPresented at AIAA SciTech 2026Aerospace systems are highly complex, often requiring many years of development and vast amounts of funding to be successfully designed. This work focuses on solving these issues by defining a scalable, integrated process between a Model-Based Systems Engineering database and a Multidisciplinary Design, Analysis, and Optimization (MDAO) architecture that minimizes manual data transfer, enabling early-stage validation of requirements, architectures, and design choices, thereby reducing the risk of error and the overall development costs of a complex system. To demonstrate this integrated process, systems models of a tactical mobility vehicle were created in SysML, connected to a separate MDAO team’s software through a custom parser, and the results were fed into a design space exploration dashboard. Through the integrated MBSE-MDAO process, we were able to easily and quickly identify stringent and flexible requirements, adjust design parameters, and find an optimal design solution to our example problem
Detection of Adventitious Lung Sounds and Respiratory Distress from Pulmonary Induced Vibrations using a MEMS Seismometer Patch
No full textPhysicians assess a patient’s respiratory health by detection of abnormal lung sounds, such as crackles and wheezes, found during pulmonary auscultation using their stethoscope and on their physical examination by interpretation of their respiratory rate combined with a visual assessment of the work of breathing (WoB) to identify common pathological lung diseases. Since these methods are subjective, a low-profile device with a capable, accurate, and quantitative remote monitoring approach, could provide valuable preemptive insights into a patient’s respiratory health, proving to be clinically beneficial. To achieve this goal, we have used a miniature lung patch consisting of a sensitive wideband MEMS seismometer that can be individually placed on the anatomical areas of a patient’s lungs to replicate traditional lung auscultation and WoB assessment. This seismometer patch captures the patient’s lung sounds as pulmonary induced vibrations (PIVs) during deep breathing and inspiratory effort via high-frequency mechanomyogram (MMG) during tidal breathing. It also detects corresponding low frequency patterns, specifically respiratory rate and pattern during the breathing cycle. To determine if the seismometer patch recordings can be used to automatically detect adventitious lung sounds, a binary classifier of wheeze versus normal breath sounds was first used to determine its applicability. The binary classifier was later expanded to a categorical classifier with using a novel data fusion deep learning model to determine if the recording contained normal, crackles or wheezing. The data fusion deep learning model was developed with combined inputs of PIV lung sounds and corresponding respiratory phase. The categorical data fusion deep learning architecture exhibited high accuracy, sensitivity, specificity, precision and F1 score of 93%, 93%, 97%, 93% and 93% respectively. The seismometer patch was able to accurately quantity a patent's work of breathing (WoB) by combining the average inspiratory effort via high-frequency mechanomyogram (MMG) signals and respiratory rate compared to the clinical standard. This work empowers remote patient monitoring via adventitious lung sound detection with PIVs acoustic map and non-invasive WoB measurement, providing essential respiratory data for tracking of pulmonary disease development.Ph.D.Electrical and Computer Engineerin
Revolution-Spaced Output-Feedback Model Predictive Control for Station Keeping on Near-Rectilinear Halo Orbits
No full text© 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works
Salt Typhoon’s Cyber Espionage: Applying the Diamond Model and Assessing Policy Governance
No full textThe People’s Republic of China (PRC)–sponsored advanced persistent threat (APT) known as Salt Typhoon conducted a sustained cyber campaign against United States telecommunications providers from 2023 to 2024, resulting in widespread compromise of critical infrastructure and exposure of sensitive communications metadata and law enforcement systems (Miller et al., 2024). This paper applies the Diamond Model to analyze the intrusion, systematically identifying the adversary, capabilities, infrastructure, and victims, extending the framework through social-political and technological meta-features. Living-off-the-land techniques, exploitation of unpatched Cisco vulnerabilities, and abuse of native network protocols enabled covert, long-term persistence and data exfiltration with minimal detection. Evaluating organizational, national, and transnational policy responses, this paper concludes that enforceable national-level centered upon vulnerability disclosure, supply chain accountability, and coordinated federal oversight are the most effective means to mitigate future nation-state campaigns