UTSA Runner Research Press (Univ. of Texas at San Antonio)
Not a member yet
6846 research outputs found
Sort by
Groundwater Nitrate Contamination and Age-Specific Health Risks in Semi-Urban Northeastern Areas of Saudi Arabia
Full text linkNitrate in groundwater (GW) poses a public-health concern in semi-urban northeastern Saudi Arabia, where households rely on untreated wells. We measured nitrate in 45 wells spanning treated/untreated commercial stations, private domestic wells, and agricultural wells, and linked contamination severity to age-specific risks using the Nitrate Pollution Index (NPI), Chronic Daily Intake (CDI), and Hazard Quotient (HQ). Nitrate ranged from 12 to 380 mg&middot;L<sup>&minus;1</sup> (35% &gt; 50 mg&middot;L<sup>&minus;1</sup> World Health Organization (WHO) guideline), with untreated private and agricultural wells most affected. Based on NPI, 65% of wells were &ldquo;clean&rdquo;, while 18% showed significant to very significant pollution. Infants and children had the highest exposure: CDI frequently exceeded the oral reference dose (1.6 mg&middot;kg<sup>&minus;1</sup>&middot;d<sup>&minus;1</sup>), and HQ &gt; 1 occurred in 56% (infants) and 51% (children) of samples from untreated sources. Treated stations consistently achieved lower nitrate and HQ &lt; 1. Sensitivity analysis identified nitrate concentration as the dominant risk driver, followed by ingestion rate, with body weight mitigating the dose. The findings suggest that monitoring based solely on compliance may underestimate risks in sensitive age groups, thereby advocating for immediate actions such as fertilizer management, septic system upgrades, extension of treatment to vulnerable households, and community monitoring. The integrated NPI&ndash;CDI&ndash;HQ framework provides a replicable methodology for associating groundwater contamination with demographic-specific health risks in arid, water-stressed regions
Experimental and Numerical Evaluation of Stacked Piezoelectrics and Composite Structures for High-Efficiency Mechanical-to-Electrical Energy Conversion
Full text linkAs energy demands grow and the push for sustainable solutions intensifies, harvesting ambient mechanical energy presents a promising route for powering low-energy electronics and autonomous sensors. This dissertation investigates the design, modeling, fabrication, and evaluation of stacked and composite piezoelectric transducers based on PZT-5H ceramics, targeting efficient mechanical-to-electrical energy conversion under low-frequency, high-force loading conditions—such as those found in roadway infrastructure.
Using finite element simulations (COMSOL Multiphysics) and experimental validation, the research explores the effects of stack geometry, electrode configuration, pre-stress conditions, and electrical load matching on energy conversion performance. Results reveal that while increasing the number of piezoelectric layers can enhance power output, stress attenuation and impedance mismatch limit efficiency beyond a critical stack height. Optimal performance was achieved by minimizing pre-stress, tuning electrical load impedance, and refining packaging designs.
To further improve energy density and mechanical compliance, novel 1–3 piezocomposite structures with embedded high-aspect-ratio PZT pillars in a polymer matrix were developed. These composites demonstrated superior electromechanical response compared to monolithic stacks, with an optimal active piezoelectric volume fraction near 20% for balancing power output and structural robustness. In parallel, an aerosol-jet-printable, high-loading PZT nanoparticle ink was formulated to enable additive manufacturing of complex piezoelectric geometries with precise control over feature resolution.
The outcomes of this research establish generalized design rules for scalable, high-efficiency piezoelectric energy harvesters suitable for embedded applications such as roadway-powered sensing. The findings advance the understanding of stress distribution, material behavior, and system integration for next-generation piezoelectric devices. This work bridges materials science, mechanical design, and electrical engineering, contributing to the broader development of energy harvesting systems and smart infrastructure.
The dissertation is organized in six chapters. After a general introduction in Chapter 1, the dissertation presents the research methods and findings in Chapters 2–5 and discusses future work in Chapter 6. Chapter 2 describes the importance of this research and specific goals that will be addressed. Chapter 3 discusses in detail the software utilized for modeling, sample description, working principles and equipment used to conduct experiments. Both computational and experimental results are presented in Chapter 4, with additional details on stress distribution within the stack and packaging design provided in Appendix II.Electrical and Computer Engineerin
Design, Construction and Reinforcement Learning Control of a Wheel-Legged Biped Robot
Full text linkThis dissertation presents the design, construction, and control of a novel hybrid bipedal robot that combines legged and wheeled locomotion to overcome the limitations of current mobile robots. While wheeled robots are efficient on smooth terrain, they struggle on uneven or soft surfaces, where legged robots offer greater adaptability but are significantly harder to design and control. To address these trade-offs, two physical prototypes were developed: Baby Rowdy, equipped with feet, and Rollie Rowdy, featuring a novel tricycle motorized skate configuration. Both robots have 17 degrees of freedom and a bird-like, 3D-printed mechanical structure with an inverted knee design, enabling natural terrain adaptability and blind walking—without cameras or pressure sensors— through backdrivable joints and proprioceptive feedback. Rollie Rowdy’s three-wheel layout improves overall stability, both during rolling and when standing or walking, making it especially suitable for hybrid locomotion. The robots incorporate a multi-level control architecture and custom-built electronic systems, enabling seamless switching between walking and rolling modes. To achieve adaptive locomotion, the robots were trained using Reinforcement Learning in simulation environments (Isaac Sim and Isaac Lab), with control policies optimized through Proximal Policy Optimization (PPO). The learned behaviors enable blind, terrain-adaptive walking and rolling without external sensors. Beyond locomotion, the robots are also designed for human interaction: they feature an intelligent head with 3 degrees of freedom, cameras, a microphone, and a speaker, allowing them to function as social robots capable of engaging with people. This work contributes an end-to-end development pipeline—from mechanical design to AI-based control—and demonstrates the viability of using learning-based locomotion strategies in real-world robotic systems, with broad applications in terrain-independent mobility.Electrical and Computer Engineerin
Development and Optimization of Multi-Well Colorimetric Assays for Growth of Coccidioides posadasii Spherules and Their Application in Large-Scale Screening
Full text link<i>Coccidioides immitis</i> and <i>Coccidioides posadasii</i>, the causative agents of coccidioidomycosis, represent a major public health concern in endemic regions of North and South America. The disease spectrum ranges from mild respiratory illness to severe disseminated infections, with thousands of cases reported annually in the United States and an increasing recognition of its global impact. Despite existing antifungal therapies, treatment remains challenging due to toxicity, drug resistance, and limited therapeutic options. High-throughput screening platforms have revolutionized drug discovery for infectious diseases; however, progress in antifungal screening for <i>Coccidioides</i> spp. has been hampered by the requirement for Biosafety Level 3 (BSL-3) containment. To overcome these barriers, we leveraged an attenuated <i>C. posadasii</i> strain that can be safely handled under BSL-2 conditions. Here, we describe the development and optimization of 96-well and 384-well plate screening methodologies, providing a safer and more efficient platform for antifungal discovery. This approach enhances the feasibility of large-scale screening efforts and may facilitate the identification of novel therapeutics for coccidioidomycosis.Molecular Microbiology and Immunolog
Communication and Humility: A Journey
No full text2nd editionThis book covers the journey of communication development over the life-span including self and other-awareness, listening, knowing, and their impacts on the processes of teaching, mentoring and leading in personal and professional settings.Communicatio
Advanced Data Science Model for Detecting and Classifying IoT Malware
Full text linkFor this study, we focused on developing a robust artificial intelligence (AI) model capable of detecting and characterizing advanced malware in Internet of Things (IoT) devices using network data. By analyzing network traffic with various machine learning (ML) models, our AI model identifies and characterizes malicious activities, significantly improving malware detection accuracy and reliability compared to traditional methods. Leveraging machine learning models such as Random Forest, Decision Trees, and Extra Trees, the model achieves a detection accuracy of approximately 92%, enhancing malware detection and characterization in IoT networks. This facilitates the identification of infected devices and their isolation to prevent further infections. The scalable framework offers real-time threat detection capabilities for IoT networks and will be expanded in future work. Future development will focus on creating an easily deployable agent capable of monitoring network traffic to detect and mitigate attacks, providing a more streamlined and robust approach to securing IoT ecosystems.Electrical and Computer Engineerin
Generation of a Multi-Class IoT Malware Dataset for Cybersecurity
Full text linkThis study introduces a modular, behaviorally curated malware dataset suite consisting of eight independent sets, each specifically designed to represent a single malware class: Trojan, Mirai (botnet), ransomware, rootkit, worm, spyware, keylogger, and virus. In contrast to earlier approaches that aggregate all malware into large, monolithic collections, this work emphasizes the selection of features unique to each malware type. Feature selection was guided by established domain knowledge and detailed behavioral telemetry obtained through sandbox execution and a subsequent report analysis on the AnyRun platform. The datasets were compiled from two primary sources: (i) the AnyRun platform, which hosts more than two million samples and provides controlled, instrumented sandbox execution for malware, and (ii) publicly available GitHub repositories. To ensure data integrity and prevent cross-contamination of behavioral logs, each sample was executed in complete isolation, allowing for the precise capture of both static attributes and dynamic runtime behavior. Feature construction was informed by operational signatures characteristic of each malware category, ensuring that the datasets accurately represent the tactics, techniques, and procedures distinguishing one class from another. This targeted design enabled the identification of subtle but significant behavioral markers that are frequently overlooked in aggregated datasets. Each dataset was balanced to include benign, suspicious, and malicious samples, thereby supporting the training and evaluation of machine learning models while minimizing bias from disproportionate class representation. Across the full suite, 10,000 samples and 171 carefully curated features were included. This constitutes one of the first dataset collections intentionally developed to capture the behavioral diversity of multiple malware categories within the context of Internet of Things (IoT) security, representing a deliberate effort to bridge the gap between generalized malware corpora and class-specific behavioral modeling
Archaeological Report, No. 518
Full text linkFrom May of 2024 to January 2025, CAR staff conducted archaeological testing in support of the Historic Structure Report
(HSR) on the Mission San José Perimeter Buildings conducted by the UTSA Center for Cultural Sustainability in response
to a request from the National Park Service (NPS). Mission San José (41BX3), located south of downtown San Antonio, is a
significant Spanish Colonial site that has been designated a State Antiquities Landmark (SAL) and a National Historic Site, as
well as part of the San Antonio Missions National Historical Park and the San Antonio Missions World Heritage Site, and as
such requires regulatory review by the Texas Historical Commission (THC) under the Antiquities Code of Texas. The project
also required review under Section 106, which was coordinated with NPS and the THC. CAR obtained Texas Antiquities Permit
No. 31739 prior to the start of fieldwork. The project area spanned approximately 2.9 ha (7.2 acres). Cynthia Munoz, CAR
Interim Director, served as the Principal Investigator, and Sarah Wigley served as the Project Archaeologist.
CAR excavated 11 test units along the foundations of the compound walls in order to allow CCS staff to examine and document
below-surface conditions of the wall foundations. Excavations were terminated when the bottom of the foundation was reached
or when obstructive features were encountered. Termination depth ranged from 20-80 centimeters below surface (cmbs). A
limestone base below the reconstructed foundations was recorded in eight of the 11 test units. Fourteen cultural features
dating to the Colonial and Historic periods were recorded in association with extensive historical and colonial deposits. These
deposits have research potential and significance to the history of Texas and the United States. Mission San Jose also has global
significance as part of the San Antonio Missions World Heritage Site. Avoidance of impact is recommended. Artifacts collected
and records generated during the course of the project are curated as CAR Accession No. 2935.Center for Cultural Sustainability, The University of Texas at San AntonioCenter for Archaeological Researc
Responsible AI for Healthcare and Community Well-Being From Ethical Design to Practical Deployment
No full textArtificial intelligence offers unprecedented opportunities to improve healthcare outcomes and inform public health policy, yet when trained on data that underrepresents certain gender groups, racial and ethnic minorities, or socioeconomically disadvantaged populations, AI can entrench and amplify existing inequities. In clinical settings, biased models may underdiagnose or mistreat vulnerable patients; in public discourse, AI-driven analyses of news and social media can skew perceptions and policy decisions, further marginalizing those already at risk. Addressing these challenges is crucial to ensure that AI delivers benefits equitably across all populations. To address these issues, my dissertation builds a human-centered framework for responsible AI that spans from identifying disparate performance across demographic groups to anticipating potential unintended harms before systems are developed and deployed. Grounded in healthcare informatics and computational social science, this research combines technical innovation with ethical foresight. I develop novel methods for bias detection, stakeholder engagement, and inclusive system design. The first component investigates gender bias in biomedical named entity recognition (NER) models used in pharmacovigilance, revealing consistent underperformance in recognizing female-associated terms and medications. The second component introduces a neural network system for extracting Social Determinants of Health from unstructured clinical notes. This system addresses the challenge of overlapping entities in clinical texts, enabling a deeper understanding of patient context, needs, and behaviors to support more personalized care. Expanding beyond clinical data, the third component explores how media framing affects public perception of cyclists. I present a large-scale dataset and a prompting framework that leverages news source bias to analyze fault attribution and sentiment in headlines, demonstrating how LLMs can be tuned for nuanced, socially aware interpretations. The final component introduce a human-centered framework for ethical foresight that combines automated user story generation with multiagent red-teaming discussions. The system generates narratives showing how an AI model might help or harm different users. Participants then engage in story-driven discussions with simulated stakeholders to explore potential risks, failures, benefits, and user values. Our findings suggest that story-driven exploration helps participants surface a wider range of potential harms and better understand how those harms are shaped by individual contexts and user needs. Together, these contributions form a blueprint for embedding fairness and accountability in AI development. By centering vulnerable populations, engaging stakeholders early, and proactively identifying harms, this research supports the creation of AI systems that are not only technically sound, but also socially responsible.Information Systems and Cyber Securit
Efficient Data Analytics and Resource Management for Space-Air-Ground Integrated Networks
Full text linkRecent years have witnessed explosive growth in mobile devices, such as smartphones, wearables, and UAVs, which generate massive amounts of data daily. This data surge places significant strain on traditional cloud networks, which are often unable to meet the increasing demand for low-latency and high-computation services required by emerging applications such as smart navigation, augmented/virtual reality (AR/VR), disaster assessment, and quantum key distribution (QKD). To address these limitations, space-air-ground integrated networks (SAGINs) have emerged as a promising solution. By combining space, air, and ground networks, SAGINs enable global wireless coverage and flexible service provisioning, making them a critical architecture for next-generation wireless communication systems. However, their dynamic, hierarchical, and heterogeneous structure poses substantial challenges in the coordination of network resources and operations.
To address the above challenges of SAGINs, this work proposes efficient data analytics and resource management strategies by leveraging advanced optimization and learning techniques for efficient resource allocation, task offloading, and network routing across diverse application scenarios. Specifically, we investigate the cooperation among low Earth orbit (LEO) satellites, high-altitude platforms (HAPs), and ground base stations (BSs) to deliver relaying and computation services to distributed IoT devices. To minimize the time-average expected service delay under energy constraints, we design a Lyapunov-based online control framework and develop a hybrid quantum-classical generalized Benders’ decomposition (HQCGBD) algorithm for jointly optimizing resource allocation and task offloading. Furthermore, we propose a novel network function virtualization (NFV)-enabled SAGIN architecture that provides end-to-end communication and computation services. The system is modeled using a multi-functional time-expanded graph (MF-TEG) to jointly optimize user association, virtual network function (VNF) deployment, and flow routing, optimized using a hybrid quantum-classical computing algorithm. In addition, we design a semi-supervised federated learning framework (Semi-FedDA) to efficiently assess building damage from satellite imagery. Finally, we consider a quantum-enabled SAGIN designed to support quantum entanglement distribution for applications such as quantum teleportation and quantum key distribution (QKD). To efficiently maximize network throughput while ensuring high entanglement fidelity, we propose a Benders’ decomposition (BD)-based algorithm that jointly optimizes routing path selection and entanglement generation rates (PS-EGR). Extensive numerical experiments are conducted to validate the effectiveness of the proposed algorithms and schemes.Electrical and Computer Engineerin