Computer Science and Electrical Engineering Electronic Theses and Dissertations (UMKC)
Permanent URI for this collection
The items in this collection are the theses and dissertations written by students of the Department of Computer Science and Electrical Engineering. Some items may be viewed only by members of the University of Missouri System and/or University of Missouri-Kansas City. Click on one of the browse buttons above for a complete listing of the works.
Browse
Recent Submissions
Item Reliable and efficient wireless communication channel management using optimization and artificial intelligence(2025) Islam, Md Tajul; Choi, Baek-YoungWith the growing ubiquity of mobile devices and wireless connectivity, ensuring efficient and reliable communication across various environments has become increasingly important. In densely deployed Enterprise Wireless LANs (E-WLANs), traditional strongest-signal-based Access Point (AP) selection can lead to network imbalance and suboptimal throughput. To address this, we propose an optimization-based AP selection scheme that considers user demand and AP capacity, leveraging relaxation and rounding techniques to enhance overall throughput, utilization, and fairness. Expanding beyond terrestrial applications, we explore the challenges of wireless communication in extreme space environments, where electromagnetic interference and radiation hinder reliable data transmission. To tackle this, we introduce a Machine Learning (ML)-based multi-stratum channel coordinator as part of the Resilient Internet of Space Things (ResIST), enabling dynamic and trustworthy channel selection through software-defined wireless topologies. Our simulation-based evaluations demonstrate the superior predictive accuracy of Feed- Forward Neural Networks (FFNN) in these conditions. Finally, we present AIR-CAV, an AI-assisted reliable channel selection framework for Connected and Autonomous Vehicles (CAVs), which integrates real-world and simulated data to dynamically predict signal quality across heterogeneous V2V communication links. Among various ML models evaluated, Convolutional Neural Networks (CNN) achieved the best performance in SNR prediction. Collectively, these contributions highlight the potential of intelligent, adaptive wireless communication strategies in both terrestrial and extraterrestrial domains.Item Enhancing dimensionality in remote sensing images(University of Missouri--Kansas City, 2025) Jiang, Hongcheng; Chen, ZhiQiangRemote sensing relies on diverse imaging modalities to capture critical information about the Earth’s surface. These modalities include panchromatic images, which are single-band grayscale representations; multispectral images, capturing a limited number of spectral bands; and hyperspectral images, encompassing a broad spectrum with numerous spectral bands. Images of the same scene are often acquired by different sensors, leading to datasets that are spatially registered but differ significantly in spatial and spectral resolutions. This variation in dimensionality presents challenges for image processing and analysis, necessitating techniques to enhance spatial, spectral, or combined spatial spectral resolutions. Dimensionality enhancement in remote sensing focuses on three primary tasks. Spatial enhancement improves the spatial resolution of images, such as through super-resolution techniques. Spectral enhancement increases spectral resolution or facilitates translation between spectral domains. Spatial-spectral enhancement combines high spatial and spectral resolutions from various sources, exemplified by pansharpening, where a high-resolution panchromatic image is fused with a low-resolution multispectral or hyperspectral image to produce a high-resolution output. These tasks are inherently complex due to the ill-posed nature of the underlying transformations. This dissertation addresses these challenges through a unified mathematical framework, employing operators to model the required transformations. However, directly solving these operator-based formulations is computationally prohibitive. As a solution, learning-based approaches are employed to reformulate the tasks as supervised or weakly supervised learning problems. These methods leverage data-driven models to approximate the mappings, offering practical and scalable solutions for dimensionality enhancement. This work introduces a unified framework addressing three critical problems in remote sensing: image super-resolution, image translation, and hyperspectral pansharpening. For each task, operator-based formulations define the mathematical principles, while learning-based formulations demonstrate the efficacy of data-driven methods. The proposed framework provides a comprehensive approach to improving spatial and spectral quality in remote sensing imagery, advancing the state-of-the-art in image enhancement techniques.Item Flocking control and intrusion detection in UAV networks(2025) Zeng, Qingli; Nait-Abdesselam, FaridUnmanned Aerial Vehicles (UAVs) have emerged as transformative technologies across numerous sectors, including commercial delivery, agricultural monitoring, disaster response, and military applications. Their ability to operate autonomously in diverse environments has led to rapid adoption and increasing deployment in swarm formations, where multiple UAVs work collectively to accomplish complex tasks. Despite their tremendous potential, this proliferation introduces two critical challenges: ensuring robust network security against cyber threats and developing efficient coordination mechanisms for drone swarms in dynamic environments. Traditional network security approaches for UAV systems face significant limitations. Existing intrusion detection datasets suffer from small sample sizes and unbalanced distributions, while conventional machine learning methods demand extensive labeled data that is both expensive and time-consuming to produce. Similarly, in the domain of swarm coordination, established rule-based methodologies like Craig Reynolds' Boid algorithm provide functional but limited frameworks that fail to capture the adaptive learning capabilities observed in natural flocking behaviors. This dissertation addresses these dual challenges through complementary research streams. First, to enhance UAV network security, we develop a novel approach combining Generative Adversarial Networks (GANs) with a Human-in-the-Loop (HITL) machine learning framework. The GAN component generates synthetic yet realistic network traffic samples to augment limited datasets, while the HITL methodology integrates human expertise to guide the learning process. Our comprehensive evaluation demonstrates that this integrated approach achieves intrusion detection accuracy of up to 99% while reducing the requirement for labeled data by up to 98%, presenting a cost-effective security solution for UAV networks. Building upon this security foundation, we then investigate advanced swarm coordination methods. We introduce a Multi-Agent Reinforcement Learning (MARL) extension to Boid modeling that enables individual drones to make autonomous decisions based on local environmental factors and collective swarm states. Unlike traditional rule-based models, our MARL-driven drones continuously optimize their behavior, mirroring the adaptability observed in natural flocks. Experimental results confirm that this approach outperforms conventional models in cohesion, alignment, and separation metrics, while exhibiting enhanced resilience against environmental variations and external perturbations. To further improve communication efficiency within these secure swarms, we develop the Dynamic BOID Routing Protocol (DBRP), which draws inspiration from biological flocking behaviors to enhance network connectivity and path stability. DBRP incorporates a motion control module that adjusts flight paths based on real-time swarm positioning and velocity, enabling adaptation to topological changes. Performance evaluations demonstrate that DBRP surpasses traditional UAV routing protocols in critical metrics such as latency, throughput, and packet loss rates, particularly in highly dynamic environments. Through these interconnected contributions, this dissertation advances both the theoretical understanding and practical implementation of secure, adaptive, and efficient UAV systems. The synergistic relationship between our security framework and swarm coordination methodologies offers a comprehensive approach to UAV network management with applications spanning civilian and defense sectors, potentially transforming how autonomous drone systems operate in complex real-world scenarios.Item Cost-efficient vision AI: challenges and solutions for real-time and stored video analytics with classical and generative AI(2024) Arefeen, Md Adnan; Uddin, Md Yusuf Sarwar (Mohammad Yusuf Sarwar)Artificial Intelligence (AI) has become integral to vision-based applications, automating tasks such as object classification, detection, and segmentation in domains such as video surveillance. Vision AI systems typically involve real-time or offline analysis. Real-time analysis processes video streams as they are captured, essential for applications like live surveillance, while offline analysis processes large video datasets post-capture, supporting use cases such as crime detection, video summarization, and interactive querying. Despite their significance, Vision AI systems face critical challenges in balancing accuracy and cost-efficiency. Key cost factors include latency, model size, redundant computations, API usage, and data privacy. These challenges hinder scalability and performance, particularly in real-time systems where high latency and large models impede responsiveness. For stored video analytics, computational demands of complex querying and inefficient data processing increase costs, especially with frequent API calls in generative AI models. This dissertation addresses these challenges by exploring innovative solutions for cost-efficient Vision AI systems. Proposed approaches include optimizing model construction, reducing real-time video processing costs, mitigating API expenses in video document analysis, and developing cost-effective generative AI techniques for video analytics. These advancements aim to build a trade-off between accuracy and cost-efficiency, enabling scalable deployment of Vision AI systems across diverse applications.Item IntelliEdgent: device-server collaborative deep learning model composition for resource-efficient edge intelligence(2024) Nimi, Sumaiya Tabassum; Uddin, Md Yusuf Sarwar (Mohammad Yusuf Sarwar); Choi, Baek-YoungDeep Learning models have achieved tremendous success lately towards analysing high-dimensional data like images, texts, audio, etc. Despite their phenomenal predictive performance, the high demands on computation resources (e.g., memory requirement is in the order of hundreds of megabytes for doing a single inference on a typical deep learning model consisting of millions of parameters associated with these models stagger their widespread deployment, particularly in the resource-constrained devices. Hence, the user devices cannot execute the tasks of model inference and maintenance in a standalone manner. Towards this end, the straightforward or Baseline solution to solve these issue is to deploy the models at a server, and the devices communicate with the server for running the models on the received user inputs and updating the models based on the users’ feedback. However, the latency involved with this Baseline setup is typically too high (in the order of hundreds of milliseconds) for each round of communication. Since both the device-only and server-only setup are not efficient enough, we propose a collaborative model composition approach for deep learning execution, based on edge computing, called IntelliEdgent, which intelligently splits the computation workload across both the device and the server, so that the resultant execution latency is optimal. At first, we study the problem of detecting Out-Of-Distribution (OOD) samples on device using a shallow detector, so that for these samples the server is not invoked for getting the classification results (since the results are going to be wrong anyways) in our device-server collaborative IntelliEdgent setup. In the next work, we study the problem deploying multitasking models in our IntelliEdgent setup. We call our approach Chimera, that can dynamically generate optimal multitasking deployment based on the dynamically changing deployment configurations. Thirdly, we study a device-server collaborative Vision Transformer (ViT) classifier called FactionFormer, based on the idea of a dynamically changing narrower deployment context compared to the off-the-shelf pretrained classifier. Besides these three works, we have other two works in the pipeline where we study the problems like personalized recommender model generation under our proposed IntelliEdgent framework.