ORCID Identifier(s)

0000-0002-0687-7925

Graduation Semester and Year

Spring 2026

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Computer Science

Department

Computer Science and Engineering

First Advisor

Dr. Habeeb Olufowobi

Second Advisor

Dr. Yonghe Liu

Third Advisor

Dr. Shirin Nilizadeh

Fourth Advisor

Dr. Ming Li

Sixth Advisor

Dr. Christos Papadopoulos

Fifth Advisor

Dr. Nadra Guizani

Abstract

Wireless communication has become the backbone of modern connectivity, enabling the Internet of Things (IoT) to support applications ranging from healthcare and transportation to industrial automation and energy systems. Yet all these advances depend on a fragile foundation: communication. While a reliable and secure communication infrastructure can transform IoT into a foundation for societal progress, its failure can jeopardize essential services and disrupt daily life. To address this fragility, this dissertation investigates the fundamental challenges of IoT wireless communication and proposes solutions that advance efficiency, security, and resilience across cyber-physical systems.

First, to strengthen communication efficiency, this dissertation begins by examining the limitations of the traditional TCP/IP framework and investigates how Named Data Networking (NDN), with features such as content naming, content-centric security, and in-network caching, can provide a more scalable and secure architectural foundation for IoT communication. Building on this foundation, it identifies the inability of NDN to support real-time guarantees and proposes a deadline-aware deterministic scheduling protocol for NDN routers that ensures the timely delivery of safety-critical IoT data. At the physical layer, it addresses the challenge of resource allocation in 5G vehicular networks, where heterogeneous traffic classes with conflicting quality-of-service requirements compete for shared spectrum, and presents a prompt-adaptive framework that leverages large language models (LLMs) for adaptive and generalizable allocation.

Second, this dissertation examines the security implications of integrating AI into communication systems. Emerging paradigms such as semantic communication merge deep learning models with traditional networks to transmit meaning rather than raw data. While this offers significant efficiency gains, it also introduces new vulnerabilities. This dissertation demonstrates how off-the-shelf language models can be exploited to attack these systems, highlighting the urgent need for robust, trustworthy communication.

Finally, recognizing that attackers will inevitably find ways to compromise communication systems, this dissertation moves beyond detection-based approaches to strengthen network resilience. It presents a systematic evaluation of time-series forecasting models for proactive attack defense in IoT networks, complemented by explainable AI frameworks that help network operators identify attack sources and critical network components for timely intervention.

Collectively, these efforts advance the vision of secure, resilient, and efficient intelligent networks capable of powering the next generation of trustworthy cyber-physical infrastructures.

Keywords

Wireless communication, Resource allocation, 5G, Semantic communication, DDoS, Large language model, Reinforcement learning, Named data networking

Download

Available for download on Saturday, May 06, 2028

Share

COinS