Author ORCID Identifier

0009-0007-3812-4024

Document Type

Dissertation

Date of Award

5-31-2026

Degree Name

Doctor of Philosophy in Electrical Engineering - (Ph.D.)

Department

Electrical and Computer Engineering

First Advisor

Nirwan Ansari

Second Advisor

Tao Han

Third Advisor

Abdallah Khreishah

Fourth Advisor

Roberto Rojas-Cessa

Fifth Advisor

Yu-Dong Yao

Abstract

The forthcoming sixth-generation (6G) and future wireless networks are envisioned to support an unprecedented range of services, delivering ultra-low latency, massive connectivity, and intelligent real-time responsiveness. These capabilities will enable emerging applications such as extended reality (XR), autonomous vehicles (AVs), industrial robotics, and the Internet of Things (IoT) to reach their full potential. Achieving this vision requires the integration of enabling technologies such as artificial intelligence and machine learning (AI/ML) and quantum computing, which are poised to play central roles in shaping the landscape of wireless communication systems.

In AI-native, data-driven, and computing-centric 6G networks, ML models will be deeply embedded throughout the system—from the radio access network (RAN) to the mobile edge and cloud core. By employing advanced ML techniques, network functionalities—ranging from modulation and sensing to resource scheduling, network design, and maintenance automation—will evolve from traditional deterministic models to adaptive, learning-based frameworks, thereby enabling the network to gain intelligence and insight from the vast amounts of data it continuously generates.

Realizing an ML model-enabled intelligent 6G wireless infrastructure presents several fundamental challenges. These include enabling seamless ML model support - from training and tuning to deployment and hosting under strict privacy constraints; facilitating secure data and model sharing across carriers and domains within a distributed, privacy-preserving framework; and efficiently accommodating the growing demand for AI-generated content (AIGC) services, such as large language models (LLMs), despite the limitations imposed by scarce spectrum and constrained computing resources at terminal devices, mobile edge nodes, and core network nodes.

To address these challenges, this dissertation designs novel architectural schemes at both the core network and the mobile edge to support ML model enabled functionalities and deliver efficient ML model services. In the core network, a federated learning—enabled Network Data Analytics Function (NWDAF) architecture incorporating Partial Homomorphic Encryption is developed to enable privacy-preserving model sharing across multiple NWDAFs and distributed data sources. This design supports collaborative model training and sharing without exposing local data, and enables practical scaling of federated learning group in loosely controlled wireless networks, such as heterogeneous 6G networks.

At the mobile edge, this work investigates how to support the rapidly growing demand for AIGC and LLM services by hosting model inference service tasks at mobile edge servers. A reinforcement learning—based offloading policy is proposed to dynamically allocates computing resources between mobile devices and edge servers to minimize model inference latency in computing resource constrained and bandwidth-limited wireless environment. The performance advantages of the proposed approach are validated through extensive evaluations.

This dissertation further explores a quantum computing-based approach to solve the challenging mobile edge computing (MEC) task offloading problem. A hybrid optimization framework is proposed, combining classical wireless bandwidth resource allocation for IoT devices connected to MEC servers with Quadratic Unconstrained Binary Optimization (QUBO)-based server selection. This hybrid classical-quantum design ensures stable results while efficiently exploring large combinational allocation spaces. Performance evaluations based on quantum circuit simulation demonstrate lower processing latency and greater stability than conventional reinforcement learning and heuristic baselines in small- and medium-scale IoT deployments. These results point toward a promising pathway for large-scale offloading as quantum technology continues to mature.

Building upon the contributions of this work, several promising directions for future research can be envisioned. First, the MEC offloading problem can be modeled by incorporating real-world datasets generated from transactional model requests and responses. Second, it would be valuable to investigate applying stringent latency constraints in each decision step of reinforcement learning (RL)—based offloading policies, enabling mission-critical communications with ultra-low latency in application scenarios such as robotic manipulation. Finally, the classical federated learning enabled NWDAF infrastructure could be explored within a quantum computing network framework, wherein each participating entity functions as a quantum ML node. These quantum nodes could collaborate via quantum communication channels, leveraging the inherent advantages of secure and high-fidelity quantum communication.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.