PhD Thesis: Programmable Smart Radio Environments: From Theory to Hardware Implementation

Author
Publication Year
2024

Type

Thesis
Abstract

Today’s wireless networks are evolving rapidly, experiencing an unprecedented surge in traffic volume, radio density, and spectral efficiency demands. This thesis addresses the critical challenges arising from this evolution of next-generation (NextG) wireless networks, focusing on three primary objectives: achieving high data rates, ultra-low latency, and massive connectivity.

To meet these diverse and demanding requirements, this thesis poses a central question: Can we build a smarter radio environment controlled and learned by software, capable of self-configuring in real-time to meet different application needs? Current approaches to handle uncontrolled wireless signals are end-to-end, but communication endpoints are limited in their ability to shape inherent propagation behavior. By focusing on changing the environment itself rather than endpoints, this thesis seeks to enhance key aspects of modern wireless networks.

Millimeter-wave technology enables multi-Gbps data rates, but its high-frequency signals are vulnerable to blockage, limiting its practical use. This thesis presents two innovative solutions to overcome this challenge. mmWall is a programmable smart surface, installed on buildings and composed of over 4,000 metamaterial elements. It can steer signals through the surface to extend outdoor mmWave signals indoors or reflect them to bypass obstacles. Wall-Street is a vehicle-mounted smart surface designed to provide robust mmWave connectivity in high-mobility environments, ensuring reliable communication even in dynamic scenarios. Extending our smart radio concepts to ultra-reliable, low-latency satellite networks, we introduce Wall-E, a dual-band smart surface that mitigates signal blockage by relaying full-duplex satellite-to-ground links, and Monolith, a smart surface that boosts data rates for inter-satellite communication. To address the growing overhead in massive Internet of Things (IoT) networks, we propose CLCP, a machine learning technique that predicts the radio environment to reduce communication overhead. This AI-driven approach complements our programmable surfaces, forming a comprehensive smart radio solution.

Given the highly complex nature of real-world systems, conceptual models alone are insufficient to fully explain them. Our solutions are implemented in physical hardware prototypes, integrated with existing network protocols, and rigorously tested through experimentation. This thesis thus offers a concrete answer to the above central question, laying the foundation for software-controlled smart radio environments in NextG wireless networks.

University
Princeton University
Documents
Publication Category