CAREER: SIM-Integrated Air-to-Ground Autonomous Vehicle Networks

Existing vehicular networks frequently fall short of the ability to process and transmit large volumes of data with ultra-low latency, high reliability, and low energy consumption. This project develops and evaluates a novel air-to-ground vehicular network to overcome energy and latency bottlenecks at the mobile edge, leveraging unmanned aerial vehicles (UAVs) as mobile communication and computing platforms to provide on-demand services. Additionally, selected signal-processing tasks are moved from power-hungry digital hardware to analog domain by exploring programmable stacked intelligent metasurface (SIM) antennas to further improve energy efficiency and resilience. The system is optimized by AI-driven resource allocation, UAV trajectory planning, and vehicle offloading strategies. It enhances intelligent transportation systems, supports safer transportation, and improves communication during emergencies. The project contains an integrated education plan including undergraduate research engagement, new graduate courses, drone coding and traffic safety challenges. It also develops a disaster-relief pop-up connectivity kit for emergency communication scenarios, and releases datasets and open-source software. This project aims to develop a comprehensive framework for SIM-integrated air-to-ground autonomous vehicle networks and investigates how programmable wave-domain beamforming, UAV-assisted vehicular edge computing, and AI-driven resource allocation can be jointly designed for uplink task offloading and downlink service delivery in highly mobile wireless environments. The research agenda is organized into three thrusts: (1) developing a predictive, multi-agent spatiotemporal intelligence scheme to predict mobility and SIM-aware quality of service; (2) designing scalable AI-driven control methods to jointly optimize UAV positioning, vehicle task offloading, radio resource management, and SIM configurations for energy-efficient uplink transmissions; and (3) developing hierarchical schemes using advanced reinforcement learning to adapt wireless resources for downlink optimization under dynamic traffic and wireless conditions. Validation is performed with simulation, hardware-in-the-loop testing, and staged experiments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.