UC Irvine

Advancing towards 6G networks emphasizes integrating communication with sensing functionalities, promising unparalleled connectivity, efficiency, and intelligence in forthcoming networks. In this context, uncrewed aerial vehicles (UAVs) emerge as pivotal assets, offering versatile solutions for both communication and sensing tasks. Leveraging their mobility and flexibility, optimizing the UAV deployment or trajectory can enhance the network performance to meet the new demands.

Anticipating next-generation 6G networks, traditional cellular architectures' constraints have led to the exploration of new network topologies, including cell-free architectures. These architectures abandon the concept of cell, allowing users to connect to multiple base stations and mitigating the effects of cellular boundaries for fairer scenarios. Combining cell-free architectures with UAVs offers substantial performance gains by leveraging UAV adaptability for dynamic coverage and capacity optimization. To fully leverage this potential, we propose a comprehensive framework for cell-free UAV networks. Initially, UAVs operate as flying base stations within a framework of perfect fronthaul connectivity. This paradigm is extended to accommodate wireless fronthaul scenarios, prompting UAVs to function as flying relays instead of flying base stations.

Moreover, UAVs hold significant potential beyond their role in communication. Equipped with sensors and video cameras, UAVs can serve a dual purpose, enabling efficient data collection and sensing tasks. One critical application is wildfire tracking, addressing the pressing need for early detection and monitoring of wildfires. With the escalating frequency and intensity of wildfires globally, efficient wildfire tracking has become imperative for mitigating their devastating impact. Integrating the strengths of cell-free UAV networks with artificial intelligence, our aim is to optimize UAV trajectories to achieve two primary objectives: (i) cover the fire perimeter with cameras and (ii) ensure reliable transmission of captured images to the network. This design significantly enhances resilience, allowing UAVs to transmit images even if certain base stations are compromised by fire incidents. However, the complexity of the overall problem presents a challenge, leading to the utilization of reinforcement learning in this scenario.

In addition to the aforementioned applications in cell-free networks and wildfire tracking, this dissertation also explores similar scenarios with cellular connectivity. This includes exploring the integration of communication, sensing, and data collection functionalities withintraditional cellular networks. These methodologies, which involve optimizing trajectories via traditional techniques or more sophisticated such as reinforcement learning, contribute to enhancing the efficiency and reliability of cellular networks as well.