Achieving Resilient and Self-Adaptive Topology Configuration in 3D UAV Networks

Abstract

3D networks with unmanned aerial vehicles (UAVs) are emerging as a cornerstone of next-generation communication infrastructure, offering flexibility and enhanced coverage in challenging environments. However, prior works predominantly focus on UAV-specific optimizations and high-throughput strategies, often overlooking the critical aspect of network reliability when incorporating these movable entities in the infrastructure. Resilience is paramount in such networks, as it ensures stable performance and connectivity in the face of dynamic conditions, such as mobile edges, transient ground devices, and significant signal interference from urban environments. To address this gap, this article proposes a topology-driven scheme from a holistic view of the 3D networks, leveraging comprehensive scene-based information to enable real-time network adaptability through topological (re)configuration. We decompose this reliability problem into three intertwined stages: topological resilience quantification, UAV self-positioning, and learning-based connectivity optimization. This framework ensures network resilience from a functional perspective, emphasizing the ability to consistently deliver high-quality performance while mitigating connectivity interruptions, essential for reliability of next-generation 3D communication infrastructure. Experimental results validate the effectiveness of our approach, demonstrating significant improvements over traditional methods in terms of bandwidth allocation to ground devices and load balancing among UAVs. Notably, our system excels in highly dynamic scenarios, where it adapts to network instability and connectivity failures on-demand, ensuring consistent and reliable communication performance.