Low-altitude UAV safeguard: analysis from connection, data and behavior perspectives

Abstract Low-altitude unmanned aerial vehicles (UAVs) are increasingly used in inspection, logistics, emergency response, public safety, sensing, and applications in the low-altitude economy. As UAV systems evolve from isolated aerial platforms into Internet of Drones (IoD), flying ad hoc networks (FANETs), B5G/6G, and edge-cloud systems, their security boundary expands from the UAV platform itself to communication links, control infrastructure, data platforms, edge/cloud services, and airspace actors. This evolution creates cross-layer risks involving malicious access, identity spoofing, privacy leakage, abnormal behavior, and non-cooperative UAV threats. Existing studies have investigated UAV security, authentication, privacy protection, intrusion detection, and counter-UAV detection, but these works are often fragmented across isolated techniques or tasks. This survey integrates recent studies relevant to low-altitude UAV security through three connected dimensions: connection security, data security, and behavior security. Connection security covers lightweight key management, trusted access authentication, cross-domain authentication, and device trust. Data security covers multimodal sensitive-data recognition, privacy-preserving learning, differential privacy, and utility-preserving data use. Behavior security covers continuous trust evaluation, intrusion detection, abnormal behavior recognition, counter-UAV detection, and airspace risk warning. The reviewed literature provides important building blocks, but an integrated, privacy-aware, resource-aware, and deployable low-altitude security infrastructure remains an open research direction.