Unmanned Aerial Vehicle Detection based on Channel State Information

Civilian unmanned aerial vehicles (UAVs) have been increasingly used in problematic ways. For instance, more and more UAVs disrupt flights and peep into privacy. This problem is likely to expand given the rapid proliferation of UAVs for commerce, monitoring, recreation, and other applications. In this paper, we propose a UAV presence detection system which identifies signal signatures by using the UAV's RF communication. We explore UAV's physical characteristics, including the mobility due to fast moving, spatiality due to UAV's 3D nature, and vibration due to its wing rotation. We consider whether the received UAV signals are uniquely differentiated from other wireless devices. We thoroughly analyze the angle of arrival (AoA) in 3D space, and flexibly apply super-resolution AoA estimation method to calculate the elevation in 3D place. We conduct spectrum on fine-grained channel state information data, and successfully detect the frequency of UAV vibration due to the rotation of UAV's propellers, and finally improve the accuracy through a clustering algorithm. Our system is prototyped and evaluated using commodity WiFi devices in real-world environment. Our system shows a good performance, which achieves 86.6% of accuracy, 87.3% of precision and 85.8% of recall.