Identifying Consumer Drones via Encrypted Traffic

To address the threats of consumer drones, effective drone countermeasures are required. As with such a process, identifying the specific type of invading drone is the critical initial step. While existing drone identification systems show some advantages, they also have several limitations, such as cost and deployability. In this paper, a unique method is proposed to identify drones based on features in the encrypted traffic between a drone and its controller. These features are constructed based on an in-depth understanding of drone communication protocols, which aids in classifying different types of popular ArduPilot drones. We collected plaintext and encrypted traffic from multiple drones, analyzed their communication patterns, and extracted features to identify specific drones. The experimental results show that the proposed framework can precisely identify target drones while outperforming existing methods. While this paper focuses on drone communications, the proposed method can also be applied to classify many other auto-control devices that exhibit similar communication behaviors.