Machine Learning-Based Detection of Unmanned Aerial Vehicles Using Radio Frequency Signals

In recent years, Unmanned Aerial Vehicles (UAVs), commonly known as drones, have rapidly gained popularity due to technological advancements and significant cost reductions. Although UAVs have demonstrated their effectiveness in a wide range of applications, from security to agriculture, and from research to rescue operations, they are also being used for malicious activities, leading to significant concerns regarding privacy, security, and safety. This study employs low-complexity machine learning methods to detect UAVs by using Radio Frequency (RF) signals emitted during real-time communication between the UAV and the control device, as opposed to deep learning methods. For this purpose, the performance of K-Nearest Neighbors (KNN), Random Forest (RF), and Decision Tree (CART) algorithms was evaluated. The results show that these algorithms are effective for UAV detection with an accuracy and F1 score exceeding 99.7%, and for UAV identification with an accuracy and F1 score of 88.4%.