UAV/Drone Detection and Classification Using Radio Frequency Machine Learning

With the increasing development and popularity of Unmanned Aerial Vehicles (UAVs, also referred to as Drones), it has become easier for civilians to purchase or build their own. Drones have become essential in various industries such as agriculture, transport, and entertainment. However, they have also become a security and safety risk. Drones can also be used as weapons and for terror, as demonstrated in recent worldwide conflicts. This raises the need for a remote wireless detection and classification system using Radio Frequency Machine Learning (RFML) to monitor, detect, and classify RF signals from the drone communication system. This project used Fieldfox and HackRF hardware to capture and train drone signals from three different DJI drone models and uses a combination of energy detection and the Orthogonal Frequency Division Multiplexing (OFDM) estimator for classification. The advantage of using energy detection is that it does not require prior knowledge of the signal transmission of the drones and significantly helps with blind detection of various drones. After data were collected, we trained and tested various machine learning models, and the model with the highest accuracy was the KNN model. We also demonstrate initial results on the efficacy of the developed model in real-time drone detection and classification with promising classification accuracy.