RF Signal-Based UAV Detection and Mode Classification: A Joint Feature Engineering Generator and Multi-Channel Deep...

Area of Science:

• Electrical Engineering
• Computer Science
• Aerospace Engineering

Background:

• Unmanned Aerial Vehicles (UAVs) are increasingly used for critical services, necessitating reliable detection and classification.
• Radio Frequency (RF) based methods offer robust UAV detection but face challenges due to signal complexity and interference.
• Accurate identification of UAVs and their flight modes is vital for safe airspace integration.

Purpose of the Study:

• To develop an advanced approach for accurate Unmanned Aerial Vehicle (UAV) detection and flight mode classification.
• To address the challenges posed by complex UAV Radio Frequency (RF) signals and inter-component interference.
• To enhance the safety and efficiency of UAV operations in shared airspace.

Main Methods:

• A novel joint Feature Engineering Generator (FEG) and Multi-Channel Deep Neural Network (MC-DNN) approach was developed.
• FEG employs data truncation, normalization, and a moving average filter for signal pre-processing.
• MC-DNN utilizes multi-channel input to effectively separate frequency components and minimize interference.

Main Results:

• The proposed method achieved high accuracy in UAV detection (98.4%) and classification (98.3% F1 score).
• Experimental validation was performed using a new dataset comprising RF signals from three UAV types across ten categories.
• The approach demonstrated superior performance compared to existing state-of-the-art UAV detection and classification techniques.

Conclusions:

• The integrated FEG and MC-DNN approach provides a highly effective solution for UAV RF signal detection and classification.
• This method significantly improves the reliability and accuracy of identifying UAVs and their operational states.
• The findings contribute to ensuring the safe and secure integration of UAVs into various critical applications and airspace.