Composite Ensemble Learning Framework for Passive Drone Radio Frequency Fingerprinting in Sixth-Generation Networks

Area of Science:

• Electrical Engineering
• Computer Science
• Cybersecurity

Background:

• The proliferation of drone technology presents significant security challenges, including unconventional attacks and swarm threats.
• Effective drone signal classification is crucial for security, resource management, and mobility in upcoming Sixth Generation (6G) networks.
• Existing deep ensemble learning methods are underexplored for 6G drone signal analysis, necessitating advanced detection techniques.

Purpose of the Study:

• To propose and evaluate a novel Composite Ensemble Learning (CEL) based neural network for classifying drone Radio Frequency (RF) signals.
• To enhance the detection of unauthorized drones and potential threats within 6G network environments.
• To improve network security, integrity, and communication efficiency against drone-related interferences.

Main Methods:

• Radio Frequency Fingerprinting (RFF) for identifying and classifying drone signals.
• A Composite Ensemble Learning (CEL) neural network integrating wavelet-based denoising.
• Combined automatic and manual feature extraction to ensure diversity, robustness, and performance.

Main Results:

• The proposed CEL method achieved superior classification accuracies compared to existing deep learning techniques.
• Demonstrated effectiveness across various Signal-to-Noise Ratios (SNRs) on open-source drone datasets.
• Validated the potential for enhanced drone detection and network security.

Conclusions:

• The novel CEL approach offers a promising solution for robust drone signal classification in 6G networks.
• This method can significantly bolster network security and integrity against unauthorized aerial vehicle threats.
• The findings pave the way for improved communication efficiency and safety in drone-integrated 6G systems.