RF eigenfingerprints, an Efficient RF Fingerprinting Method in IoT Context.

Area of Science:

• Cybersecurity
• Wireless Communication
• Signal Processing

Background:

• Node authentication is critical for Internet of Things (IoT) networks.
• Radio Frequency (RF) fingerprinting offers a non-cryptographic approach to device authentication.
• Existing methods may lack scalability or explainability for diverse IoT environments.

Purpose of the Study:

• To propose a novel RF fingerprinting method, termed RF eigenfingerprints, for enhanced IoT device authentication.
• To leverage techniques inspired by facial recognition for feature extraction and selection.
• To validate the method's performance through simulation, real-world experiments, and hardware implementation.

Main Methods:

• Utilized Singular Value Decomposition (SVD) for automatic feature learning.
• Employed the Ljung-Box test for selecting significant RF signal features.
• Developed a statistical model for device authentication.
• Created a novel RF fingerprinting impairments model for simulation.

Main Results:

• Demonstrated the effectiveness of RF eigenfingerprints through simulation, real-world experiments, and FPGA implementation.
• The proposed method exhibits good scalability, low complexity, and high explainability.
• Validated the performance of the novel RF fingerprinting impairments model.

Conclusions:

• RF eigenfingerprints present a promising solution for secure and efficient authentication in IoT networks.
• The method's inherent properties make it well-suited for resource-constrained IoT environments.
• This work contributes a robust and explainable physical-layer security technique for the IoT.