Radio Frequency Fingerprint Identification for 5G Mobile Devices Using DCTF and Deep Learning

Area of Science:

• Cybersecurity
• Wireless Communications
• Signal Processing

Background:

• Fifth-generation (5G) networks face significant security vulnerabilities.
• Physical layer authentication using Radio Frequency Fingerprinting (RFF) offers a promising defense.
• RFF can detect sophisticated attacks like spoofing and Distributed Denial of Service (DDoS).

Purpose of the Study:

• To evaluate the performance of RFF identification for 5G mobile phones.
• To propose novel DCTF-based RFF identification schemes for enhanced security.
• To assess the accuracy of these schemes under varying conditions.

Main Methods:

• Extraction of Differential Constellation Trace Figures (DCTFs) from Physical Random Access Channel (PRACH) preambles.
• Development of an index-based DCTF identification scheme utilizing a full PRACH preamble database.
• Proposal of a group-based DCTF identification scheme for scenarios with partial preamble training.
• Experimental validation using six 5G phones across three models and an OpenAirInterface-based 5G gNB.

Main Results:

• The index-based DCTF identification scheme achieved 92.78% classification accuracy at a 25 dB signal-to-noise ratio.
• The group-based DCTF identification scheme demonstrated 89.59% classification accuracy.
• The experimental setup confirmed the practical feasibility of the proposed methods.

Conclusions:

• DCTF-based RFF identification is effective for authenticating 5G mobile devices.
• The proposed index-based and group-based schemes offer robust solutions for 5G physical layer security.
• These findings contribute to securing the evolving 5G ecosystem against emerging threats.