Towards Safer Connections: Secure Authentication in 5G Networks Leveraging Radio Frequency Fingerprinting

The advancement of Fifth-Generation (5G) technology has greatly enriched individuals' access to a broad array of convenient services, offering significant benefits in mobile communications, autonomous driving, smart homes, and the Internet of Things. However, with the increase in the number of network access devices, large and frequent data interactions have brought potential security risks to wireless networks, which are related to user privacy security, traffic safety, and even social security. Malicious intrusions targeting wireless communications have resulted in significant disruptions and incurred substantial losses. Presently, user identities in wireless communications rely on software-based data identifiers, which can be forged easily. This paper proposes a hardware-level device authentication mechanism that uses the intrinsic hardware characteristics of the transmitter (shown as impairments in the radio signal of the transmission link) for identity authentication. The resulting fingerprint is naturally distinctive and highly resistant to counterfeiting attempts. We use 5 devices across different models and 10 devices of the same model for experimental evaluation. Then we apply machine learning algorithms to construct a fingerprint database and device authentication, achieving an average of 82.4% precision, 89.9% recall, and 85.9% F1-score, which can help to safeguard the security of 5G communication.