Privacy-Preserving RF Fingerprint Authentication Under Device Aging and Temperature Variations

Abstract

This study introduces a robust, physical-layer privacy-preserving radio frequency fingerprinting (RFF) authentication framework designed to mitigate the harmful effects of environmental thermal fluctuations and long-term hardware degradation. Empirical evaluations show that the proposed architecture yields consistently low equal error rates (EER) and superior classification accuracy across the full range of evaluated temperatures and operational lifecycles. Notably, the framework exhibits exceptional robustness under concurrent environmental and temporal extremes, substantially suppressing false negative rates for highly degraded devices. Furthermore, the cryptographically obfuscated fingerprint templates provide strong physical-layer privacy, minimizing device linkability and cross-database tracking vulnerabilities in database-leakage scenarios. Computational profiling indicates that the proposed methodology remains operationally feasible in real time, requiring a cumulative execution latency of only 121.1 µs per authentication instance on a single CPU core. Collectively, these findings validate that the proposed approach achieves an optimal trade-off among classification efficacy, environmental resilience, data confidentiality, and structural throughput, offering a scalable paradigm for secure wireless device validation. Furthermore, under an adversarial model where the template database is compromised, the obfuscated templates successfully suppress device-tracking accuracy from 99.5% down to a nominal chance level of 2.5%, while demanding a computational overhead of merely 121.1 µs per authentication instance on a single CPU core.