At a time when 6G networks demand ever faster, more energy-efficient and more secure communications, Xuanbang CHEN’s PhD thesis explores an original approach to strengthening identification and authentication in visible light communication (VLC). By leveraging the physical fingerprint of LEDs—a unique signature inherent to each component—the thesis proposes a physics-informed method that is more interpretable and lighter than black-box approaches. It also proves more robust to temperature and bias current variations thanks to a drift-cancellation strategy.

This approach relies on an interpretable equivalent-circuit model to extract fingerprint features at the circuit level, while reducing computational load compared to deep learning methods. Experiments conducted on commercial LEDs highlight the impact of real-world operating conditions and the benefits of drift compensation in stabilizing performance. These results open promising perspectives for more reliable VLC systems, suited to the latency and energy constraints of next-generation networks.