Delayed self-homodyne detection enabled optical fingerprint identification of ONUs in IMDD passive optical networks

To address the growing security threats, such as identity spoofing, in passive optical networks (PONs), we propose a delayed self-homodyne-enabled optical fingerprint identification method for optical network units (ONUs). By introducing a delayed self-homodyne (DSH) structure at the receiver, device-dependent fingerprint features-primarily laser phase fluctuations-are effectively incorporated into the intensity signal without an additional local oscillator. Optimized feature extraction schemes are designed to isolate these signatures from intensity modulation interference, which are then classified using a convolutional neural network (CNN). The feasibility and robustness of the scheme are validated through numerical simulations and physical experiments using commercial ONU modules. The results demonstrate that the DSH-enhanced fingerprints provide superior discriminability and maintain high identification accuracy under practical low signal-to-noise ratio conditions. Long-term testing further confirms the temporal stability of the proposed solution, indicating that delayed self-homodyne detection offers a robust and physically interpretable approach to enhancing physical-layer security in next-generation PONs.