We propose and demonstrate an identity authentication method in the orthogonal frequency division multiplexing passive optical network (OFDM-PON) by recognizing device fingerprints of optical network units (ONUs). Signal decomposition methods based on wavelet transform are implemented to extract feature matrixes during the pre-process of samples, and then a trained two dimensional convolutional neural network (2D-CNN) is applied to classify and identify these feature matrixes. Experimental results show that the identity of legitimate ONUs can be successfully recognized and 97.41% identification accuracy is achieved. A rogue ONU can be detected with an identification accuracy of 100%, which indicates that the ability of PONs to resist identity spoofing attack is effectively improved. The robustness of the scheme is also verified. With the proposed strategy, the security level of PON system at the physical layer can be increased markedly.
We propose a time delay signature extraction method for optical chaos systems based on a convolutional neural network. Through transforming the time delay signature of a one-dimensional time series into two-dimensional image features, the excellent ability of convolutional neural networks for image feature recognition is fully utilized. The effectiveness of the method is verified on chaos systems with opto-electronic feedback and all optical feedback. The recognition accuracy of the method is 100% under normal conditions. For the system with extremely strong nonlinearity, the accuracy can be 93.25%, and the amount of data required is less than traditional methods. Moreover, it is verified that the proposed method possesses a strong ability to withstand the effects of noise.