An adversarial learning based knowledge distillation optical performance monitoring scheme based on has been proposed for 7 core fiber in this paper. Adversarial learning-based knowledge distillation simplified the architecture of the neural network for optical performance monitoring, including modulation format recognition (MFR) and optical signal-to-noise ratio (OSNR) estimation, in spatial division multiplexing (SDM) fiber transmission systems. On account of the knowledge distillation technologies, the knowledge in large teacher model is transferred to lightweight student model to reduce the complexity of the neural network and the difficulty of deployment. In addition, the adversarial learning is applied to the teacher-student architecture in order to enhance the generalization ability of the student model. After adversarial learning-based knowledge distillation, the student model is suitable for the deployment of the services in optical nodes. Experimentation results indicate that the student model has the 100 % modulation format recognition success rate for QPSK, 8QAM and 16QAM while the RMSE of optical signal-to-noise ratio (OSNR) estimation is below 0.1 dB. Due to its excellent performance and being easy to implement, the proposed scheme has the potential for the next-generation multiple core fiber based optical network.
This paper proposed a novel chaotic physical security scheme based on Variational Auto-Encoder (VAE) for optical frequency division multiplexing-passive optical networks (OFDM-PON). We adopt the deep generative model VAE to generate chaotic sequences for the encryption of OFDM symbols. Different chaotic security schemes are included to improve the key space and sensitivity of chaotic models, thus enhancing the security of the OFDM-PON system. With the training materials of different chaotic security schemes, VAE can learn the complex structure of data distribution in various chaotic models and finally has the ability to generate the key group with a large space. Meanwhile, the benchmark performance of the OFDM system is experimentally investigated in terms of the bit error rate (BER). Moreover, owing to the parallel computing of GPU, the time consumed for training of VAE can be reduced to a large extent, and the time for generation of chaotic sequences via VAE is only 1.38% of that via repeated iteration of equations, which highlights the remarkable reduction in complexity of the chaotic physical security scheme.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.