Acta Optica Sinica, Volume. 43, Issue 7, 0715002(2023)
Simplified Multi-Channel Parallel Optical Performance Monitoring Based on Deep Learning
Fig. 2. Process of spectrum preprocessing. (a) Original spectrum; (b) spectrum after downsampling; (c) spectrum after filtering
Fig. 3. Obtained Ahs under different modulation formats based on principle described in Fig. 1. (a) 64QAM-16QAM-4QAM; (b) 64QAM-64QAM-16QAM; (c) 4QAM-4QAM-64QAM; (d) 4QAM-4QAM-4QAM
Fig. 6. Schematic diagram of three-channel WDM coherent optical communication system
Fig. 8. Monitoring results of MT-DNN with different optimizers. (a) MFI accuracy and MAE of OSNR monitoring; (b) loss of MFI and OSNR monitoring
Fig. 11. MAE of OSNR monitoring for ten different modulation formats and OSNR monitoring error for all signals in testing phase. (a) MAE of OSNR monitoring; (b) OSNR monitoring error
Fig. 12. MFI accuracies and MAEs of three-channel OSNR parallel monitoring of TL-MT-DNN and w/o TL-MT-DNN varying with epoch in training phase.(a) MFI accuracy; (b) MAE of three-channel OSNR parallel monitoring
Fig. 13. Confusion matrix of MFI accuracy and OSNR monitoring errors of three channels for trained TL-MT-DNN based on Eq. (6) in testing phase. (a) Confusion matrix of MFI accuracy; (b) OSNR monitoring error of channel 1; (c) OSNR monitoring error of channel 2; (d) OSNR monitoring error of channel 3
|
Get Citation
Copy Citation Text
Mengyan Li, Jintao Wu, Jingyu Yang, Lifu Zhang, Yong Tan, Tian Qiu, Yuebin Li, Heming Deng, Fengguang Luo, Liu Yang. Simplified Multi-Channel Parallel Optical Performance Monitoring Based on Deep Learning[J]. Acta Optica Sinica, 2023, 43(7): 0715002
Category: Machine Vision
Received: Nov. 22, 2022
Accepted: Jan. 7, 2023
Published Online: Apr. 6, 2023
The Author Email: Luo Fengguang (fgluo@hust.edu.cn), Yang Liu (liuyang89@hust.edu.cn)