Performance Monitoring of PAM4 Optical Communication System Based on Principal Component Analysis and Support Vector Regression

As a popular signal transmission technique, PAM4 is widely used in short and medium distance optical communication networks. Though there exists techniques to monitor its performance, it’s still necessary to do more research on techniques for increasing resource utilization and monitoring effects in PAM4 optical network. In this paper, the experimental data of chromatic dispersive (CD) and optical signal to noise ratio (OSNR) under the condition of nonlinear optical fiber is generated by setting relevant parameters in PAM4 optical communication system we constructed. Then the eigenvectors of experimental data are obtained by constructing asynchronous amplitude histograms and we use principal component analysis technique to reduce the dimension of eigenvectors by 20.8%. Finally, the dimensionality reduction results are used as the input of support vector regression algorithm to complete the prediction of CD and OSNR. The prediction errors of CD and OSNR are varied in the range of- 0.02 to 0.02 dB and -0.8 to 0.8 ps/nm respectively. The simulation results show that the proposed method is effective and accurate in monitoring the performance of PAM4 optical communication system and it can also be applied to monitor the performance of PAM-N optical communication system.