Optical Fiber Multi-Parameter Measurement Based on Machine Learning

This paper proposes a method to extract the parameters to be measured from the incomplete information of the signal by machine learning. Instead of the data containing all the pulse amplitude and phase information, the method employs the power spectrum amplitude data containing only part of the signal information for parameter extraction. It overcomes the difficulty in measuring the phase information of complex optical signals. Simulations verify the ability to utilize machine learning algorithms to extract the parameter information of transmission medium from pulse evolution and the feasibility of using the power spectrum of pulse without phase information to realize optical fiber multi-parameter measurement. The simulation results show that the mean square error of this method can be controlled below 0.3% with proper machine learning algorithms.