Joint Estimation of Linear and Nonlinear Coherent Optical Fiber Signal-to-Noise Ratio

Abstract

This letter proposes an estimator based on the neural network (NN) to jointly estimate the linear and nonlinear signal-to-noise ratios. The proposed NN-based estimator utilizes new input features based on the entropy extracted from the received signal. Moreover, the computational complexity of the proposed estimator is analyzed. The dataset utilized for training and testing is constructed from dual-polarization 16-ary quadrature amplitude modulation format over different system configurations of the standard single-mode fiber, such as launch power, transmission distances, and the number of wavelength division multiplexed channels. Numerical results reveal the superiority of the proposed NN-based estimator in terms of accuracy and computational complexity compared to the existing NN-based estimators in the literature.