学术文献库

In the recent context of Software Defined Optical Network, the fast and accurate Quality of Transmission (QoT) estimation of the transmission link is essential. Gaussian Noise models are shown to yield a fast estimation of the average QoT derived from deterministic system parameters, but do not capture the QoT variability. In order to assess numerically the stochastic joint effect of Polarization Mode Dispersion (PMD) and Kerr nonlinearities, system designers generally use the Split Step Fourier Method (SSFM) based on Manakov-PMD equation neglecting the nonlinear-PMD term which is faster than using Coupled NonLinear Schrodinger Equation (CNLSE) and enough accurate for fiber with short birefringence correlation length (around less than 10m). In this work, we present insights of the way to tune the parameters of this Manakov-PMD method and its limitation when seeking an accurate estimation of the Non-Linear Interference (NLI) noise statistical distribution for all fibers potentially installed in the current optical network. In particular we compare this Manakov-PMD method results with respect to the one obtained by CNLSE while varying the fiber birefringence correlation length, PMD coefficient and the fiber type. Our results highlight a potential discrepancy of 0.5 dB in the estimation of the Q^2 factor in one span Polarization Division Multiplexed Quadrature Phase Shift Keying (PDM-QPSK) transmission with optimal launch power per channel and yield guidelines to choose the most suitable numerical estimation method.