System Performance Prediction With the Gaussian Noise Model in 100G PDM-QPSK Coherent Optical Networks

We demonstrate that the transmission BER, OSNR penalty, and system margin can be accurately predicted for multiple fiber types using the back-to-back response together with the Gaussian Noise model of nonlinear penalties. We first experimentally quantify the 1600 km link performance of SMF, MDF, and LAF fiber types in a coherent, WDM PDM-QPSK system at both 28 and 32 GBaud employing all-EDFA amplification and nearly identical span lengths to isolate fiber performance effects. We quantify the BER, OSNR transmission penalty, and link margin versus per-channel launch power in both linear and nonlinear transmission regimes. We demonstrate that the total system performance can be directly and accurately predicted using the fiber parameters α, D, and γ, the number spans and noise figures, and back-to-back performance of the transmitter-receiver pair. We also show that the system margins scale as (α |D|/γ
²
)
^1/3
as predicted by the Gaussian Noise model of nonlinear penalties in uncompensated systems.