Data-driven Enhancement of the Time-domain First-order Regular Perturbation Model

A normalized batch gradient descent optimizer is proposed to improve the first-order regular perturbation coefficients of Manakov equation, often referred as kernels. The optimization based on linear parameterization offered by and targets enhanced low-complexity models for fiber channel. We demonstrate that optimized model outperforms analytical counterpart where kernels are numerically evaluated via their integral form. provides same accuracy with a reduced number while operating over an extended power range covering both nonlinear highly regimes. $6-7$~dB gain, depending metric used, obtained respect conventional perturbation.