Modeling extended L-band fiber amplifiers using neural networks trained on experimental data.

Producing high performance amplifiers requires accurate numerical models. As the optimization space is large, computationally efficient models are of great value. Parameter-based models for L-band amplifiers have accuracy limited by difficulty in estimating the Giles-parameter. The use a neural network model can avoid parametrization. We exploit a rich, experimentally captured training set to achieve a high accuracy neural network model. Our approach creates independent models for gain and noise figure. We examine both core and cladding pumping methods, again with independent models for each. The neural networks outperform parameter-based models with higher accuracy (variance of error reduced by 50%) and extremely fast simulation times (400 times faster), greatly facilitating amplifier design. As an example application, we design an amplifier to optimize optical signal-to-noise ratio by exhaustive search with our fast neural network models.

Optimization criteria and design of few-mode Erbium-doped fibers for cladding-pumped amplifiers.

We propose a novel optimization method that combines two design criteria to reduce the differential modal gain (DMG) in few-mode cladding-pumped erbium-doped fiber amplifiers (FM-EDFAs). In addition to the standard criterion that considers the mode intensity and dopant profile overlap, we introduce a second criterion that ensures that all doped regions have the same saturation behavior. With these two criteria, we define a figure-of-merit (FOM) that allows the design of FM-EDFAs with low DMG without high computational cost. We illustrate this method with the design of six-mode erbium-doped fibers (EDFs) for amplification over the C-Band targeting designs that are compatible with standard fabrication processes. The fibers have either a step-index or a staircase refractive index profile (RIP), with two ring-shaped erbium-doped regions in the core. With a staircase RIP, a fiber length of 29 m and 20 W of pump power injected in the cladding, our best design leads to a minimum gain of 22.6 dB while maintaining a DMGmax under 0.18 dB. We further show that the FOM optimization achieves a robust design with low DMG over a wide range of variations in signal power, pump power and fiber length.