Deep Learning Approach to Determine the Optical Characteristics of Photonic Crystal Fiber for Orbital Angular Momentum Transmission

A Deep Learning model will be a promising method for evaluating the parameters of the photonic crystal fiber for transmitting orbital angular modes, which is a challenge for the finite element method (FEM) owing to the FEM's high processing power and time requirements. This paper evaluates multiple Artificial Neural Network (ANN) models built from scratch to map design parameters and optical characteristics: Effective Refractive Index (Neff), Mode Purity (
$\eta$
), Dispersion(D), Effective Area (A
eff
) & Nonlinearity (
$\gamma$
) of Ring Core Photonic Crystal Fiber (RC-PCF) employed for OAM transmission with notable precision having mean squared error of 0.02815 requiring significantly less time (approximately 4% of the time required in COMSOL Multiphysics), which can be regarded as a considerable substitute to simulative processes in terms of computational resources and complexity associated with the rigorous modeling through trial and error approach conventional numerical simulation techniques.