Simple and efficient ANN model proposed for the temperature dependence of EDFA gain based on experimental results

Abstract

This study deals with the Artificial Neural Network (ANN) model of erbium-doped fiber amplifier (EDFA) gain in C band based on our experimental results at the temperature range of 0–60°C. An ANN with three inputs and one output is considered where the inputs are signal power, wavelength, temperature and the output is EDFA gain. The network parameters are optimized by monitoring mean square error (MSE) at the output. The proposed dynamic model tremendously reduces the computational in the order of milliseconds which computes the EDFA gain at different operating conditions and is in very good agreement with our experimental findings.