AGC EDFA transient suppression algorithm assisted by cognitive neural network

Abstract

This work proposes an EDFA electronic automatic gain control (AGC) scheme assisted by a cognitive neural network algorithm, providing gain control with transient suppression for potentially any EDFA operation point (input/output power condition) within a DWDM dynamic optical network. The idea is to use the neural network to estimate pump power levels and speed up the AGC proportional integral gain controller convergence. For experimental testing, the algorithm was embedded in a microprocessor inside an EDFA module, which was then placed in a fully-loaded reconfigurable DWDM optical link (80 × 112 Gbits/s DP-QPSK channels). By assuming a central point in the controller power mask, results show that gain control is kept below 2 dB for a giving surviving channel, with strong transient suppression during add/drop of 79 out of 80 channels (19 dB input power variation), leading to minimum undershoot/overshoot below 3.1 dB.