Machine Learning for Ultrawide Bandwidth Amplifier Configuration

Abstract

Ultrawide bandwidth optical amplifier designs based on multiple pump wavelengths additionally benefit from gain control per sub-bands, at the cost of an increased configuration complexity. In such systems it becomes possible to adapt system parameters such as spectral tilt and optical output power per amplifier, in response to changing conditions in the network, such as wavelength re-configuration and repairs. Machine learning gives a high accuracy performance in configuring and monitoring amplifiers easing the complexity incurred by human-effort.

Identifiers

DOI

10.1109/icton.2019.8840453

Citations and references