A novel tool, based on Bayesian filtering framework and expectation maximization algorithm, is numerically and experimentally demonstrated for accurate frequency comb noise characterization. The tool is statistically optimum in a mean-square-error-sense, works at wide range of SNRs and offers more accurate noise estimation compared to conventional methods.
@article{arxiv.1904.11951,
title = {Optical Frequency Comb Noise Characterization Using Machine Learning},
author = {Giovanni Brajato and Lars Lundberg and Victor Torres-Company and Darko Zibar},
journal= {arXiv preprint arXiv:1904.11951},
year = {2019}
}