FPGA Implementation of Complex Value-based Clustering Filter for Chromatic Dispersion Compensation in Coherent Metro Links with Ultra-low Power Consumption

TL;DR

This paper presents an FPGA-based implementation of a machine learning-assisted chromatic dispersion filter that significantly reduces power consumption in coherent metro links, outperforming traditional FFT-based methods.

Contribution

It introduces a novel FPGA implementation of a machine learning-assisted filter for chromatic dispersion compensation, achieving ultra-low power consumption.

Findings

Energy efficiency gain of up to 63.5% on FPGA

Superior power efficiency compared to FFT-based filters

Validated on FPGA with experimental results

Abstract

This paper introduces a new machine learning-assisted chromatic dispersion compensation filter, demonstrating its superior power efficiency compared to conventional FFT-based filters for metro link distances. Validations on FPGA confirmed an energy efficiency gain of up to 63.5\% compared to the standard frequency-domain chromatic dispersion equalizer.