Geometric Constellation Shaping with Low-complexity Demappers for Wiener Phase-noise Channels

TL;DR

This paper demonstrates that separating in-phase and quadrature components in machine-learning based demappers for optical systems with geometric constellation shaping reduces complexity without sacrificing performance.

Contribution

It introduces a low-complexity demapper design that maintains high performance in Wiener phase-noise channels for optical communications.

Findings

Reduced computational complexity in demappers.

Maintained performance with separated in-phase and quadrature processing.

Applicable to optical communication systems with geometric constellation shaping.

Abstract

We show that separating the in-phase and quadrature component in optimized, machine-learning based demappers of optical communications systems with geometric constellation shaping reduces the required computational complexity whilst retaining their good performance.