Laser Phase Noise Tolerance of Uniform and Probabilistically-shaped QAM Signals for High Spectral Efficiency Systems

TL;DR

This study investigates how laser phase noise affects the performance of probabilistically shaped and uniformly shaped QAM signals in high spectral efficiency optical communication systems through simulations and experiments.

Contribution

It provides a comprehensive comparison of PS and US QAM signals under different laser linewidths and pilot ratios, revealing their respective robustness and limitations.

Findings

Probabilistically shaped QAM performs well with narrow linewidth lasers.

Uniform QAM outperforms PS-QAM with high phase noise and low pilot ratios.

Performance depends on laser linewidth, pilot ratio, and phase recovery scheme.

Abstract

We numerically and experimentally investigate the laser phase noise tolerance of probabilistically shaped (PS) and uniformly shaped (US) quadrature amplitude modulation (QAM) signals. In the simulations, we compare PS-64QAM to US-16QAM, PS-256QAM to US-64QAM, and PS-1024QAM to US-256QAM under the same information rate (IR). We confirm that a sufficient shaping gain is observed with narrow linewidth lasers, whereas degradation of the shaping gain is clearly observed when large phase noise and high order modulation formats are assumed. In our experiments, we compare polarization-division-multiplexed (PDM) 16-GBd PS-1024QAM and US-256QAM under the same IR using lasers with 0.1-kHz and 40-kHz linewidths. For carrier phase recovery (CPR), we employ a pilot-assisted digital phase locked loop. Results reveal that PS-1024QAM achieves high performance with the 0.1 kHz-laser or > 5% pilot ratio, whereas US-256QAM outperforms PS-1024QAM when lasers with 40-kHz linewidth and < 5% pilot ratio are used. We also evaluate the pilot ratio dependency of the required optical signal-to-noise ratio at the forward error correction limit and the achievable information rate. Additionally, we compare the performance of two types of CPR updating schemes: updating phase estimation at only the pilot symbol or at all symbols.