Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Optimizing Bipolar Constellations for High-Rate Transmission in Short-Reach Fiber Links with Direct Detection

Thomas Wiegart, Daniel Plabst, Norbert Hanik, Gerhard Kramer

TL;DR

This work targets high-rate short-reach fiber links with direct-detection by optimizing bipolar modulation through a modulator bias offset and pairing it with a neural-network equalizer that employs successive interference cancellation (SIC). The NN-based receiver leverages channel memory to approach joint digital detection performance while maintaining manageable complexity, and it demonstrates substantial gains over conventional direct-detection receivers. Key findings show net bit rates exceeding 400 Gbit/s on a 10 km O-Band link (ROP around -15 dBm) and SIC delivering enhancements of over 100 Gbit/s compared to standard SDD, with an optimal offset around c ≈ 0.6 for mid-to-high symbol rates. The results suggest a practical pathway to ultra-high-rate short-reach links using bipolar signaling and NN-SIC processing, with experimental validation identified as future work.

Abstract

Bipolar modulation increases the achievable information rate of communication links with direct-detection receivers. This paper optimizes bipolar transmission with a modulator bias offset for short-reach fiber links. A neural network equalizer with successive interference cancellation is shown to gain over 100 Gbit/s compared to standard receivers.

Optimizing Bipolar Constellations for High-Rate Transmission in Short-Reach Fiber Links with Direct Detection

TL;DR

This work targets high-rate short-reach fiber links with direct-detection by optimizing bipolar modulation through a modulator bias offset and pairing it with a neural-network equalizer that employs successive interference cancellation (SIC). The NN-based receiver leverages channel memory to approach joint digital detection performance while maintaining manageable complexity, and it demonstrates substantial gains over conventional direct-detection receivers. Key findings show net bit rates exceeding 400 Gbit/s on a 10 km O-Band link (ROP around -15 dBm) and SIC delivering enhancements of over 100 Gbit/s compared to standard SDD, with an optimal offset around c ≈ 0.6 for mid-to-high symbol rates. The results suggest a practical pathway to ultra-high-rate short-reach links using bipolar signaling and NN-SIC processing, with experimental validation identified as future work.

Abstract

Bipolar modulation increases the achievable information rate of communication links with direct-detection receivers. This paper optimizes bipolar transmission with a modulator bias offset for short-reach fiber links. A neural network equalizer with successive interference cancellation is shown to gain over 100 Gbit/s compared to standard receivers.
Paper Structure (6 sections, 4 equations, 3 figures)

This paper contains 6 sections, 4 equations, 3 figures.

Table of Contents

  1. Introduction
  2. System Model
  3. Neural Network Equalizer
  4. System Parameters and Offset Optimization
  5. Offset Optimization
  6. Conclusions

Figures (3)

  • Figure 1: System model.
  • Figure 2: Net bit rates $R_b$ (top figures) and AIRs $I_S$ (bottom figures) for different offset values with $8$-ary modulation, transmission over $10km$ of SSMF in the O-Band, ROP of $-15dBm$ and $S=3$ SIC stages.
  • Figure 3: Net bit rates $R_b$ (top figures) and AIRs $I_S$ (bottom figures) for different offset values with $8$-ary modulation, transmission over $10km$ of SSMF in the O-Band, ROP of $-13dBm$ and $S=3$ SIC stages. See Fig. \ref{['fig:offset7dB']} for the legend.