Neural Probabilistic Amplitude Shaping for Nonlinear Fiber Channels

2026-02-02

AI Summary

This work addresses the performance limitations of high-order modulation signals in nonlinear fiber channels by proposing a neural probabilistic amplitude shaping framework, which introduces neural probabilistic modeling into amplitude shaping design for optical communications for the first time. The method leverages deep learning to jointly optimize the amplitude distribution of transmitted signals and the characteristics of the nonlinear channel, enabling end-to-end learning of dual-polarization 64-QAM signal distributions. Experimental results over a 205-km single-span coherent optical link demonstrate that the proposed approach achieves a 0.5 dB signal-to-noise ratio gain compared to conventional sequence selection methods, significantly enhancing transmission performance.

Technology Category

• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP
• Machine Learning: Optimization

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Social networks and social learning
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

We introduce neural probabilistic amplitude shaping, a joint-distribution learning framework for coherent fiber systems. The proposed scheme provides a 0.5 dB signal-to-noise ratio gain over sequence selection for dual-polarized 64-QAM transmission across a single-span 205 km link.

Problem

• nonlinear fiber channels
• amplitude shaping
• coherent fiber systems
• 64-QAM transmission
• signal-to-noise ratio

Innovation

• neural probabilistic amplitude shaping
• joint-distribution learning
• nonlinear fiber channels
• coherent optical communication
• 64-QAM