Noise Models in the Nonlinear Spectral Domain for Optical Fibre Communications

TL;DR

This paper introduces the first analytical noise models for spectral amplitudes in soliton-based optical fiber communications, enabling more efficient encoding and opening new avenues for capacity analysis.

Contribution

It presents the first analytical noise models for spectral amplitudes of N-solitons, expanding the encoding possibilities beyond the imaginary eigenvalue component.

Findings

Developed analytical noise models without Gaussian approximation

Enables encoding information using spectral amplitudes

Facilitates capacity and information theoretic analysis

Abstract

Existing works on building a soliton transmission system only encode information using the imaginary part of the eigenvalue, which fails to make full use of the signal degree-of-freedoms. Motivated by this observation, we make the first step of encoding information using (discrete) spectral amplitudes by proposing analytical noise models for the spectral amplitudes of $N$-solitons ($N\geq 1$). To our best knowledge, this is the first work in building an analytical noise model for spectral amplitudes, which leads to many interesting information theoretic questions, such as channel capacity analysis, and has a potential of increasing the transmission rate. The noise statistics of the spectral amplitude of a soliton are also obtained without the Gaussian approximation.