Signal shaping in optical communications—Beyond the Gaussian channel

With the advent of new bandwidth-hungry cloud-based services such as autonomous traffic systems, remote medicine, and the internet of things, higher and higher demands are placed on the digital backbone infrastructure, which is based almost entirely on fiber-optic communication links. This project aims to boost the performance of such links using signal shaping.

Signal shaping means adapting the transmitted signal alphabet to the capacity-achieving distribution for the underlying channel. Shaping theory and algorithms are well developed for linear channels but not for nonlinear. The fiber-optic channel is inherently nonlinear, and furthermore, existing optical transmitters exhibit nonideal, nonlinear characteristics at the ultrahigh data rates where future fiber systems will operate.

To understand the potential shaping gains and tailor shaped transmission schemes to such channels, we will first model the combination of nonlinear fiber and nonideal hardware as a mathematical input-output relation and then develop fundamental shaping theory for nonlinear channels in general and the fiber channel in particular. Using this theory, novel shaping schemes with memory will be designed specifically for fiber links, and their performance will be evaluated experimentally. Finally, the models, theory, and shaping schemes will be extended to spatially multiplexed systems using multicore or multimode fibers, where multidimensional shaping techniques are expected to incur even larger gains.

Start date 01/01/2018
End date 31/12/2021

Published: Fri 12 Jan 2018. Modified: Fri 30 Mar 2018