Efficient channel modeling of structured light in turbulence using generative adversarial networks.

Abstract

We present a fast and efficient simulation method of structured light free space optics (FSO) channel effects from propagation through a turbulent atmosphere. In a system that makes use of multiple higher order modes (structured light), turbulence causes crosstalk between modes. This crosstalk can be described by a channel matrix, which usually requires a complete physical simulation or an experiment. Current simulation techniques based on the phase-screen approximation method are very computationally intensive and are limited by the accuracy of the underlying models. In this work, we propose to circumvent these limitations by using a data-driven approach for the decomposition matrix simulation with a conditional generative adversarial network (CGAN) synthetic simulator.