Abstract
Transmitting a vortex beam (VB) with orbital angular momentum encounters significant aberrations due to atmospheric turbulence (AT) in free-space optical communication systems. Historically, research has focused on adaptive optics (AO) to correct distorted VB intensity distributions by generating phase compensation screens. In this paper, we introduce an image-translation-based machine learning (ML) framework using generative networks to directly produce corrected VB intensity profiles, thus mitigating AT-induced aberrations. Our network was trained with numerical simulation datasets to learn the mapping relationship between distorted and pristine intensity distributions. Subsequently, we evaluated its generalization capabilities using datasets with varied AT conditions and propagation distances. The trained network consistently generated corrected intensity profiles closely matching the desired outcomes, significantly reducing the required training data. Our methodology effectively compensates for distorted VB intensity, achieving a remarkable 40 dB enhancement compared to the distorted profiles.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability
The data are not publicly available but can be accessed by reasonable request.