CNN-Based Phase Matching for the OAM Mode Selection in Turbulence Heterodyne Coherent Mitigation Links

Chunyong Yang,Jun Chen,Shaoping Chen,Jin Hou,Kaige Shan

doi:10.1109/jphot.2020.3025944

Chunyong Yang, Jun Chen + Show 3 more

Open Access

https://doi.org/10.1109/jphot.2020.3025944

Copy DOI

Abstract

A novel method is proposed to select vortex beams carrying a specific orbital angular momentum (OAM) mode in turbulence heterodyne coherent mitigation (THCM) link. It is worth mentioning that intelligent phase matching (IPM) of the OAM beams based on the convolutional neural network (CNN) is the remarkable feature. Namely, CNN is particularly trained as the OAM modes classifier by the light intensity distribution patterns of different modes. The classifier actually acts as a mode detector to distinguish OAM modes by the map between the light intensity distribution and OAM mode, and then output mode information (MI). Specially, the phase matching technology is demonstrated to realize selection of specific OAM mode, where exploiting MI to select a specific phase mask is a characteristic of IPM. Subsequently, the phase mask is attached to the Gaussian beam to obtain the OAM beam carrying a special mode. Numerical results show a high IPM accuracy of 99% under medium strength atmospheric turbulence (AT).

Highlights

In recent years, optical orbital angular momentum (OAM) has attracted much attention in the field of free space optical communication (FSO) due to its capacity expansion potential [1]–[3] and [4]
convolutional neural network (CNN) is trained as the OAM modes classifier by the light intensity distribution patterns of different modes
The CNN is trained as the OAM modes classifier by the light intensity distribution patterns of different mode

Summary

Introduction

Optical OAM has attracted much attention in the field of free space optical communication (FSO) due to its capacity expansion potential [1]–[3] and [4]. Researchers have reported a variety of compensation methods to mitigate turbulence distortion of OAM beam, such as channel coding, adaptive optics (AO) and heterodyne coherent [13], [14] Both [15] and [16] demonstrated that the transmission system of low-density parity check (LDPC) -coded OAM based FSO can achieve higher coding gains. In reference [23], the radon-cumulative distribution transform (R-CDT) technique was employed to preprocess OAM intensity maps, and it was combined with a 1-Layer shallow CNN to realize the pattern classification of the OAM beam It was shown in [1] that CNN could be utilized for the detection of atmospheric turbulence and adaptive demodulation of multi-channel OAM beams.

Intelligent Phase Matching

Convolutional Neural Network

Numerical Simulation

Findings

Conclusion