Efficient Recognition of the Propagated Orbital Angular Momentum Modes in Turbulences With the Convolutional Neural Network

Abstract

The vortex beam carrying orbital angular momentum (OAM) has attracted great attentions in optical communication field, which can extend the channel capacity of communication system due to the orthogonality between different OAM modes. Generally, atmospheric turbulence can distort the helical phase fronts of OAM beams, which presents a critical challenge to the effective recognition of OAM modes. Recently, convolutional neural network (CNN), as a model of deep learning, has been widely applied to machine vision. In this paper, based on the CNN theory, we make a tradeoff between the computational complexity of the system and the efficiency of recognition by establishing a specially designed six-layer CNN structure in CPU station to efficiently achieve the recognition of OAM mode in turbulent environment through the feature extraction of the received Laguerre-Gaussian beam's intensity distributions. Furthermore, we examine the performances of our designed CNN with respect to various turbulence levels, transmission distances, mode spacings, and we have also compared the performances of recognizing single OAM mode with multiplexed OAM modes. The numerical simulation shows that basing on CNN method, the coaxial multiplexed OAM modes can obtain higher recognizing accuracy about 96.25% even under long transmission distance with strong turbulence. It is anticipated that the results might be helpful for future implementation of high-capacity OAM-based optical communication technology.

Keywords

• Pattern recognition
• machine vision
• optical vortices
• free-space optical communication
• atmospheric turbulence