OAM mode recognition based on joint scheme of combining the Gerchberg–Saxton (GS) algorithm and convolutional neural network (CNN)

Abstract

Optical vortex carrying orbital angular momentum (OAM) has attracted a lot of attentions in the field of free-space optical (FSO) communication. Generally, after transmitting in atmospheric turbulence, the helical phase-front of Laguerre-Gaussian (LG) beam carrying OAMs will be severely distorted, thus result in intermodal crosstalk, which is a critical challenge to the effective recognition of the OAM modes. In this paper, we have proposed a joint scheme of combining Gerchberg–Saxton (GS) algorithm and convolutional neural network (CNN) (GS-CNN) to achieve the efficient recognition of the multiplexing LG beams under turbulence environment. The feasibility of the scheme is verified by investigating the recognition performances under various turbulence-strength levels and mode spacings. Moreover, we also discuss the performances of recognition LG beam with different radial indices. The numerical simulation shows that after initial restoration of the beams’ intensity distribution by the GS algorithm, the following CNN can effectively extract the features of retrieved intensity distribution and achieve higher accuracy on recognizing multiplexing LG beams in severe turbulence environment. Besides, the good performance can also be achieved in recognizing LG beams with different radial indices. The results demonstrate great potentials of the GS-CNN for the implementation of actual OAM-based communication system.