Experimental realization of optical communication link with intensity degenerate orbital angular momentum beams

Abstract

This study introduces a significant advancement in optical communication by exploring the experimental astigmatic speckle patterns generated from orbital angular momentum (OAM) beams. Seemingly identical speckle patterns, due to their symmetry, pose challenges for classification by the convolutional neural network (CNN). By introducing astigmatism with a cylindrical lens, this symmetry is broken, enabling these speckle patterns to be effectively classified. In this study, a 3-layer customized CNN with 14.5 million learnable parameters is employed for the classification of the OAM speckle images. The study conducts two separate experimental investigations on the 1f and 2f Fourier transform (FT) systems, revealing that the 1f FT system achieves a classification accuracy of 95% and above, while the 2f FT system yields an accuracy of over 89%. These achievements facilitate the establishment of a robust communication link, enabling effective encoding and decoding of both image and alphanumeric data. The findings underscore the potential of utilizing speckle patterns from OAM beams to significantly enhance the capacity and security of optical communication.