Recognition of the orbital angular momentum of vortex beams under the transmission errors of a Cassegrain antenna based on deep learning

When vortex beams carrying orbital angular momentum (OAM) propagate through free space in optical communication systems, transmission errors between antennas can cause beam distortion, which presents a challenge for OAM mode recognition. To effectively recognize the OAM of vortex beams that have undergone transmission errors, we propose a solution using convolutional neural networks (CNN). In this study, we simulated and analyzed the transmission characteristics of light using geometrical optics. Through experiments, we obtained actual received images of higher-order radial index fractional-order vortex spots. The images were compiled into a dataset to train the CNN. The training results demonstrate that our network exhibits high recognition capability for OAM of vortex beams under different transmission errors. This approach provides more feasible solutions for free-space optical communication based on the recognition of higher-order vortex beam OAM modes.