Top three intelligent algorithms for OAM mode recognitions in optical communications

Abstract

Vortex optical communication employing orbital angular momentum (OAM) has been a hot research field in recent years.  Thanks to the orthogonality of the OAM, several multiplexing and modulation techniques have been developed that can effectively improve communication capacity. However, to achieve this, accurate mode recognition in the OAM-based free-space optical (FSO) communication system is essential. Generally, perturbations in the free space link significantly affect the transmission efficiency and distort the helical phase-front of OAM beams, which will result in intermodal crosstalk and poses a critical challenge in the recognition of OAM modes. To date, artificial intelligence (AI) technologies have been widely applied to address the aforementioned bottleneck of insufficient accuracy of existing techniques for OAM mode detection. Therefore, a review paper that discusses the recent developments and challenges of the most widely used AI algorithms for OAM mode recognition schemes, i.e., feedforward neural network (FNN), convolutional neural network (CNN), and diffractive deep neural networks (D2NN) is urgently required. By elaborating on the principles of these algorithms and analyzing recent reports, encompassing both experimental and simulated results, we established their profound importance in enhancing the accuracy of OAM mode recognition. Moreover, this work provides an outlook on the recent trends in this newly developed field and the critical challenges faced in effectively using AI for improving the reliability of the OAM-based FSO communication system in near future.