Machine Learning-Based Classification of OAM Modes in FSO Systems Under Fog Conditions Using SVM and KNN

The rapid advancement of Free Space Optical (FSO) communication presents a compelling alternative to traditional radio frequency methods. FSO technology has garnered significant momentum in recent years, driven by its capacity for highspeed data transmission, enhanced capacity, heightened security, streamlined installation processes, immunity to Electromagnetic Interference (EMI), and utilization of unlicensed spectrum. Despite these advantages, atmospheric turbulence poses a formidable challenge, manifesting in forms such as rain, fog, and haze, which can significantly degrade system performance. One of the primary issues stemming from atmospheric turbulence is fading, which detrimentally impacts FSO communication. Orbital Angular Momentum (OAM) multiplexing is a promising technique to enhance the capacity and performance of FSO communication systems. By leveraging multiple OAM modes, the system can achieve high data rates and improved spectral efficiency, making it suitable for nextgeneration optical networks. However, accurately distinguishing between different Laguerre-Gaussian (LG) modes in an OAM-FSO system is challenging, especially under adverse weather conditions. To address this, machine learning (ML) models, specifically Support Vector Machine (SVM) and K-Nearest Neighbors (KNN), are employed to classify received signals into distinct $L G$ modes. The performance of both models is analyzed, demonstrating that SVM significantly outperforms KNN in terms of classification accuracy and robustness. The findings highlight the potential of ML-based classification in enhancing the reliability of OAM-FSO systems, paving the way for intelligent and adaptive optical communication networks.