K-nearest neighbor model for classification between four different Hermite gaussian beams in MDM/FSO systems under rainy weather

In this paper the K-Nearest Neighbor (KNN) model which is one of machine learning (ML) algorithm is applied to distinguish between the four distinct Hermite Gaussian (HG) beams (HG00, HG01, HG02, and HG03) in mode division multiplexed (MDM)-free space optics (FSO) communication system. First, the performance of the FSO system using these different HG beams is investigated for rainy weather conditions in various Indian cities having distinct geographical locations. Transmission performance is evaluated in terms of quality (Q)-factor, signal-to-noise ratio (SNR), bit error rate (BER), and eye diagrams. Then, the values of the BER for the different HG beams are applied to the KNN ML model as to classify between them. HG modes are used for capacity enhancement. Results indicate that 80 Gbps can be transmitted over 1000 m in Hyderabad, 800 m in Pune, 580 m in Chennai, and 475 m in Mumbai at log (BER) ~ \({10}^{-15}\), Q-factor less than 8, and SNR ~ 10 dB. Additionally, the K-Nearest Neighbor (KNN) ML model shows 94% classification accuracy between four HG beams.