Machine Learning approach for predicting suitable wavelengths in OFDM-FSO system

Free Space Optics (FSO) is one of the branches of optical electronics which integrates two kinds of science: electronics and light. It is one of the modern technologies of communications that transmits signals from one place to another through atmosphere. This research studies the effect of different laser wavelengths (1550nm,1250nm,850nm) in the OFDM-FSO system on distance transmission. Empirical study has been done in different weather conditions (Fog, Rain, Clear…). The tool used for conducting the experiments is by Optisystem v.15 software. The obtained results show that the longer the wavelength used, the further the distance can be reached. Therefore, when using 1550nm, the signal arrives farther than the distance using 1250nm and 850nm. Machine Learning Algorithms (MLAs) have been applied on dataset collected using Optisystem v.15 for the purpose of predicting the best wavelength to be used in OFDM-FSO systems. The empirical results in this work proved that the Random Forest (RF) algorithm gives the best prediction accuracy in comparison with Linear Regression(LR) algorithm.