Machine Learning Techniques In Wdm-Fso Systems: Comparative Study

Abstract

Free space optical (FSO) communication has recently received much attention from researchers due to its great advantages. FSO offers large bandwidth, ease of installation and deployment, security compared to other wireless communications, and FSO systems do not require spectrum license compared to radio frequency systems, which makes FSO an important and advanced communication system. However, FSO systems often have drawbacks. One of the disadvantages of FSO communication is that it requires line of sight (LOS) between the transmitter and the receiver. And this technique (FSO) requires clear, turbulence-free weather conditions in the transmission channel to ensure successful transmission and correct information delivery. Artificial intelligence techniques have been widely introduced into optical communication systems in recent years, especially in the performance prediction of these systems. This paper presents experiments on the design and simulation of an FSO system with wavelength division multiplexing modulation (WDM) under rainy, foggy and snowy climatic conditions and calculates the distances at which the signal can correctly reach the receiver using the Optisystem simulator. Furthermore, machine learning algorithms were used to predict the quality factor of the proposed system and then the performance metrics R2 and RMSE (Root Mean Square Error) were used to compare between the algorithms. The obtained results show that the Random Forest (RF) algorithm gave the lowest RMSE value and the highest R2 value in comparison with the Decision Tree (DT) and K-Nearest Neighbors algorithm (KNN) algorithms. Therefore, we can say that the RF algorithm gave better results in the predicting the accuracy of the quality factor in the WDM-FSO system.

Keywords

• machine learning
• fso
• wdm
• optisystem