Channel Classification for Free Space Optical Communication Network based on Machine Learning Techniques

Hamouda, Yousef E. M.; Aljuaidi, Ahmed A.; Younis, Hassan A.

doi:10.21608/mjeer.2025.322921.1095

	Channel Classification for Free Space Optical Communication Network based on Machine Learning Techniques
Menoufia Journal of Electronic Engineering Research
Volume 34, Issue 1, January 2025, Page 40-53 PDF (1.03 MB)
Document Type: Original Article
DOI: 10.21608/mjeer.2025.322921.1095
View on SCiNiTO
Authors
Yousef E. M. Hamouda ¹; Ahmed A. Aljuaidi²; Hassan A. Younis²
¹Al Aqsa University
²MSc. Student
Abstract
Free Space Optical (FSO) communication is an optical communications technology that uses unguided light propagation in free space. Comparing with other wireless network systems, FSO links have high bandwidth, free licensed frequencies, high security, and low transmitted power. Nevertheless, FSO wireless channel suffers from different atmospheric attenuation factors. In this paper, Free Space Optical Channel Classification (FSO-CC) approach is presented to predict the suitability of FSO link. The FSO-CC uses machine learning classifiers to predict the FSO channel status. The classifiers decision depends on the satisfaction of the FSO communication performance such as the minimum received signal to noise ratio, and the transmitter capabilities such as the maximum transmitted power. The features inputs of the proposed classifiers are the distance between the FSO transmitter and receiver nodes, and the current weather conditions. The label output of the proposed classifiers is the channel suitability class. The simulation environment of the proposed FSO-CC scheme is implemented using a real dataset of weather conditions, and actually FSO location nodes. The simulation results show that the proposed FSO-CC efficiently estimates the suitability of FSO wireless channel to use FSO communication or other kinds of communication systems, according to the current weather conditions and distance between the FSO transmitter and receiver nodes. Furthermore, the decision-making feature in the proposed FSO-CC are dynamically controlled. Compared with logistic regression, decision tree, and random forest classifiers, the support vector machine classifier gives better performance in terms of prediction errors, F1 score, precision score, accuracy, score, and recall score.
Keywords
Free Space Optical; Machine Learning; Channel Estimation; Weather Conditions; Classification


Statistics Article View: 159 PDF Download: 178