Efficient Quality Factor Prediction of Artificial Neural Network Based IsOWC System

This paper presents a novel approach utilizing an artificial neural network (ANN) for optical wireless communication (OWC) between satellites in geosynchronous earth orbit and lower earth orbit, covering a distance of 45000 km. The objective of this ANN based intersatellite optical wireless communication (IsOWC) system is to intelligently predict the quality factor considering different wavelengths. To enhance the transmission performance between these satellite systems, the mean squared error (MSE) is minimized using the Levenberg–Marquardt optimizer. Remarkably, after 25 epochs, the MSE value reaches an impressive 0.000373. The results demonstrate that the ANN-based learning outperforms other machine learning algorithms, exhibiting a significantly lower MSE. Furthermore, this system has a high convergence rate as well as resistant to outliers and overfitting. Even if the number of features is small, it can be predicted accurately. Such systems hold great promise for future wireless designs and integrations, spanning from satellite to terrestrial and underwater OWC systems.