AI-Enhanced VLC/RF Hybrid for Smart IoT: A Revolution

Abstract

This paper presents a comprehensive survey of artificial intelligence (AI) techniques deployed in the context of hybrid systems, illuminating their utility and proposing distinct areas for AI application. The paper further highlights the most recent advancements in this rapidly evolving field. A rigorous performance analysis is conducted on a selection of ensemble learning, namely AdaBoost C4.5, Bagging, Gradient Boosting Machine (GBM), XGBoost, and LightGBM. These models are evaluated on a dataset with varying training sizes to assess their robustness and scalability. In addition to the evaluation, the paper explores the impact of hyperparameter optimization on the performance of these models. The results consistently demonstrate an enhancement in the classifiers' performance commensurate with an increase in training size. Specifically, GBM and AdaBoost C4.5 exhibit significant performance improvements with larger training sizes. In the absence of optimization, GBM achieves the highest accuracy at 96.62%. However, when hyperparameter optimization is employed, both GBM and XGBoost display substantial improvements, reaching an accuracy nearing 98.81%. The paper concludes with a discussion on future research directions related to the application of AI in hybrid systems. The findings from this study provide a robust foundation for further investigations and advancements in this promising and dynamic field.