Path Loss Modeling for Flying Ad-Hoc Networks:An Ensemble Learning Approach

Several Machine Learning (ML) approaches have been proposed for path loss (PL) prediction in emerging fifth generation and beyond (5G/B5G) cellular communication networks. In the domain of next-generation high-availability communications, flying ad hoc networks (FANETs), which function as clusters of deployable relays or base stations for the on-demand extension of cellular coverage, represent a promising alternative. Thus, the highly accurate path prediction is crucial for the FANETs deployment and for emerging B5G wireless networks design, planning, and optimization. In this paper, we model the path loss in an urban environment for different heights using different ensemble learning approaches. More specifically, we apply five different boosting approaches are, as well a new ensemble model based on a stacked generalization technique for more robust and accurate predictions. The proposed ensemble approaches exhibit great accuracy and efficiency in predicting the path loss for FANETs deployment in cellular networks.