A machine learning approach to evolving an optimal propagation model for last mile connectivity using low altitude platforms

Abstract

This paper develops a machine leaning framework that evolves an optimal propagation model for the last mile with Low Altitude Platforms from existing propagation models. Existing propagation models reviewed exhibit both advantages and shortcomings in relation to a set of factors that affect performance across different terrains, i.e. path loss, elevation angle, altitude, coverage, power consumption, operational frequency, interference, and antenna type. A comparison of the predictions between the optimized and the existing models in relation to above set of factors reveals significant improvements are achieved with the optimal model.