Application of Machine Learning in Urban Base Station Placement for 5G Communications and Beyond
Irfan Farhan Mohamad Rafie
,
Soo Yong Lim
and
Michael Jenn Hwan Chung
Optimal placement of wireless base stations in urban areas allows for maximum coverage and performance whilst maintaining minimal cost. In this paper, we propose a novel machine learning approach to place base stations rapidly in an urban environment for 5G communications and beyond. This is a noteworthy approach as 5G, especially those that involve millimeter wave frequencies tend to require significantly higher number of base stations for any particular area, unlike their counterpart low frequencies where a small number of base station is sufficient to cover a good geographical area. Our machine learning empowered path loss model is developed to tackle this change in gameplay head-on, and it bridges the gap between empirical and ray tracing methods where we achieve accuracy closer to ray tracing yet at a significantly lower computation cost. Promising preliminary results are obtained, with a minimum coverage area of 80% with potential for future improvements.
