Spatial Statistical Modeling for Heterogeneous Cellular Networks - An Empirical Study

Abstract

Modeling the spatial distribution of multi-tier base stations (BSs) is an important issue for understanding and validating the analysis and design of heterogeneous cellular networks (HCNs). In this paper, we use different spatial statistical models to describe the spatial distribution of BSs, by fitting real data sets of BS locations in HCNs. Classical statistics like the L function, and cellular network performance metrics like the coverage probability are used to evaluate the goodness-of-fit. The results reveal that notable distinctions exist between modeling HCNs and single-tier networks. Although each tier in a HCN may be most accurately fitted by statistical models other than the Poisson spatial distribution, it is surprising that multiple tiers of independent Poisson spatial distributions provide an accurate description of the overall HCN. The impacts of different network parameters and the dependency between tiers on the modeling accuracy are extensively investigated.