What is the Best Spatial Distribution to Model Base Station Density? A Deep Dive into Two European Mobile Networks

Abstract

This paper studies the base station (BS) spatial distributions across different scenarios in\ud urban, rural, and coastal zones, based on real BS deployment data sets obtained from two European\ud countries (i.e., Italy and Croatia). Basically, this paper takes into account different representative statistical\ud distributions to characterize the probability density function of the BS spatial density, including Poisson,\ud generalized Pareto, Weibull, lognormal, and -Stable. Based on a thorough comparison with real data sets,\ud our results clearly assess that the -Stable distribution is the most accurate one among the other candidates\ud in urban scenarios. This nding is conrmed across different sample area sizes, operators, and cellular\ud technologies (GSM/UMTS/LTE). On the other hand, the lognormal and Weibull distributions tend to t\ud better the real ones in rural and coastal scenarios. We believe that the results of this paper can be exploited\ud to derive fruitful guidelines for BS deployment in a cellular network design, providing various networ

Metrics and barycenters for point pattern data

“…From biomedical imagery over geo-referrenced disease cases and positions of mobile phone users to climate change related space-time events, such as landslides, we have more and more complicated data available. See Samartsidis et al (2019), Konstantinoudis et al (2019), Chiaraviglio et al (2016), Lombardo et al (2018) for individual examples and the textbooks Diggle (2013), Baddeley et al (2015Baddeley et al ( ), B laszczyszyn et al (2018 for a broad overview of further applications. While a few decades ago data consisted typically of a single point pattern in a low dimensional Euclidean space, maybe with some low-dimensional mark information, we have nowadays often multiple observations of point patterns available that may live on more complicated spaces, e.g.…”

Cellular Network Traces Towards 5G: Usage, Analysis and Generation

“…Among the most recent ones, [22] fits log-normal distributions to both the deployment (in space) and the demand (in time) of a mobile network in China. In a similar spirit, the authors of [23] study which, among several distributions, best matches the space patterns of real-world cellular networks in Italy. Both [22] and [23] employ different models for different types of areas, e.g., rural and urban ones; furthermore, [22] also separately models on-and offpeak hours.…”

“…Most existing works [22], [23] solve this problem by fitting a distribution to the observed B(t) values, and then extracting samples from it. However, as discussed earlier, such samples will be i.i.d., i.e., the number of BSs in a tile would be independent of the number of BSs in neighboring tiles around it.…”

New Trends in Stochastic Geometry for Wireless Networks: A Tutorial and Survey

“…Hence, realistic deployments have commonly an increasing tendency towards clustering in user hotspots (e.g., events, urban area) and a tendency towards repulsion and regularity when users are equally likely scattered [92]- [94]. In this way, since the received SINR is sensitive to the interaction degree between nodes location, capturing the geometry of such nodes through an appropriate PP will directly impact the accuracy of network performance evaluation [45], [69], [92]- [96].…”