Exploring Spatial Relationships and Identifying Influential Nodes in Cellular Networks

Abstract

Abstract-This work provides an up to date measurement-driven examination of the spatial characteristics of network resource usage. The data set used is from a large nationwide 3G cellular network comprised of several thousand base stations. Firstly, we discuss our data set and its potential application. Next, we examine the spatial correlation between base stations in terms of radio resource usage. We find significant spatial correlation, particularly for proximate base stations. We examine the causality structure in the network using Granger causality to identify key influential indicator base stations within sub-networks. These indicator base stations act as hubs in the wider network and provide additional information about the future states of their neighbors. The penultimate section examines the influential indicator base stations in more detail. Finally, we conclude with a brief discussion of the key points and how we aim to progress this work.

Predictive modelling of cellular load

“…As operators move to add capacity, a detailed understanding of the underlying dynamics of resource usage is increasingly important. To this end, some recent works have begun to make use of large scale data sets provided by network operators to identify important facets of network usage [3][4][5][6][7][8][9]. Understanding traffic patterns and predicting load in individual cells and groups of cells is becoming ever more important with the emergence of Self-Organising Networks (SON).…”