Spatial-Temporal Distribution of Mobile Traffic and Base Station Clustering Based on Urban Function in Cellular Networks

With the rapid development of mobile internet, it’s essential to understand the spatial-temporal distribution of mobile traffic. Based on the mobile traffic data collected from a large 4G cellular network in northwestern China, this paper presents detailed analyses of the traffic data on base stations in two aspects: (1) spatial-temporal distribution, (2) clustering based on physical context, i.e., urban function. We introduce the concept of traffic density to measure the traffic level, according to the Voronoi diagram to partition the covering area of BSs. Both spatial and temporal dimensions show distinct inhomogeneity property of mobile traffic. Furthermore, we cluster BSs utilizing urban function information, which enables us to identify and label base stations. The diverse application usage patterns of each cluster of BSs are obtained, which could be applied in resource cache policy and BS loading allocation.