Spatial-Temporal Analysis of Erlang Measurement in Large-Scale GSM Cellular Networks

In this paper, we conduct a detailed analysis on voice traffic intensity measurement of large-scale mobile communication base stations. Our data set tracks over 700 base stations inside one city in Southern China over a continuous period of 18 days. We cluster base stations with k-means method based on their average traffic and the result shows four distinctive clusters, among which one cluster consisting of base stations near university campuses reveals unique calling behavior of college students. The spatial characteristics of base station traffic from global and local perspectives demonstrates that the highest global spatial correlation is around 6 AM and local non-stationarity is identified. Our insights of spatial-temporal traffic dynamics in mobile network can be leveraged into network monitoring and resource management.