GIS-based mobile coverage estimation to reduce uncertainty of cellular trajectories

Abstract

Mobile positioning data are normally generated as scattered points in CDRs (Call Detail Records), which include the IDs of connected cell towers for call activities. To interpret mobile associations with towers, one needs to model the area where the mobile phone user is likely to be when connected. Voronoi cells, which define the area closest to each tower, are usually employed to this end. However, the accuracy of Voronoi tessellation is influenced by various environmental factors due to propagation of signals in different manners, such as diffraction, scattering and reflection. This non-specificity of cell shapes causes a major source of uncertainty in, say, trajectory reconstruction. This paper evaluates the effect of imprecision of cell shape on accuracy of CDR-based trajectory reconstruction. It further aims to reconstruct more accurate trajectories from the coarse CDR data through considering the wave propagation models in estimating the location of mobile phone users. Since many of the factors in radio wave propagation are geographical, GIS techniques will be incorporated into this procedure.