Method for an automatic identification of urban dense areas from cell phones records

Abstract

Method for an automatic identification of urban dense areas from cell phones records, by using a computing device that receives as inputs: a geographical region R, a time period Δt for which dense areas in the region R need to be computed, a set of BTSs in the region R, a set of CDRs generated by individuals during the time period Δt using the set of BTSs of the region R, a coverage C and a granularity ε. The method includes constructing a graph G=(V, E), being V=vertexes and E=edges, using Delaunay triangulation, where each vertex vi of V corresponds to btsi of BTS in the geographical region R, and each edge ei,j of E represents connection between btsi and btsj; eliminating from E all the edges in E with a distance between two connecting BTS larger than c, so that a desired spatial granularity is ensured; associating a weight wi,j to each edge ei,j of E that has not been eliminated, the weight representing the average density of the area covered by btsi and btsj during the time period Δt; constructing a data structure L that contains the dense areas using the edges of E; and applying a “Maximum Spanning Tree” type algorithm to detect dense areas given by the data structure L.

Dependencies are inferred from the claim text.