Generating driving trajectories with mobile subscription data for mobile network simulations

In this paper we present a method to estimate the individual mobility behavior of mobile network subscribers. This is very helpful for real network simulations where statistical approaches cannot fulfil the required grade of representation of reality. We focus on the estimation of trajectories on the road network of mobile network subscribers using signaling information of network operators. Basis of our investigations are anonymised signalling data in the mobile network operator's infrastructure. We use these data to estimate the user's position on the road network over time. The estimation of the trajectory's starting and end point is based on the population density distribution in the cell area. Metrics are introduced that help to determine the best trajectory estimation either by timing or geometry validation. The data our investigations are based on are taken from an Austrian mobile network operator captured in November 2010. We calculated a trajectory for a mobile subscriber who was moving along the Austrian road network. The calculated trajectory covered 96% of the actual path the user was driving. We discovered that mobile subscription data can be used to generate possible driving trajectories.