Appliance of optimized Bayesian networks for location estimation [mobile radio applications]

Location based mobile applications will yield high profitability in urban areas with their high density of customers. In these types of areas, localization techniques based on runtime measurements often lack performance due to multipath propagation. We present a positioning method especially suited for heavy urban areas based on power-level measurements or any other location dependent parameter of the serving and adjacent cells. The main focus of this paper is on improving the performance of the Bayesian-network-based algorithm by utilizing prior knowledge gained from neighboring positions. Analysis of measurements performed in the inner city of Vienna reveals that utilizing the prior information halves the number of necessary samples and concurrently improves the accuracy.