Enhanced cell ID based terminal location for urban area location based applications

Mobile location based applications are in the first place designed for urban areas where providers find a high density of customers. Exactly in these environments, conventional trilateration location techniques often lack performance due to multipath propagation. We present results from a localization related usability study with a fully implemented mobile tourist guide, called LoL@ (local location assistant) in the City of Vienna. We show that GPS or cell ID based localization techniques are not sufficient to achieve user satisfaction. Based on these results, we introduce a new localization approach utilizing Bayesian networks to improve localization accuracy for mobile location based services in urban areas. The method uses power level measurements at the mobile station and can be generalized to use any location dependent parameters. The achieved accuracy, based on measurements in the City of Vienna, is less than 20 meters in 67% of all estimates, and about 50 meters in 90% of all estimates. This is adequate for most location applications, including emergency services.