Learning Geographic Regions using Location Based Services in Next Generation Networks

In this paper we apply classification to learn geographic regions using Location Based Services (LBS) in Next Generation Networks (NGN). We assume that the information in Local Network (cellular network) can be freely exchanged with Global IP Network (NGN) and the information can be gathered in a database. Location Based Services (LBS) in the IP Multimedia Subsystem (IMS) also provide location information for the data sets. Statistic classification methods are applied to the data sets in the database. We distinguish two cases: a) Learning the geographic regions in which certain events happen. Depending on the information provided by the users, they are divided into different user groups (event classes) using Type Filters (TF). Then discriminant analysis is applied to the position information offered by LBS in IMS to determine the geographic regions of the different classes. b) Learning events that happen inside certain geographic regions. The observed area is divided into different geographic regions (location classes) using Location Filter (LF). Then discriminant analysis is applied to determine patterns of behavior in these regions. The learned geographic regions supporting up-to-date information can be used to establish services for this region or for other regions over NGN. The presented concept can be applied to any scenario with location-based events.