Fusion of Neural Network positioning and Database Correlation in localizing a Mobile Terminal

Mobile phone location activities have been intensified recently. This work will examine the possibility of combining the Neural Network (NN) positioning and the database correlation (DC) localization system in order to realize a robust RSS based positioning system within a GSM network context. The NN is trained first offline before it use in the positioning unit. Function approximation and classification properties of the NN will be investigated and the best NN architecture will be chosen from a series of candidates. During positioning phase both positioning results from NN and DC are fused with a robust tested method in order to increase the precision of the position estimate. Results show that, the fusion process reduces the positioning error. Keywordsdata fusion, database correlation, localization, neural network, mobile positioning.