Data fusion based on neural network for the mobile subscriber location

The position estimation of a cellular mobile subscriber is a requirement driven by emergency calls, and also by the emergence of new location based services. In order to reach a better accuracy than the one provided by each base station separately, one approach is to fuse the measurements of several base stations like the direction of arrival, the time of arrival,... This paper presents the application of an artificial neural network to fuse radiolocation measurements and confidence of measurements. Based on radiolocation data provided by a CDMA simulator an accuracy of 65 m in 67% of cases has been reached. In order to avoid the use of a neural network fuser specifically dedicated to a cell, it is necessary to generalize the training process. It allows to localize the mobile station in any circumstances. This process allows to have a low cost fuser compatible with FCC requirements.