A New Approach to Mobile Position Tracking

Abstract — Recently, there has been much interest in accurate determination of mobile user locations in cellular environments. A general approach to this geolocation problem is to gather measurements from a number of base stations and to estimate user locations using a least squares approach [1]. However, in non-line-of-sight (NLOS) situations, measurements are signifi-cantly biased. Hence, very large errors in location estimation may be introduced. In this paper, an approach developed in learning theory, namely, Support Vector Regression (SVR), is applied to the geolocation problem with the addition of Kalman-Bucy filtering to smooth location estimates in a mobile tracking scenario. I.