Indoor Positioning Using Periodical Analysis

LBS (Location Based Service) has been more and more integrated into the various industries, simultaneously, the development of the industry further promote the intensive research and widespread application of LBS. In this paper, we present an excellent resolution for indoor location estimation based on local geomagnetic features. Multi-stage periodical analysis can sufficiently mine the different features for high precision. Initial stage: Kalman filter was used to smoothing the fluctuation of magnetic measurements; Second stage: the filtered data were normalized to construct mean generation matrix of periodical extension, which was the basis for multiple stepwise regression modeling; Third stage: location estimation was finally speculated by KNN (K-nearest neighbor) method in view of the local magnetic characteristics. The magnetic measurements (X,Y,Z) over 82 days, but only the data of 46 days were applied in this paper, in entire implementing processes were from an office environment in Institute of Computing Technology Chinese Academy of Sciences. Our results showed that multi-stage analysis can efficiently estimate the indoor location with an overall accuracy of 83%, completely incorporated the superiority of smart-phone and natural resources, and can be conveniently extended to the relative LBS fields. KEYWORD: Indoor Positioning; Magnetic Field; Periodical Analysis; Multiple Stepwise Regression