Introducing a decision tree-based indoor positioning technique

Positioning a user is an essential ingredient of a location-based system. For the outdoor positioning, GPS is practically used. For the indoor positioning, Active Badge, BAT, Cricket, and so on have been introduced. These methods are very accurate but require special equipments dedicated for positioning. Instead of using special equipments, using existing equipments is more economical. For this reason, positioning methods of using existing wireless LAN access points have recently been introduced. Among the methods employed by them, the fingerprint methods are the most promising. Probabilistic method, K-NN (Nearest Neighbor), and Neural networks are the techniques used by the most location fingerprinting. We are proposing a new technique which is more efficient than these three. Our technique builds a decision tree during the off-line phase and determines a user’s location referring to the tree. Time complexity analysis and experimental accuracy analysis of the proposed technique are presented in this paper.