Fingerprint DB generating system exploiting PDR based dynamic collection for indoor localization of smart-phone users

The fingerprinting has been considered as a promising method for indoor localization due to the robustness from the multi-path than other methods exploiting wireless signals. However, the fingerprinting method has serious limitation which it requires too much time and cost to construct a DB (Data Base) before positioning. In conventional static collection based fingerprint DB generation system, it usually takes up to several minutes to obtain a radio pattern for a single reference point comprising very small part of whole service areas. To overcome the drawback, the fingerprint DB generation system based on the dynamic collection, which breaks the tradition of the radio pattern collection, is proposed. In contrast to the conventional collection method, a collector carrying a smart-phone moves pre-designated routes in service areas, and the position and the radio patterns are autonomously gathered by the collecting application in the smart-phone at every seconds. With the dynamically collected data, then, the fingerprint DB is finally generated by averaging method. From the experimental results, it is validated that the proposed collection method remarkably improves the efficiency of the collection. Also, the positioning using the generated DB shows reasonable accuracy, 2.62m, in an indoor office environment.