부산대학교 도서관

All Traditional geomagnetic fingerprint location is characterized by serial algorithm, which is realized through pair-wise comparison in in-door Geomagnetic big data which is generated by earth's magnetic field. The system is bottle-necked when processing massive smart phone location data. In this paper, a new Hadoop-MapReduce based method is put forward, which can be seen as a parallel algorithm. Through introducing more computers, the new method can greatly enhance the performance of in-door geomagnetic localization. In the meantime, a dynamic spatial-temporal window is used to promote the detecting accuracy.