BASISMAP: sequence-based similarity search for geomagnetic positioning

Abstract

Indoor localization has become a popular topic with the development of location-based services (LBS) and indoor navigation systems. Beside these circumstances indoor positioning has been the focus of attention for researchers as the most important component of these applications. Many signals are used as distinguishable features for indoor positioning. RF-based Wi-Fi and BLE systems are the most popular ones and these have been preferred because of their high distinguishable feature. The use of geomagnetism, a natural signal found all over the world, has also been of interest to many researchers. Geomagnetic signals being distorted in the indoor area due to the effect of the structure by using that information takes opportunity to determine the relevant location. In this study, a new method is proposed to convert these unknown signals into location data using a magnetic fingerprint database. The sequential data collected using a dynamic comparison buffer in motion is evaluated with the help of the similarity search method called matrix profile, and position is obtained. The study was compared with other methods in the literature and its prominent and weak points were shared. The performance of the study was evaluated using site-survey by collecting data in an office environment. It has been concluded that the cumulative error is below 2.2 m in the normal operating phase of the system on a 100-m-long path. Compared to the literature, a low complexity and efficient solution is proposed. Furthermore, matrix-profile-based path matching method was used for the first time in magnetic sequence-based localization.