OHTLoc: an online heterogeneous transfer method on wifi-based indoor localization system: work-in-progress

Abstract

With the development of wireless network technology, the WiFi-based indoor localization methods incorporating machine learning have attracted wide attention due to its easy deployment and low cost characteristics. However, the existing learning methods are limited to locating homogeneous and tagged target data. Such strict conditions do not exist in the actual indoor positioning environment, and therefore cannot meet people's locational needs. In this article, we design an Online Heterogeneous Transfer method in Indoor Localization(OHTLoc), a novel transfer learning approach that can realize online location prediction based on the RSS(Received Signal Strength) fingerprint and CSI(Channel State Information) data using WLANs. In particular, OHTLoc does not require any tags on the target data. This is the first time this type of algorithm has been proposed in the field of indoor localization. The prediction results of the target demonstrate showed in the experiment part demonstrate the effectiveness of the proposed technique.