Wi-Fi fingerprint and pedestrian dead reckoning-based indoor localization with supervised learning

Abstract

The Global Positioning System is not suitable for indoor localization due to signal loss in enclosed environments. Hence, this research designed and developed a hybrid indoor localization approach by integrating the Wi-Fi fingerprinting approach with the pedestrian dead reckoning. Limitations of the Wi-Fi fingerprinting-based localization are compensated by the pedestrian dead reckoning approach which is implemented on the mobile platform. Supervised learning models such as artificial neural networks and K- nearest neighbors have been used to map offline and online datasets. The predicted locations obtained through the hybrid localization approach are compared with the true locations via Euclidean and Manhattan distance calculations. The results prove that the Wi-Fi Fingerprinting and Pedestrian Dead Reckoning together has given promising results for localization than using them alone.