Enhancing sparse fingerprint using signal interpolation for indoor positioning

Abstract

Fingerprinting technology used for localization can be extensively applied in both indoor and outdoor settings, playing a role in scenarios such as autonomous driving and robotic cruising. However, this method is costly due to the high expenses associated with data collection and the substantial computational power required for processing. This project involves utilizing the well-established UJIIndoorLoc dataset, performing data cleaning and sampling, and enhancing the dataset using Kriging and Inverse Distance Weighting (IDW) interpolation methods. The performances of these methods in positioning scenarios are then compared. Finally, the potential and efficiency of further enhancing the fingerprinting process using machine learning models will be discussed.