Fingerprint Augmentation Evaluation in LoRaWAN

This document is a research article that investigates the effectiveness of a fingerprint augmentation method called ProxyFAUG in enhancing the positioning accuracy and label efficiency of localization systems within Low Power Wide Area Networks (LPWANs), specifically utilizing datasets from Antwerp LoRaWAN. The study begins with a detailed analysis of existing literature on localization methods and evaluation methodologies, followed by an evaluation of ProxyFAUG's capacity to improve performance using a voluminous training set. Results indicate a significant reduction in median localization error by 19% when the training set is augmented by ProxyFAUG. Furthermore, it demonstrates that the performance achieved with the full training set can be matched using only 40% of the original data. The research replicates ProxyFAUG's initial demonstration in a Sigfox setting, confirming consistent results across different technologies. The work emphasizes the importance of open data and open code in promoting transparency and reproducibility in research.