Towards Industry 5.0 Positioning with k-NN-based Enhanced LoRaWAN Localization

Abstract

The usability of Long Range Wide Area Network (LoRaWAN) for localization remained a white space on the navigation map of modern Industry 5.0 wireless technology. Many traditional approaches proved to be not usable for meaningful accuracy. Nonetheless, this paper serves as a culmination of our investigation into the potential of LoRaWAN for localization, synthesizing our key discoveries and advancement. Besides that, the paper brings two key contributions. First, it introduces two new underground LoRaWAN datasets, serving as assets for researchers and practitioners in the field. Secondly, it proposes two innovative k–nearest neighbors (k–NN)-based algorithms aimed at enhancing position estimation accuracy by optimizing nearest neighbor selection. The integration of the preprocessing strategies with the proposed algorithms leads to accuracy improvement of 17.2%, underscoring the potential of LoRaWAN-based localization in sectors where high precision is not paramount.