Evaluation of Real-Life LoRaWAN Localization: Accuracy Dependencies Analysis Based on Outdoor Measurement Datasets

Modern outdoor localization is commonly dependent on various Global Navigation Satellite Systems (GNSSs). On the other hand, they are known to be power-hungry and not suitable for resource-constrained devices currently flooding the Industrial Internet of Things (IIoT). Nonetheless, some of those devices may be equipped with Low-Power Wide-Area Network (LPWAN) communication chip that could be utilized for positioning. Current work examines two outdoor datasets collected using LoRa Wan in Brno, Czech Republic, to assess the possibility of applying technology for localization solutions for industrial outdoor scenarios. The main localization approach applied in this is work is k-NN fingerprinting. For the first dataset gathered over the whole city, the minimal mean localization error turned out to be not stable, while accuracy for the second one covering a small rectangular area 8.5×70 m is 6.42 m that sounds promising in terms of LoRaWAN-based localization. Moreover, by analyzing data collected in two independent measurements campaigns, this work provides some derivations related to the accuracy dependencies on parameters of the measurement campaign (gateways (G W s), coverage area, the average distance between measurement points). It makes a step towards comparing the results of published papers in this area obtained for different datasets.