Improving LoRaWAN RSSI-Based Localization in Harsh Environments: The Harbor Use Case

Recently, LoRaWAN communications have become a widely used technology within IoT ecosystems due to their long-range coverage, low-cost, and native RSSI-based location capabilities. However, RSSI-based localization has low accuracy due to interference in propagation, such as the multipath and fading phenomena, becoming more critical in harsh and dynamic environments like airports or harbors. A harbor has a wide area with a combination of distinct landscapes (sea, river, urban areas, etc.), and distinct infrastructures (buildings, large steel structures, etc.). In this paper, we evaluate and present a harbor assets localization system that uses a LoRaWAN-based multi-slope path-loss modeling approach. For this purpose, a harbor scale LoRaWAN testbed, composed of three gateways (GWs) and a mobile end node, has been deployed and used to characterize RSSI-based multi-slope path-loss modeling under realistic conditions. Experimental data have been collected over three days in different dynamic scenarios and used for ranging and location estimation comparison using distinct methods, i.e., RSSI-based and fingerprinting. Based on the achieved results, by a correct partitioning of the test environment based on its specific environmental conditions, a decrease between 4 and 10 dBm in path-loss estimation error can be achieved. This path-loss estimation error decrements provide 50% improvement in distance estimation accuracy.