Is LoRaWAN Really Wide? Fine-grained LoRa Link-level Measurement in An Urban Environment

Internet-of-Things (IoT) aims to connect billions of low-date rate and energy-constrained end-devices in the near future. Although many IoT systems have been commercialized, most of them focus on home and body scale applications. To establish a low-cost IoT at the city scale, LoRa Wide Area Networks (LoRaWAN) have become attractive in recent years due to their desirable kilometer or even longer communication distance with low energy consumption. However, due to the expensive cost of densely deploying end-nodes, the understanding of LoRa link behavior is still coarse-grained, and hard to fully realize the link dynamics, networking coverage, and localization accuracy of LoRaWAN in an urban environment. This paper shows a fine-grained LoRa link-level measurement via mobile end-nodes. We deploy two gateways and six mobile end-nodes and collect data packets over four months at a $6\times 6\ km^{2}$ urban area. The evaluation mainly focuses on answering three questions: 1) Does a LoRa link stably perform in both spatial and temporal dimensions? 2) How large area can be covered for reliable communication by each gateway in the urban environment? 3) What accuracy can be achieved to localize an end-node through LoRa links? According to our measurement, our key findings are 1) The spatial and temporal behavior of LoRa links is quite dynamic due to the different types of land covers and the frequent micro-environment changes in the urban areas; 2) Each gateway can cover about 11.3 km ² area and marginal SNR gains (e.g., 2 dB) of LoRa links are efficient enough to enlarge 32.6% coverage area of a gateway; and 3). The median localization error is about 400 m. Without densely deployed LoRa gateways, the SOTA LoRa localization can support road-level localization, even when an end node is close to one of the gateways.