How To Conduct LoRaWAN Site Surveys

LoRa/LoRaWAN enables long-range links, which result in anisotropic propagation around the gateway as the links are likely to experience different environments. This property is not captured by popular distance based channel models, which makes them inaccurate to estimate LoRaWAN coverage. An alternative is conducting a site survey to build a path loss model, however, despite the many published LoRaWAN site surveys there is no accepted survey methodology. Therefore, in this paper, we propose a well-defined methodology that specifies how to plan and execute a site-survey for LoRa/LoRaWAN. Several disjoint survey paths are explored simultaneously to capture the high spatial variations of LoRaWAN coverage. The proposed methodology was examined in a small, hilly, town in rural Ireland and the measurements were fitted in both a single and dual component (near and far) log-distance path loss model, the latter being more accurate. We also investigated the impact of terrain elevation on the path loss and packet delivery rate, the impact of multipath reflection, and the variation in link quality from indoor and outdoor transmissions. Finally, using the collected measurements, we evaluated how increasing the number of survey paths affects the accuracy of the resulting channel model.