Measurements and Analysis of Large Scale LoRa Network Efficiency

The LoRa networks allow building long-range and low power wireless networks using a star of stars topology, with multiple gateways serving low-cost devices. We present the measurements obtained in an urban environment from two independent real-life network topologies located in two Polish cities. These topologies consist of about 6400 and 4400 location points and included more than eight mln data points. We focus on the variability of RSSI (Received Signal Strength Index), SNR (Signal to Noise Ratio), probability of packet delivery, and these measures' dependence on distance. We also present an estimated path loss derived from the measurement data. The analysis shows significant variability of packet delivery probability and the RSSI. Approximately 65% of the population provides > 80% packet delivery probability in setup A, and 85% of the population provides > 80% packet delivery probability in setup B. The calculated path loss model is slightly more pessimistic than shown by most studies, as at longer distances, the measured SNR and RSSI are higher than the actual average since all transmissions with RSSI lower than gateway sensitivity (reception threshold) are not considered in the analysis as there is no data recorded. We also compare the measurements with simulated path loss, showing that commonly used simulation models may overestimate the path loss for larger distances.