Probabilistic Analysis of Packet Losses in Dense LoRa Networks

Nowadays low-power wide area networks (LPWAN) technologies, such as LoRa, have attracted interests from researchers and practitioners with the ability to achieve robust communication, which is suitable for harsh environments with many obstacles. Moreover, LoRa is capable of delivering data over very long communication distances, which can be beneficial for varieties of applications. Due to its wider range, however, a LoRa network may be subject to a high probability of packet collisions when deployed with a large number of end devices. To help LoRa network planners analyze the performance in terms of packet losses, and adjust communication parameters to improve reliability, a proper mathematical model is needed. In this paper, we develop a packet loss model that takes into account individual end devices' packet arrival rates, transmission powers, and distances from a gateway. Experiments show that packet loss probabilities predicted by the model mostly conform with the results from simulations and the model accurately predicts the aggregated interference, along with the expected loss rate, under moderate to heavy data traffic.