Quantile Classification of Variance from the Mean for Spreading Factor Allocation in LoRaWAN

The adaptive data rate (ADR) is an important scheme for optimizing data rate or spreading factor (SF), time on air (ToA), and power consumption of the end nodes in the long range wide area network (LoRaWAN). However, the ADR scheme can still be improved for better network performance. This paper describes the novel ADR scheme based on the quantile classification of variance from the mean (QCVM). The QCVM method enhances the distribution of the SFs to reduce a probability of packet collision and loss. The simulation is performed to evaluate the network performance before and after the implementation of the QCVM method. The results show that the data extraction rate (DER) is improved over the original ADR scheme when the number of nodes and payload size increase. DER is however, smaller in larger coverage area, probably due to the larger portion of the nodes being allocated to high SFs, resulting in more packet loss.