Construction and performance evaluation of a LoRaWAN testbed

Abstract

The Internet of Things (IoT) is the creation of a large network of interconnected devices that can communicate wirelessly to enable innovation across multiple fields. A newly developed wireless technology is the Long Range Wide Area Network (LoRaWAN) protocol and the performance of LoRaWANs is the focus of this work.

Literature revealed that existing performance evaluations were mainly performed over short time periods and with only a few nodes. Additionally, communication was only evaluated over short distances in urban areas and long distance evaluations were mainly performed in rural environments. A research gap, therefore, existed for the evaluation of this technology over long and short distances and over long time periods in an urban environment.

The contributions of this research can be laid out as three objectives. The first was to determine how effective a LoRaWAN would be for nodes at larger distances (> 1.5km) from a gateway in an urban environment. The second was to determine how the Adaptive Data Rate (ADR) scheme impacts the performance of a LoRaWAN. The final objective was to determine the impact of other LoRaWAN parameters on the performance of a LoRaWAN.

A LoRaWAN testbed was constructed consisting of 18 nodes and one gateway, and several experiments were performed. Each node was sent configuration commands remotely and each experiment was executed until every node sent 1000 packets. Packet delivery in the form of the Packet Delivery Ratio (PDR) was chosen as the performance metric to be used and this was calculated from the ratio of sent versus received packets.

The first objective was to examine LoRaWAN performance from the perspective of long range nodes. This was answered by examining the data from 5 nodes, located at distances between 1.98 km and 5.19 km from the gateway. The results show that a LoRaWAN should only be used if a PDR of between 60~\% and 80 \% is acceptable for long range nodes. Furthermore, there can be significant differences in the PDRs between nodes and these differences can become bigger over long distances. The data revealed that all aspects, but especially the ADR scheme, should be considered for long range nodes.

The LoRaWAN protocol's ADR scheme was the focus of the second research contribution. This scheme aims to optimise network throughput by adjusting the transmission settings of individual nodes in the network. When the ADR scheme was enabled, the PDR was consistently worse for all three groups (the 18 nodes were divided into three groups based on distance). An examination of the data revealed that, out of 5 possible data rates, the scheme dominantly assigned either the slowest or the fastest data rate. The slower data rates allow for greater range, but the scheme algorithm did not assign these rates to nodes at longer distances, and as a result their performance suffered. The data also revealed that the data rate assignments oscillated between choices, forcing nodes to switch between rates very frequently.

The last research contribution was to determine the impact of other LoRaWAN parameters on the performance of a LoRaWAN. There are numerous parameters and combinations, and thus only the impact of payload length, link checks and waiting time on the PDR were examined and these were found to be not significant. The evaluation of enabling acknowledgements did reveal a significant improvement to the PDR, but whilst this feature works well in small networks (such as the testbed), it would be detrimental in large networks. LoRaWAN gateways can either be in receiving or transmitting mode and gateways are unable to receive packets when transmitting acknowledgements, therefore resulting in an increase in missed packets for a gateway in a large and congested network.