Modelling of IoT Traffic and Its Impact on LoRaWAN

Abstract

The recent growth of IoT use-cases in a wide array of industrial, utility and environmental applications has necessitated the need for connectivity solutions with diverse requirements. Connectivity through BLE, Zigbee and 6LoPAN are examples of short-range IoT deployments. But to provide connectivity to a high density of devices over larger coverage areas, Low-Power Wide-Area Network (LPWAN) technologies in both licensed as well as unlicensed bands have been considered. In this paper, we consider modelling the traffic from IoT devices connected through LPWAN technologies. Due to diverse applications of IoT, it is not trivial to have a single traffic model to represent all of them, but the traffic can be broadly classified as either periodic, event-triggered, or a combination of both. We evaluate the performance of LoRaWAN, one such LPWAN technology, in the presence of a hybrid of both traffic types, where the event propagates spatially over time. In a practical deployment of sensor based IoT devices, the devices are usually densely deployed to ensure sufficient & reliable measurement. Thereby, when an event occurs, they exhibit spatial & temporal correlation in their traffic rate due to the natural phenomena of the metric they measure. We use the CMMPP model to represent such characteristic traffic from independent IoT devices triggered by an event. The characteristics of LoRa, the physical layer of LoRaWAN, is abstracted based on required signal strength and interference thresholds for different modulation parameters. Through system simulations, we demonstrate that there is a significant performance hit in LoRaWAN based networks, during the occurrence of events. In particular, using the packet delivery rate (PDR) as the metric, we found that while the system was able to handle regular updates from the devices with a PDR > 80%, event-driven traffic nearly impaired the network causing the PDR to drop below 10%.