Efficient graph-based gateway placement for large-scale LoRaWAN deployments

Low Power Wide Area Networks with Long Range Wide Area Networks (LoRaWANs) as one of their most prominent representatives are promising solutions for future Internet of Things deployments. The technology is characterized by low energy requirements leading to long battery lifetimes. However, the drawbacks are limited throughput rates and the unreliable nature of the LoRa technology. Specifically its random channel access approach leads to significant message collisions and thus, data loss in larger deployments. From a network planning point of view, the reduction of potentially colliding messages at the frequency bands in combination with limiting the transmission duration of messages is a promising approach. For that reason, we present a novel graph-based gateway placement approach. We show that our approach performs similar to state-of-the-art related work in the worst case and reduces the required number of gateways by up to 40% while reducing the collision probability by up to 70%. Furthermore, we overcome the challenge of scalability of our approach when placing gateways in very large networks. By splitting the problem into smaller problem instances by pre-clustering, we can solve arbitrarily large instances efficiently without a significant increase in number of required gateways in real networks.

Automated summary

Low Power Wide Area Networks (LoRaWANs) are a promising solution for future IoT deployments due to their low energy requirements and long battery lifetimes. However, the technology is limited by low throughput rates and the unreliable nature of LoRa. To address these challenges, we propose a novel graph-based gateway placement approach that reduces collision probability and the number of required gateways. We also overcome scalability challenges by pre-clustering and efficiently solving large network instances.