Enabling Edge computing over LoRaWAN. A Device-Gateway Coordination Protocol

The freedom of the LoRaWAN license-free ad-hoc deployment model can significantly reduce the complexity of network deployment, however, it introduces certain problems that hinder network scalability and performance. First, Network Gateways simply forward frames to the centralised Network Server without the possibility of becoming Edge computing processing elements. Second, Network Gateways may have overlapping areas of network coverage (a) resulting in an increase of network traffic at the network backbone as frames are relayed to the Network Server multiple times, (b) may cause unexpected frame collisions and duty-cycle exhaustion. In this paper, a novel decentralised algorithm is presented that assigns each IoT Device to a single Network Gateway so that (a) duplicate message deliveries are completely avoided, (b) Network Gateways can become an intermediate operations layer between the IoT devices and the Network Server providing computational and storage resources. The proposed protocol is implemented in the OMNeT++ simulator and evaluated based on datasets collected from long-term real-world deployments. To improve the accuracy of the experiments in large-scale and dense urban deployments the OMNeT++ LoRa interference model is extended by considering the non-perfect orthogonality of the LoRa Spreading Factors and the channel adjacency. The performance evaluation highlights the benefits of the distributed approach proposed here and provides valuable indications on the overall performance of the network.