Estimating the number of gateways through placement strategies in a LoRaWAN network
One of the promising technologies for the Internet of Things (IoT) is represented by the Long Range (LoRa) and its communication protocol Long Range Wide Area Network (LoRaWAN). Considering that the communication between end devices and the network server occurs through gateways (GW) acting as an exchange point or forwarding messages, network planning and optimization are considered significant issues that impact device performance as well as capital expenditure (CAPEX) and operational expenditure (OPEX) costs, planning is required to deploy a number of GW based on device positioning. In this context, we propose a comparative study of gateway positioning strategies that use the Fuzzy C-Means (FCM), Gustafson-Kessel (GK), and K-means (KM) algorithms; adding one that segments the scenario into 2km grids called Grid25 and Rand22 that randomly arranges the GWs, analyzing the Received Signal Strength Indication (RSSI), Signal to Noise Ratio (SNR), delay, distance, CAPEX, and OPEX metrics, making it possible to establish the number of GWs and costs as well as the performance of the strategies. The results show a reduction in the amount of GW as well as CAPEX and OPEX with approximately the same proportion of delivery compared to Grid25. FCM and GK resulted in superior performance compared to KM starting at 22 GW for RSSI and SNR, while FCM and KM obtained lower distances and delays than GK, starting at 18 GW.