A Clustering-Based Adaptive Data Rate Technique for Industrial LoRaWAN-IoT Networks

For large-scale, remote, and harsh environments-based Industrial-IoT applications, the long-range wide-area network (LoRaWAN) is a low-power protocol that enables reliable, robust, and long-range communication. A critical aspect of LoRaWAN is the adaptive data rate (ADR), which allows end devices in Industrial Internet of Things (IIoT) applications to modify the data rate dynamically according to channel conditions. This guarantees efficient connectivity and prolongs the battery life of an end device. Data loss in harsh IIoT situations can be huge because the channel traffic tends to vary, which might include significant congestion and interference. There are several issues with the present ADR method, including its slow convergence time and lack of adaptability. Furthermore, in large IIoT applications, the ADR approach is overwhelming in high-traffic networks since it only considers link-level performance and assigns configuration parameters like spreading factor (SF) and transmission power (TP) to individual end devices. As a result, network performance could be improved. This article proposes an effective cluster-based ADR mechanism (Cluster-ADR) to minimize collisions and packet loss. The proposed Cluster-ADR clusters the end devices based on the estimated path loss. To address the aforementioned issues, this article proposes an effective Cluster-ADR by employing signal orthogonality to allocate SFs to minimize collisions and packet loss in dense LoRa-based IIoT networks. A clustering method based on measured path loss is also presented that clusters end devices, assigns different channels to each cluster, and allocates optimal SFs for each end device within the cluster. Additionally, this article presents an analytical computation of the energy consumption and convergence time of standard and the proposed ADR mechanisms using a Markov model. The performance of Cluster-ADR is examined in terms of packet success rate, power consumption, and convergence time using simulations and testbed implementation.