Energy-efficient multi-hop LoRa broadcasting with reinforcement learning for IoT networks

Low power wide area networks (LPWAN) have grown significantly in popularity recently, and long-range (LoRa) technologies have drawn notice as a branch of LPWAN. Nevertheless, most current research primarily concentrates on optimizing communication protocols or mechanisms for the LoRa uplink. Considering the demand for large-scale data distribution in the IoT environment, we propose a novel mechanism for LoRa broadcasting with formula derivation and parameter analysis. This scheme adopts the advantages of both LoRa protocols and multi-hop technology that make the data quickly spread to all devices from the center of an area.This scheme optimizes transmission energy consumption by selecting proper relays to alleviate the problem of power shortage in LoRa devices. In this paper, we design an algorithm based on machine learning and reinforcement learning to reduce transmission costs for LoRa devices. The superiority of the proposed scheme in saving communication resources has been demonstrated compared to traditional methods. When broadcasting data downstream, it can save approximately 87.4% of the time. Moreover, through simulation analysis, the proposed algorithm can save at least 12.61% transmitting energy under constraints comparing with the benchmark algorithms.