Improvement of LoRa Communication Scalability using Machine Learning Based Adaptiveness

Abstract

Number of embedded devices connected to the Internet is rapidly increasing, especially in the era of the Internet of Things (IoT). The growing number of IoT devices communicating wirelessly causes a communication-parameters selection problem, implying the increasing number of communication collisions. Multiple factors of IoT networks signify this problem, such as inability to communication-channel listening prior to the transmission (due to longer distances), energy constrains (due to inability of powering devices from the grid), or limitation of duty cycle and high interference (due to usage of unlicensed band in communication technologies). This article is focused on alleviating this problem in LoRa networks, which is one of the most promising technology for long-range and low-powercommunication. We utilize the existing LoRa@FIIT protocol to achieve energy-efficient communication. The scalability of the LoRa network is increased by modifying the communication-parameters selection algorithm. By ensuring of quality of service mechanism at each node in the infrastructure, the application domain of the proposed architecture is widened. The simulation-based experimental results showed a significantly reduced number of collisions for mobile nodes, which reduces the channel congestion and the wasted energy by retransmissions.