Efficient Slot Selection in LoRaWAN Using Reinforcement Learning and Gateway-side Guidance

Abstract

Low Power Wide Area Networks (LPWANs) support large-scale, energy-efficient Internet of Things (IoT) systems. Among them, Long Range Wide Area Networks (LoRaWANs) are widely used due to its versatility in real-world applications. This project addresses the scalability problem in LoRaWAN, where high collision rates reduce network performance significantly. The main objective is to improve packet reception rate (PRR) and fairness in such networks. To achieve this, we propose a learning-based approach that enables devices to make more informed transmission decisions using both local experience and lightweight network-wide information. The system is evaluated through simulation and real-world experiments. The results demonstrate consistent improvements in reliability and fairness compared to baseline approaches, while also showing that similar performance can be.