Dynamic Spreading Factor and Power Allocation in LoRaWAN Networks Using Reinforcement Learning
Résumé
LoRaWAN has emerged as a leading solution for supporting long-range, low-power communication in massive IoT deployments. However, ensuring reliable data transmission in dense and dynamic environments poses significant challenges, particularly with interference mitigation and energy efficiency.To address these limitations, this paper proposes an innovative approach that integrates reinforcement learning for decentralized and adaptive resource allocation. The approach enables end devices to autonomously configure their transmission parameters, such as spreading factors and power levels, based on local environmental feedback. By eliminating reliance on centralized Adaptive Data Rate mechanisms, the proposed approach significantly enhances energy efficiency and improves scalability in high-traffic LoRaWAN networks. The approach is validated through real-world experimental deployments along with extensive simulations, demonstrating notable improvements in energy effectiveness. This work highlights the potential of reinforcement learning to optimize resource management and enhance the communication robustness in LoRaWAN environments.