Design and Performance Evaluation of an AI-Driven Hybrid Simulation Model for LoRaWAN Networks

This paper proposes an attempt to combine simulation and artificial intelligence (AI) to facilitate the operation of the LoRaWAN network. Our network parameters include transmit power, spreading factor, and coding rate, which we predict using deep learning (ANN, DNN, GRU), reinforcement learning (DQN, multi-agent RL), and ensemble strategies to find optimal parameters and maximize energy efficiency, packet success rate, and throughput. The approaches are tested through simulation with NS-3 and LoRaEnergySim platforms to validate them under realistic conditions of interference and traffic. Findings show that AI methods can substantially enhance resource allocation and conserve energy than the state-of-the-art methods, offering a solid framework of adaptive LoRaWAN network management.