AoI-Aware Data Collection and Energy Replenishment for Multi-UAV-Enabled IoT Systems
Kaijin Shi, Juan Liu, Ling Xie, Zheng Zhou, Hua Chen, Guinian Feng · IEEE Transactions on Green Communications and Networking · 2025
Autonomous Aerial Vehicles (UAVs) have emerged as crucial enablers of efficient data collection and energy providers for Internet of Things (IoT) networks. Given the inherent limitations of UAV battery capacity, it is imperative to devise effective energy replenishment strategies. This paper delves into the fresh data collection and energy replenishment problem in multi-UAV-enabled IoT networks, aiming to minimize the Age of Information (AoI) of Sensor Nodes (SNs). To ensure consistent operation, the SNs rely on wireless power transfer from the UAVs, while the UAVs, in turn, recharge their batteries at charging stations. As the UAVs fly over the SNs, they efficiently gather up-to-date data in a timely manner. We model this complex problem as a Partially Observable Markov Decision Process (POMDP). Then, we employ two Multi-Agent Reinforcement Learning (MARL) algorithms: the Value-Decomposition Network (VDN) and the Q-Mixing (QMIX), to solve the problem. In these algorithms, each UAV serves as an intelligent agent, independently learning the environment to make strategic decisions regarding its flight, association with SNs, and recharging. Simulation results demonstrate significant advantages of two MARL algorithms over baseline approaches.