Energy-Fairness-Aware Multi-UAV Trajectory Planning Algorithm for AoI-Constrained Data Collection

Multi-UAV assisted data collection has been widely applied to enhance data freshness in wireless sensor networks. Most existing studies focus on minimizing the maximum or average Age of Information (AoI), overlooking the heterogeneous AoI requirements of the sensor nodes. These varying constraints increase the difficulty of scheduling UAVs' flight paths and corresponding speeds, resulting in higher total energy consumption and imbalanced energy usage. To address this issue, we optimize the cooperative trajectories of UAVs to meet diverse AoI constraints while considering their limited on-board energy. The cooperative trajectory planning is essential not only to reduce overall energy consumption but also to ensure energy fairness, which is critical for balancing workloads and avoiding excessive burden on individual UAVs. In this paper, the trajectory of each UAV is represented by the data collection positions and flight speeds, where each position is determined by a target sensor node (SN) and an associated hovering offset. We propose a deep reinforcement learning-based algorithm, Cooperation Hybrid Independent Proximal Policy Optimization (CHI-PPO), which jointly schedules UAVs' access sequences to target nodes, hovering offsets, and flight speeds. Since target nodes are modeled as discrete actions, while offsets and speeds are continuous actions, handling such hybrid decision spaces poses a significant challenge. Our algorithm draws inspiration from the distributed framework of Independent Proximal Policy Optimization (I-PPO) and establishes a Cooperation Hybrid Proximal Policy Optimization (CH-PPO) network to select hybrid actions for scheduling the trajectory of each UAV. Simulation results demonstrate that CHI-PPO achieves stable convergence and significantly outperforms state-of-the-art baselines in both energy efficiency and fairness.