Optimizing Energy Efficiency in UAV-Assisted Wireless Sensor Networks With Reinforcement Learning PPO2 Algorithm

The present study aims to investigate the problem of minimizing energy consumption in unmanned aerial vehicle (UAV)-assisted wireless sensor networks (WSNs). In the current context of scarce energy resources, efficient management of energy in UAVs and sensor nodes (SNs) is crucial for enhancing the overall network performance and prolonging network lifespan. To address this, a comprehensive approach is proposed, integrating cluster analysis and path planning to optimize energy utilization and path selection. Initially, a honey badger algorithm (HBA) optimized possibilistic fuzzy C-means (PFCM) algorithm is using to select more suitable clusters, mitigating energy wastage near the data center (DC) and reducing energy consumption in nodes far from the cluster center. This optimization process also simplifies UAV path planning complexity. Subsequently, the clustering results are input into the proximal policy optimization 2 (PPO2) algorithm for UAV path planning, aiming to determine the optimal trajectory. The PPO2 algorithm uses reinforcement learning to balance energy consumption and path optimization, thus minimizing energy consumption across the entire network. To further enhance the path, the two-opt algorithm is used to improve the trajectory of the strategic network. Through a local search algorithm, the path of the strategic network is optimized, thereby improving energy utilization efficiency and network performance. Extensive simulation experiments are conducted to validate the effectiveness of the proposed algorithm. Comparative analysis with other algorithms indicates significant improvements in energy consumption, effectively prolonging the network’s lifespan. This method allows for substantial energy savings in UAVs and SNs, consequently enhancing the overall energy efficiency of the network.