ABSTRACT Unmanned aerial vehicles (UAVs) have emerged as efficient mobile data collectors capable of mitigating the communication overhead and uneven energy consumption inherent in large‐scale wireless sensor networks (WSNs). However, current clustering, cluster‐head (CH) selection, and UAV routing strategies typically operate in isolation, limiting adaptability, scalability, and long‐term energy efficiency. This paper proposes IF‐UAV, an integrated intelligent framework that enables energy‐aware clustering, hybrid machine learning–driven CH selection, continuous‐action UAV trajectory optimization, and communication‐efficient federated data aggregation. First, an enhanced K‐Medoids with Adaptive Learning Rate (KMA‐LR) algorithm forms energy‐balanced clusters that dynamically respond to residual energy gradients. Next, a Hybrid GWO‐CNN model combines global metaheuristic search with learned node‐suitability predictions for high‐accuracy CH selection. For mobility, a Deep Deterministic Policy Gradient (DDPG) agent learns smooth and energy‐optimal UAV trajectories that adapt to network conditions in real time. To reduce communication overhead, FedFusion, an edge‐assisted federated aggregation scheme using parameterized autoencoders, compresses local cluster information while preserving essential structure. Simulation results demonstrate significant performance gains over LEACH, PEGASIS, FCM, PSO‐Clust, and ACO‐UAV, including a 40%–60% extension in early stability, over 30% reduction in UAV flight distance, and up to 50% throughput improvement. The findings confirm that tightly integrated clustering, CH selection, learning‐based routing, and federated fusion provide a scalable and energy‐efficient solution for UAV‐assisted IoT deployments.
Ali Kadhum M. Al‐Qurabat (Wed,) studied this question.