Lightweight Joint Optimization of Clustering and Trajectory for UAV-Assisted WSNs

UAV-assisted data collection in wireless sensor networks offers flexibility and cost-efficiency, yet faces challenges from complex air-ground coupling and joint trajectory-resource optimization. Traditional deep reinforcement learning (DRL) methods still struggle with limited deployment capability on embedded platforms. This paper proposes a lightweight DQN framework with structured pruning to jointly optimize UAV path planning and cluster head selection. The approach employs forward-activation-based pruning for model compression. Simulations show that the proposed method outperforms ACO, with model size reduced by up to 90% while maintaining real-time performance suitable for UAV deployment.