Efficient Data Collection in Large-Scale UAV-aided Wireless Sensor Networks

Dispatching unmanned aerial vehicles (UAVs) to collect sensor data from distributed sensor networks is expected to significantly improve the data collection efficiency of traditional Wireless Sensor Networks (WSNs). In this paper, we consider data collection for distributed large-scale UAV-aided WSNs. We first divide the target area into clusters, and determine the cluster head election and data forwarding rules within the cluster based on the value of information (VoI) and power of nodes. Then sensor nodes collect information around by event-driven, and the UAV acts as a mobile sink to collect the data of cluster head nodes. The direct future prediction (DFP) model is used to plan the UAV path, which allows the UAV to handle multi-target tasks, maximizing the total VoI collected while ensuring that the UAV is charged at low power. Simulation results show that our DFP model is superior to traditional Q-Learning and Deep Q-Learning Network (DQN) in performance.