Satellite-controlled UAV-assisted IoT Information Collection with Deep Reinforcement Learning and Device Matching
Wei Xin, Guobin Zhang, Zhu Han
Abstract
A space-air-ground integrated network composed of satellite, unmanned aerial vehicles (UAVs) and Internet of Things (IoT) devices is constructed to collect the wide-area information in this paper. The IoT devices sense the environment information, and the UAVs fly to collect the data and transmit to the satellite. In order to enhance the collected information freshness, the age of information (AoI) of the system is minimized by the UAV trajectory design with the pairing between the UAVs and IoT devices. The satellite works as the central controller to collect the minimal AoI values between each UAV and IoT device, which are obtained by the UAV training. The deep reinforcement learning is employed during the training to achieve the AoI along with the optimal destined position of the UAV. After then, the satellite builds the prefer lists of the UAVs and IoT devices based on the AoI values and finishes the matching by the Gale-Shapley algorithm. The simulation verifies the superiority of the proposed scheme over the existing approach.