A Robust Connectivity Maintenance Method for Multi-UAVs Network Under Obstacle Environment

The robustness of multi-UAVs communication network decreases when some UAVs are destroyed, most existing methods construct robust networks by adding a large number of links. However, as the links increase, the consumption of network bandwidth increases, and the interference between signals also increases. In this paper, a method based on position control and link optimization has been proposed. For the position control, artificial potential field (APF) method is used to guide UAVs to their destination and avoid collisions. For the link optimization, reinforcement learning is used to explore as few edges as possible for each UAV while ensuring network robustness. In addition, in order to prevent the network from over relying on some nodes, the sampling method based on second-order neighbor is designed. In 3D space simulation, we verify that our method has very approximate robustness to other methods even under the network edges are reduced by \(30\%\) .