Trajectory Planning of Multiple Dronecells in Vehicular Networks: A Reinforcement Learning Approach

The agility of unmanned aerial vehicles (UAVs) have been recently harnessed in developing potential solutions that provide seamless coverage for vehicles in areas with poor cellular infrastructure. In this letter, multiple UAVs are deployed to provide the needed cellular coverage to vehicles traveling with random speeds over a given highway segment. This letter minimizes the number of deployed UAVs and optimizes their trajectories to offer prevalent communication coverage to all vehicles crossing the highway segment while saving energy consumption of the UAVs. Due to varying traffic conditions on the highway, a reinforcement learning approach is utilized to govern the number of needed UAVs and their trajectories to serve the existing and newly arriving vehicles. Numerical results demonstrate the effectiveness of the proposed design and show that during the mission time, a minimum number of UAVs adapt their velocities in order to cover the vehicles.