A Deep Reinforcement Learning Approach to Energy-harvesting UAV-aided Data Collection

Unmanned aerial vehicles (UAVs) can be used as mobile relays to assist in wireless communications due to their high mobility. This paper considers UAV-assisted data collection in wireless sensor networks (WSNs), where energy harvesting is used to provide sustainable energy for the UAV. In particular, the transmission opportunities of the ground sensor nodes and the flight trajectory of the energy-harvesting-powered UAV are jointly optimized to minimize the age of information (AoI) while maintaining the UAV's energy consumption as low as possible. This problem is modeled as a Markov decision process (MDP) with relatively large state and action spaces. To break the curse of dimension and speed up the convergence, the Asynchronous Advantage Actor-Critic (A3C) algorithm is employed to make real-time decisions in the deep reinforcement learning framework. Simulation verify the effectiveness of the proposed data collection approach.