UAV-Enabled Wireless Powered Communication Networks for Over-the-Air Computation

This paper investigates an unmanned aerial vehicle (UAV)-enabled wireless powered communication network for over-the-air computation, where the UAV is employed as a hybrid access point to serve multiple terrestrial sensors. Specifically, the sensors harvest energy from the signal sent by the UAV, which is then used by the sensors to support their uplink data aggregation of sampled data. We aim to maximize the sum computation rate while satisfying the energy constraints at the sensors and the mobility constraints at the UAV. The transmit power of the sensors, the UAV trajectory, as well as the time allocation, are jointly optimized. Since the UAV may either hover or cruise while communicating with the sensors, both static and mobile UAV transmission schemes are considered. When the UAV is hovering, we propose a globally optimal closed form-based method. When the UAV is cruising, we propose a locally optimal successive inner convex approximation-based method. Finally, numerical results are provided to validate the effectiveness of our proposed two possible transmission schemes and demonstrate that our proposed locally optimal trajectory optimization scheme can effectively mitigate the doubly near-far effect by leveraging the mobility of the UAV.