3D placement of UAVs with optimal beamforming for multi-user communications

This article proposes the three-dimensional (3D) deployment of multiple unmanned aerial vehi-cles (UAVs) as relays between internet of things (IoT) access points (APs) and terrestrial base stations (BSs). Limited by the total battery capacity of UAVs, the 3D placement of multiple UAVs is designed to minimize the number of required UAV flights to serve the IoT APs within a pre-defined time while maximizing the number of successfully served IoT Aps, which is a nonlinear and NP-hard problem. To achieve the above goals, we linearize the problem into an integer linear programming (ILP) model and propose two genetic algorithms (GAs) by adjusting the number, 3D positions, and beamwidths of UAVs. Simulation results show that the use of flexible beamwidths for UAVs can significantly improve performance. Moreover, for different signal-to-interference-noise-ratio (SINR) thresholds, the proposed consecutive GA (ConsGA) clearly outperforms other heuristics. Moreover, the SINR and UAV bandwidth significantly affect the overall performance. Furthermore, our simulations show that the proposed discrete GA (DisGA (960)) achieves the lowest number of UAV flights compared to other algorithms while successfully meeting the SINR threshold, and its performance is the closest to that of the opti-mization model (i.e., ILP model).