Performance modelling and optimal design in UAV-connected wireless networks

Unmanned aerial vehicles (UAVs), also known as drones, have already been widely implemented in wireless networks for promoting network performance and enabling new services. To efficiently explore the diversity introduced by the mobility of UAV, many efforts have been made in the design of the UAV trajectory under various wireless scenarios. However, the continuous trajectory designs for UAVs generally suffer from the infinite number of variables and the inherent continuity of the trajectories. To tackle these difficulties, we develop two highly efficient approaches in this thesis for the continuous UAV trajectory designs, as well as the joint network designs with resource allocation.More specifically, focusing on the UAV-enabled wireless power transfer (WPT) network, we rigorously proved the optimality of the successive-hover-fly (SHF) structure in the corresponding continuous UAV trajectory design and accordingly proposed an SHF structure based approach. More specifically, we have proved that within the corresponding optimal trajectory, the UAV is supposed to successively hover at a sequence of hovering points and fly between hovering points with always the maximum speed. Based on the SHF structure and the successive convex approximation techniques, an iterative algorithm with guaranteed convergence behaviour is proposed for efficiently obtaining a suboptimal solution. The significantly reduced number of variables under SHF structure ensures a much lower design complexity. Afterwards, the developed SHF structure based design approach is further extended to efficiently address the time minimization problems for completing wireless tasks in different networks, i.e., for completing the WPT tasks in UAV-enabled WPT networks requiring pure UAV trajectory designs and the data collection tasks in UAV-aided communication scenarios demanding joint user scheduling and trajectory designs. Furthermore, targeting at achieving the trajectory optimality, we concentrated on the single-user scenario and creatively developed a ...