Abstract
This work examines the deployment of multiple fixed-wing unmanned aerial vehicles (UAVs) for data-gathering from ground IoT devices, and the corresponding device association policy. Each UAV is assumed to hover above its associated devices following a circular trajectory. The device association and the UAVs' trajectory centers and radii are jointly optimized to maximize the energy-savings relative to a constant transmission power scheme. Given the trajectory centers and radii, the device association problem is modeled as a multiple 0-1 knapsack problem, taking into consideration the load demands of different devices as well as UAVs' service capacities. A two-stage maximum energy-saving device association policy is proposed, where each UAV first solves a single knapsack problem based on all connectable devices, and then resolves conflict with others by a maximum profit assignment. Moreover, given the device association, the UAVs' trajectory centers and radii are optimized by an iterative load-balancing algorithm, where the trajectory centers are chosen as a load-dependent weighted sum of the associated devices' locations. The device association and the UAV deployment are optimized in turn until convergence. Simulation results show that our proposed schemes outperform candidate algorithms in terms of the total energy-savings of IoT devices.