Millions of Internet of Thing (IoT) devices have been widely deployed to support applications such as smart city, industrial Internet, and smart transportation. These IoT devices periodically upload their collected data and reconfigure themselves to adapt to the dynamic environment. Both operations are resource consuming for low-end IoT devices. An edge computing enabled unmanned aerial vehicle (UAV) is proposed to fly over to collect data and complete reconfiguration computing tasks from IoT devices. Distinct from most existing work, this paper focuses on flight speed scheduling that allocates proper flight speed to minimize the energy consumption of the UAV with a practical energy model, under the constraints of individual task execution deadlines and communication ranges. We formulate the Energy-Efficient flight Speed Scheduling (EESS) problem, and devise a novel diagram to visualize and analyze this problem. An optimal energy-efficient flight speed scheduling (Offspeeding) algorithm is then proposed to solve the offline version of the EESS problem. Utilizing Offspeeding and the optimal properties obtained from the theoretical analysis, an online heuristic speed scheduling algorithm is developed for more realistic scenarios, where information from IoT devices keeps unknown until the UAV flies close. Finally, simulation results demonstrate our online heuristic is near optimal. This research sheds light on a new research direction, e.g., deadline driven UAV speed scheduling for edge computing with a practical propulsion energy model.

数以百万计的物联网（IoT）设备已被广泛部署，以支持智能城市，工业互联网和智能交通等应用。这些物联网设备会定期上传其收集的数据并重新配置以适应动态环境。对于低端物联网设备，这两种操作都消耗资源。提出了一种支持边缘计算的无人机（UAV）飞越以收集数据并完成IoT设备的重新配置计算任务。与大多数现有工作不同，本文着重于飞行速度调度，该飞行速度调度在单个任务执行截止日期和通信范围的约束下，通过实用的能量模型分配适当的飞行速度，以最大程度地降低无人机的能耗。我们制定了节能飞行速度计划（EESS问题），并设计出新颖的图表来可视化和分析该问题。一个Ø ptimal能源-E FF icient飞行速度正点荷兰国际集团（Offspeeding）算法，然后提出解决的离线版本EESS问题。利用超速驾驶和从理论分析中获得的最佳属性，开发了一种在线启发式速度调度算法，用于更现实的场景，其中来自物联网设备的信息一直保持未知，直到无人机飞近为止。最后，仿真结果表明我们的在线启发式算法接近最佳。这项研究为新的研究方向提供了启示，例如 最后期限驱动的无人机速度调度，以实用的推进能量模型进行边缘计算。