As one of promising distributed multi-robot system, Unmanned Aerial Vehicles (UAVs) can collaborate to offload complex tasks in edge networks. A Smart and Trust Multi-UAV Task Offloading (STMTO) system is established to offload tasks from Internet of Thing (IoT) devices to edge severs through UAVs with a trust style. In STMTO system, first, a group of UAVs is dispatched to relay tasks from devices to edge servers with rich computing resource. A collaborative task collection scheme is proposed to minimize energy consumption and the task processing delay by dividing working area for each UAV and designing the flight trajectory. Secondly, a many-to-many task double auction model is established for devices and edge servers to maximize the offloading utility, where devices act as buyers, edge servers as sellers, and UAVs as auctioneers. Last, to resist attack of malicious edge servers and ensure the task security, a novel trust evaluation method based on the comparison of true utility and expected utility is integrated in auction mechanism. The theoretical analysis and implementation results show that the proposed STMTO system not only achieve the best utility for devices and edge servers simultaneously, but also identify the malicious edge servers and protect task from attacks.

作为有前途的分布式多机器人系统之一，无人驾驶飞行器 (UAV) 可以协作卸载边缘网络中的复杂任务。建立了智能和信任多无人机任务卸载（STMTO）系统，通过具有信任风格的无人机将任务从物联网（IoT）设备卸载到边缘服务器。在STMTO系统中，首先调度一组无人机将任务从设备中继到具有丰富计算资源的边缘服务器。提出了一种协同任务收集方案，通过划分每个无人机的工作区域和设计飞行轨迹来最小化能耗和任务处理延迟。其次，针对设备和边缘服务器建立多对多任务双拍卖模型，以最大化卸载效用，其中设备作为买家，边缘服务器作为卖家，无人机作为拍卖者。最后的，为抵御恶意边缘服务器的攻击，保证任务安全，拍卖机制中融入了一种基于真实效用与期望效用比较的信任评估方法。理论分析和实现结果表明，所提出的 STMTO 系统不仅可以同时实现设备和边缘服务器的最佳效用，而且可以识别恶意边缘服务器并保护任务免受攻击。