Cooperation between the cloud and the fog in the Internet of Things (IoT) system can enhance the efficiency of limited-power and computationally-constrained IoT devices in terms of delay and energy consumption by offloading delay-sensitive and computation-intensive tasks to nearby fog nodes. The purpose of computation offloading is to minimize the energy consumption of IoT devices, meanwhile assuring maximum tolerable delay of tasks. In this paper, we propose a computation offloading scheme in an IoT-fog-cloud architecture considering a multiuser multi-fog nodes scenario. The proposed scheme considers multiple offloading through collaboration between fog nodes, which aims to optimize the offloading probability and the transmit power allocation jointly. Since the formulated optimization problem is non-convex and NP-hard, exploiting successive convex approximation (SCA) and Dinkelbach method, an iterative two-steps algorithm is proposed to solve the problem efficiently. The simulation results expose the tradeoff between energy consumption and task completion time in IoT devices. Further, the convergence of the proposed algorithm is verified.

物联网（IoT）系统中云与雾之间的合作可以通过将延迟敏感和计算密集型任务卸载到附近的雾中来提高延迟和能耗方面的受限功率和受计算限制的IoT设备的效率节点。计算分流的目的是最大程度地减少IoT设备的能耗，同时确保最大可容忍的任务延迟。在本文中，我们考虑了多用户多雾节点的情况，提出了一种IoT-雾云架构中的计算分流方案。提出的方案通过雾节点之间的协作考虑了多个卸载，旨在共同优化卸载概率和发射功率分配。由于拟定的优化问题是非凸的且NP-困难的，利用连续凸逼近法（SCA）和Dinkelbach方法，提出了一种迭代两步算法来有效地解决该问题。仿真结果揭示了物联网设备的能耗和任务完成时间之间的权衡。此外，验证了所提出算法的收敛性。