In various Internet-of-Things (IoT) applications, a number of devices and sensors are used to collect data sets. As devices become more capable and smarter, they can not only collect data sets, but also process them locally. However, since most devices would be limited in terms of computing power and energy, they can take advantage of offloading so that their tasks can be carried out at mobile edge computing (MEC) servers. In this paper, we discuss computation offloading for devices in IoT applications. In particular, we consider users or devices with sporadic tasks, where optimizing resource allocation between offloading devices and coordinating for multiuser offloading becomes inefficient. Thus, we propose a two-stage offloading approach that is friendly to devices with sporadic tasks as it employs multichannel random access for offloading requests with low signaling overhead. The stability of the two-stage offloading approach is considered with methods to stabilize the system. We also analyze the latency outage probability as a performance index from a device perspective.

翻译：在各种互联网应用中,许多装置和传感器被用于收集数据集。随着装置变得更加有能力和智能,它们不仅能够收集数据集,而且可以在当地处理。然而,由于大多数装置在计算电能和能源方面受到限制,它们可以利用卸载来完成移动边缘计算(MEC)服务器的任务。在本文件中,我们讨论为IoT应用中的装置计算卸载。特别是,我们考虑到用户或任务零星的装置,在卸载装置与多用户卸载协调之间优化资源分配变得效率低下。因此,我们建议了两阶段卸载方法,在使用多通道随机访问以低信号间接卸载请求时,对零星性任务装置是友好的。两阶段卸载方法的稳定性与稳定系统的方法一起考虑。我们还从设备角度分析延载概率的性能指数。