Résumé : 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.