The IoT network allows IoT devices to communicate with other devices, applications, and services by exploiting existing network infrastructure. Recently, a promising paradigm, MEC, emerging for alleviating high latency data services in cloud computing framework plays an important role in the IoT network. Network performance and intelligence can be improved by integrating cognitive and cooperative mechanisms in the MEC framework. However, the QoS of computation-intensive tasks may degrade because of the limited available computational resources in MEC servers. Moreover, the characteristics of resources belonging to MEC servers and cloud servers are commonly different. In order to optimize the strategy of resource assignment, the tasks of assigning the limited computational resources in MEC servers and resolving the high latency problem in cloud servers have attracted growing interest from researchers. In this article, we propose a joint optimization paradigm for task-driven resource assignment based on evolutionary computation considering the power consumption and computation/communication delay simultaneously. The MEC framework consists of MEC servers, mobile devices, and cloud servers, and offloads the computational resources to the edge of end users. Additionally, we introduce and analyze three typical task-driven cases, which are the server-determined condition, server-flexible condition, and server-uncertain condition, respectively. Finally, we present the existing technical challenges and discuss the open research issues.

中文翻译：

---

利用边缘计算的移动边缘计算中的任务驱动资源分配

物联网网络允许物联网设备通过利用现有的网络基础架构与其他设备，应用程序和服务进行通信。最近，为缓解云计算框架中的高延迟数据服务而出现的一种有前途的范例MEC在物联网网络中起着重要的作用。通过将认知和协作机制集成到MEC框架中，可以提高网络性能和智能。但是，由于MEC服务器中可用的计算资源有限，因此计算密集型任务的QoS可能会降低。此外，属于MEC服务器和云服务器的资源的特征通常是不同的。为了优化资源分配策略，在MEC服务器中分配有限的计算资源以及解决云服务器中的高延迟问题的任务吸引了研究人员的越来越多的兴趣。在本文中，我们提出了一种基于演化计算的任务驱动资源分配联合优化范例，同时考虑了功耗和计算/通信延迟。MEC框架由MEC服务器，移动设备和云服务器组成，并将计算资源卸载到最终用户的边缘。此外，我们介绍并分析了三种典型的任务驱动情况，分别是服务器确定的条件，服务器灵活的条件和服务器不确定的条件。最后，我们提出了现有的技术挑战并讨论了开放研究问题。