基于联合决策模型的物联网边缘计算资源分配  

Resource Allocation of IoT Edge Computing Based on Joint Decision Model

作　　者：刘明 龚伟 LIU Ming;GONG Wei(Langfang Branch Campus,Hebei University of Technology,Langfang Hebei 065000,China)

机构地区：[1]河北工业大学廊坊分校,河北廊坊065000

出　　处：《计算机仿真》2021年第12期299-303,共5页Computer Simulation

摘　　要：随着应用需求的增加,一些场景要求物联网能够支持密集型计算任务。传统物联网只能提供单机资源,且负载能力有限,无法有效解决时延、资源与任务的配置问题。于是提出基于联合决策模型的物联网边缘计算资源分配方法,利用边缘网络的计算优势来弥补物联网节点本地计算资源的不足,从而提高任务时延与峰值负载的性能。先从时延、能耗、计算资源和带宽资源方面进行分析,并考虑了节点移动、数据传输和卸载等情况带来的问题。根据时间和各类资源模型的分析,建立联合模型来得到资源分配调度的最佳决策,将最小卸载模型推演至最高总效用模型,并通过最速下降法对模型进行分解,在任务卸载率一定时,求解得到资源分配情况。通过动态时变物联网环境下的仿真,得到所提方法能够在较短的执行时间内,达到较高的任务完成率,且保持较低的能耗和资源分配数量。结果表明所提方法能够适应动态时变的物联网应用需求,有效完成任务与资源的卸载决策与调度分配。With the increase of application requirements, some scenarios require the Internet of things to supportintensive computing tasks. The traditional Internet of things can only provide single machine resources, and the loadcapacity is limited, which can not effectively solve the problems of delay, resource, and task configuration. There-fore, a resource allocation method of IoT edge computing based on a joint decision model is proposed. The computingadvantage of edge networks is used to make up for the lack of local computing resources of IoT nodes, so as to im-prove the performance of task delay and peak load. Firstly, the delay, energy consumption, computing resources andbandwidth resources were analyzed, and the problems caused by node mobility, data transmission and unloading wereconsidered. Then, according to the analysis of time and various resource models, a joint model was established to getthe best decision of resource allocation and scheduling, and the minimum unloading model was deduced to the maxi-mum total utility model. The model was decomposed by the steepest descent method, and the resource allocation wasobtained when the task unloading rate was constant. Finally, through the simulation experiment in the dynamic timevarying Internet of things environment, the results showed that the proposed method could achieve a higher task com-pletion rate in a shorter execution time, and maintain a lower energy consumption and resource allocation. The resultsshow that the proposed method can adapt to the dynamic and time-varying application requirements of the Internet ofthings, and effectively complete the unloading decision and scheduling allocation of tasks and resources.

关 键 词：物联网 任务卸载 联合决策 资源分配 边缘计算 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]