ThriftyEdge: Resource-Efficient Edge Computing for Intelligent IoT Applications
ThriftyEdge: Resource-Efficient Edge Computing for Intelligent IoT Applications
复制标题
ThriftyEdge：适用于智能物联网应用的资源高效型边缘计算
DOI：
10.1109/mnet.2018.1700145
复制
发表时间：
2018-01-01
期刊：
影响因子：
9.3
通讯作者：
Xu, Jie
中科院分区：
文献类型：
--
作者：
Chen, Xu;Shi, Qian;Xu, Jie
在本文中，我们提出了一种新的资源高效边缘计算范式，用于新兴的智能物联网应用，例如用于精准农业、电子健康和智能家居的飞行自组织网络。我们设计了一种资源高效的边缘计算方案，使得智能物联网设备用户可以通过在本地设备、附近的辅助设备和附近的边缘云之间进行适当的任务卸载来很好地支持其计算密集型任务。与现有的移动的计算卸载研究不同，本文从资源效率的角度出发，设计了一种有效的计算卸载机制，包括延迟感知的任务图划分算法和最优虚拟机选择方法，以最小化智能物联网设备的边缘资源占用，同时满足其QoS要求。性能评估证实了所提出的资源高效的边缘计算方案的有效性和上级性能。
In this article we propose a new paradigm of resource-efficient edge computing for the emerging intelligent IoT applications such as flying ad hoc networks for precision agriculture, e-health, and smart homes. We devise a resource-efficient edge computing scheme such that an intelligent IoT device user can well support its computationally intensive task by proper task offloading across the local device, nearby helper device, and the edge cloud in proximity. Different from existing studies for mobile computation offloading, we explore the novel perspective of resource efficiency and devise an efficient computation offloading mechanism consisting of a delay-aware task graph partition algorithm and an optimal virtual machine selection method in order to minimize an intelligent IoT device's edge resource occupancy and meanwhile satisfy its QoS requirement. Performance evaluation corroborates the effectiveness and superior performance of the proposed resource-efficient edge computing scheme.