Abstract

With the development of Mobile Edge Computing, driverless, 5 G, and related techniques, Edge-Cloud based Urban Intelligent Video Systems are extremely promising to support public safety through powerful analysis and timely response. Furtherly, flexible Unmanned Ground Vehicles(UGVs), which are equipped with edge devices, can enhance these edge systems to withstand these abnormalities: natural disasters, abnormal crowd flows, and other emergencies. In this regard, as a critical issue in edge systems, task placement in these systems needs to consider these “mobile” edge nodes: ICVs(UGVs). Therefore, a novel and effective framework named Optimized Centroids K-means based Task Placement framework is proposed: we firstly involve the clustering approach to optimize initial centroids as the positions of ICVs in terms of Edge Nodes, various typical optimization methods can be utilized to place related edge tasks effectively. The experimental results demonstrate that our novel framework has a great improvement over several existing typical strategies and supports multiple optimization methods well in this paper.

中文翻译：

---

基于边缘云的城市智能视频系统中的 UGV 感知任务放置

摘要

随着移动边缘计算、无人驾驶、5G及相关技术的发展，基于边缘云的城市智能视频系统非常有希望通过强大的分析和及时的响应来支持公共安全。此外，配备边缘设备的灵活无人地面车辆（ UGV ）可以增强这些边缘系统以抵御这些异常情况：自然灾害、异常人群流动和其他紧急情况。在这方面，作为边缘系统中的一个关键问题，这些系统中的任务放置需要考虑这些“移动”边缘节点：ICV（UGV）。因此，提出了一种新颖有效的框架，称为基于优化质心 K 均值的任务放置框架：我们首先采用聚类方法来优化初始质心作为ICV在边缘节点上的位置，可以利用各种典型的优化方法来有效地放置相关的边缘任务。实验结果表明，我们的新框架比现有的几种典型策略有了很大的改进，并且很好地支持了本文中的多种优化方法。