Web使能的物端计算系统  

Web Enabled Things Computing System

作　　者：彭晓晖 张星洲 王一帆[1,2] 朝鲁 Peng Xiaohui;Zhang Xingzhou;Wang Yifan;Chao Lu(Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190;University of Chinese Academy of Sciences,Beijing 100190)

机构地区：[1]中国科学院计算技术研究所,北京100190 [2]中国科学院大学,北京100190

出　　处：《计算机研究与发展》2018年第3期572-584,共13页Journal of Computer Research and Development

基　　金：中国科学院率先行动"百人计划"(Y704061000)~~

摘　　要：近年来兴起的边缘计算试图将部分计算从云端移到设备端,从而减少云端计算负载和网络传输负载.物端计算系统是边缘计算系统中面向物理世界的终端设备组成的计算系统.由于物端设备具有多样性,设计一个统一的体系结构来支持物端智能应用十分具有挑战.现代Web系统的体系结构是解决多样性的有效方案之一,但由于大部分物端设备的资源受限的特性,应用Web体系结构十分困难.1)阐述了现代Web系统、边缘计算系统和物端计算系统的概念,从组成物端计算系统的设备多样性和资源受限特性出发分析其面临的挑战;2)针对这些问题和挑战调研了一些基于REST的用于边缘计算系统的应用层协议;3)详细调研和评估了4个Web系统代表性脚本语言,总结了一些试图将这些语言应用于物端设备的工作;4)调研了传统嵌入式系统的调试技术.通过调研得出结论:目前的物端计算系统虽然市场规模巨大,但是仍未形成高效的、统一的体系结构来支撑人工智能应用的大量部署;5)列出了物端计算系统的一些重要研究方向,包括统一的体系结构、高能效Web、支持物端智能和物端调试技术.The rising edge computing paradigm tries to shift some computing tasks from cloud to devices recently,which reduces the computing load of cloud and traffic load of the Internet.The things computing system consists of the devices which are physical world oriented with physical functionalities.It is a great challenge to design a unified system architecture for things computing system because of the system diversity.The architecture of the modern Web system is an efficient solution for the diversity issue.However,due to the resource-constrained feature extending the Web architecture to the things computing system is also very difficult.In this paper,we first introduce the concept of edge computing system and things computing system,and summarize the challenges brought by diversity and resource-constrained features of things computing system.Then,a detailed study of the state-of-the-art technologies,including REST principle,script languages and debugging technique for extending the Web to things computing system,is presented.Most of the related work tried to modify the“Uniform Interface”principle to adapt to edge system.We conclude from the examined literature that things computing system is a massive market,but there is still no unified system architecture which supports both the Web and intelligence.Finally,we present some future research directions for things computing system including the unified system architecture,efficient Web technologies,supporting intelligence and debugging techniques.

关 键 词：物端计算系统 资源受限 万物互联 表述性状态转移原理 物端设备 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]