面向物联网的不确定情境感知计算研究

项目名称： 面向物联网的不确定情境感知计算研究

项目编号： No.61272177

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 林树宽

作者单位： 东北大学

项目金额： 80万元

中文摘要： 物联网是智能感知、识别技术与普适计算、泛在网络的融合应用，被称为继计算机、互联网之后世界信息产业发展的第三次浪潮。目前，我国已经将其列为国家战略，并纳入今后国家重点发展和扶持的新兴战略性产业之一。在物联网的三层架构中，情境感知计算是第三层- - 智能处理层的核心，是物联网软件处理平台的重要组成部分。但作为新兴技术领域，目前的情境感知计算在面向物联网应用时尚存在许多问题。本课题在深入分析这些问题的基础上，突破已有的情境感知计算技术的局限，针对物联网中情境数据的特点，提出了基于不确定频繁情节挖掘、双向概率推理及不确定SVM学习建模的情境感知计算的新思路，在面向物联网的情境感知计算中增加了可回溯性、容错性和不确定数据处理能力。本课题力争在情境感知计算的情境数据建模、情境推理和情境挖掘算法等方面做出创新性工作，最终将建立一个集研究成果于一体的原型系统

中文关键词： 情境感知计算；不确定频繁情节挖掘；不确定情境SVM学习建模；时态和多维属性约束；双向概率推理

英文摘要： The Internet of things is a fusion application of intellisense, identification technology, pervasive computing and ubiquitous network, and is called the third tidal wave of information industry development after computer and Internet in the world. At present, it has been upgraded to a national strategy of China as an emerging industry of strategic importance which will be specially supported. In the three-tier architecture of the Internet of things, context-aware computation is the kernel of the third layer, the processing layer, and an imortant part of the software platform of the Internet of things. However, as an emerging technical field, the current context-aware computation has many problems in applications of the Internet of things. The project analyzes the problems deeply. On this basis, it will breakthrough the limitation of existing tecniques of context-aware computation, and proposes new ideas of uncertain frequent episode mining, bidirectional probabilistic reasoning and uncertain SVM modelling according to characteristics of context data. It will have retrospective property, fault tolerance and uncertain data processing capability. The project will finish creative work at context data modelling, context reasoning and context mining, and will develop a prototype system including all research achieveme

英文关键词： context-aware computation；uncertain frequent episode mining；SVM learning and modeling of uncertain context；temporal and multi-attribute constraints；bidirectional probability reasoning