基于订阅匹配树的发布订阅系统高效内容匹配的关键技术研究

项目名称： 基于订阅匹配树的发布订阅系统高效内容匹配的关键技术研究

项目编号： No.61272544

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： 乔保军

作者单位： 河南大学

项目金额： 80万元

中文摘要： 基于内容的发布订阅系统已广泛地应用于开发大规模分布式系统，高效的基于内容的发布订阅系统必须要解决高效的匹配问题，在这种系统中匹配结构和匹配算法的设计是一个关键问题。本课题的主要研究工作是：针对基于内容的发布订阅系统如何设计一种高效的匹配树结构- - 订阅匹配树（Submatree），将订阅过滤器中的一个谓词建模为一个区间间隔，将发布的事件值建模为一个数据点,用订阅匹配树索引所有的订阅间隔，从而将发布内容与订阅过滤器中谓词的匹配问题转换成一个数据点在区间上的刺查询问题，降低匹配操作的复杂度；基于订阅匹配树，进一步研究基于成本模型的间隔划分和Submatree的合并等算法，构建一个高效的基于内容的分布式发布订阅系统架构- - DISCOPS，使该系统架构具有较高的匹配效率，较少的网络通信流量和较低的传输延迟。

中文关键词： 订阅匹配树；发布订阅系统；内容匹配；间隔划分；

英文摘要： Publish/Subscribe system based on content is often applied in a distributed environment, and the system must resolve the problem of efficient content matching. The efficiency of matching structure and algorithm is the key issue for content publish/subscribe systems.The main research of this proposal is designing a efficient subscription match tree structure(Submatree) and framework(DISCOPS) for distributed content publish/subscribe sysems. In this framework, we model a predicate in each subscription filter as an interval and published content value as a data point. Given the model, matching content value against predicates can be seen as a stabbing query of point against inteals and reducing the complexcity of matching operation. Based on the Submatree, we will research the interval division and Submatree maintaining which is based on cost model.Using the subscrition match tree and the associated techniques which we proposed, we can construct a efficient distribued content Publish/Subscribe system which can significantly outperform two counterpart with low cost and fast forwarding.

英文关键词： Subscription Match Tree；Publish/Subscribe System；Content Matching；Interval Division；