【作者】 施颖男；

【导师】 李德敏；

【作者基本信息】 东华大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 随着移动通信技术和计算机网络技术的不断发展和广泛应用，在不久的将来，移动计算机可以在任何时间、地点接入各类网络，以获取所需的数据信息，这种特殊的分布式计算环境被称为移动计算。它正受到越来越多的关注。 本课题主要以国家自然科学基金资助项目(70271001)—移动决策理论与移动决策支持系统研究作为研究的背景，所以我们首先需要在对整个系统环境-移动计算环境的特点和主要问题进行讨论的基础上，对该环境中存在的大量数据信息进行合理的分析、分类，然后针对各类数据分别进行有效的数据挖掘，从中提取出潜在的、有用的规则，用来为移动客户提供决策支持服务。 本文对以上任务的讨论顺序安排如下：首先是对移动计算环境的技术特点和难点进行讨论，包括移动计算的概念和典型系统模型、主要挑战、移动联网以及软件应用这几个大的方面；其次根据移动环境的移动特性把移动计算环境中的数据分为普通数据，时间数据，空间数据以及时空数据，提出了在移动计算环境中数据挖掘的一般流程；接下来分别对这四类数据进行挖掘算法的讨论：每一部分都是先介绍该类数据的挖掘方法研究现状，对于普通数据，针对我们已提出的一种挖掘算法-粗糙集算法(RS)，提出了对应的规则更新算法，对于后三种数据，本人根据其在移动计算环境中的特点分别提出了与移动用户相关的该类数据的一种具体的处理方法和算法流程图，包括基于移位连接方法的多属性时间序列的挖掘算法，基于Apriori算法的空间关联规则数据挖掘方法以及关于移动用户移动模式的时空数据挖掘方法，并用MATLAB对其中的规则更新算法和时间序列的挖掘算法这两方面进行了实例仿真。 在移动计算环境中通过对数据的挖掘和信息的处理，从而快速的移动计算中的规则提取与规则更新研究提取出对移动用户有价值的信息，不仅提供给移动用户日常的决策支持和紧急状况下的应急服务支持，而且简洁的规则信息相对原始的大量数据而言，可以大大的减少无线链路上的传输数据量，减少传输的时间，最重要的是节省了无线资源，降低了网络阻塞的情况，从而提高了整个移动计算系统的可用性和高效性。本课题的研究在军事、应急决策、医疗和股票等领域都有着广泛的应用价值。更多还原

【Abstract】 With the continuous developments and extensive applications of technologies in the fields of both mobile communication and computer network, mobile computers can access kinds of network at any time and any place to get the data information they need in the near future. This special distributed computing environment is called mobile computing. More and more attention has been put on it these years.In the context of National natural scientific fund-Research of mobile decision-making theory and its support system, we will take rational analysis, classify on the data information existing abundantly in this environment after an in-depth study about the characteristics and issues of mobile computing environment, and then do some effective data mining of all kinds of data to find the potential and useful rules, which can be used by the mobile clients for decision-making.The order of our discussions’ about these tasks is as follows: Firstly, we pay more attention to the characteristics and difficulties of its environment including the concept, typical system model, main challenges, mobile network connection and soft application. Secondly, according to mobile specialties of the environment we make the sort of data into four kinds: general data, time series, spatial data and time-spatial data, and present general processing of data mining. Lastly, we discuss the methods of data mining of these four kinds respectively: After the introduction of the actuality of data mining of every kind, an algorithm of rule updating based on Rough Set is given, then put forward the processing of data related to mobile users and flow chat according to characteristics of the other three kinds. It includes mining in multi-attribute time series based on shift and connect, mining in spatial associated rule based on Apriori and mining in mobile patterns of users. At the same time, we have emulations of algorithms of rale updating and time series mining with MATLAB.In mobile environment, extracting valuable information quickly through data mining provides daily decision-making support and emergency service-support for mobile users, and compact rule information compared to a great deal of original data can reduce the quantity of data translated in a radio link and can save time. Foremost, it saves the wireless resources, lightens congestion of bandwidth in wireless network, so that it will enhance the usability and high-efficiency of the whole mobile computing system. The research in this paper has broad application value in fields of military affairs emergency decision-makings medical treatments and stocks.更多还原