摘要
在移动通信网络环境中,如何合理地组织和存储移动对象的配置信息,从而有效地降低查询和更新代价是位置管理中的一个重要问题.将数据挖掘应用到移动计算环境中是一项具有挑战性的研究课题,具有广阔的应用前景.区域划分能够优化位置数据库的拓扑结构,有效地降低查询和更新代价.但是随着时间的迁移,用户的移动模式会发生改变,导致原有区域的划分与当前的移动模式不符,因此产生了动态区域划分这一亟待解决的重要问题.聚类可以很好地解决区域划分问题,而对于动态区域划分问题,如果仍然采用聚类来解决,就等于重新划分,没有充分利用原有划分的信息,所需代价很大.提出了一种增量的聚类算法来解决动态区域划分问题.该方法以较小的代价调整原有划分,使得新得到的划分仍然满足区域划分所需满足的条件.
How to effectively organize and store the profile of moving objects in a mobile environment, where then can effectively lower the paging and update cost, is an important problem in location management. Combining data mining into the mobile environment is a challenging research task, which has broad applications. Zone partition can effectively optimize the topology of location databases and efficiently reduce the cost of location paging and location update. But with the evolving time, the mobile users’ moving patterns may change, so the original partitions may not match the current moving patterns. Thus one of the important problems, which need to be solved, is how to partition the zones dynamically. Clustering method can solve the static zone partition well, but face with the dynamic zone partition problem. If the clustering method is still used to solve this problem, it means that the zones are partitioned again from scratch, which doesn’t utilize the original partitions and need great cost. In this paper an incremental clustering method is provided to solve the dynamic zone partition problem, which adjusts the original zone partitions with less cost and guarantees all the conditions needed for zone partition problem in the meanwhile.
出处
《软件学报》
EI
CSCD
北大核心
2004年第9期1351-1360,共10页
Journal of Software
基金
国家高技术研究发展计划(863)
国家重点基础研究发展规划(973)
北京大学-IBM创新研究院项目资助~~
关键词
增量聚类
数据挖掘
位置数据库
位置管理:移动通信
incremental clustering
data mining
location database
location management
mobile communication
