一种基于关联规则挖掘的IP定位方法  

Method for IP geolocation based on association rule mining

作　　者：王婷[1] 宋俊德[1] 宋美娜[1] 

机构地区：[1]北京邮电大学计算机学院,北京100876

出　　处：《东南大学学报（自然科学版）》2015年第4期657-662,共6页Journal of Southeast University：Natural Science Edition

基　　金："十二五"国家科技支撑计划资助项目(2012BAH01F02;2013BAH10F01;2013BAH07F02);国家自然科学基金资助项目(61072060)

摘　　要：针对IP定位数据库如何及时更新IP定位数据并同时保证其定位准确率和覆盖率的问题,提出一种基于关联规则挖掘的IP定位方法(IPGEL).首先,结合traceroute网络测量数据和已有IP定位数据库中IP定位数据,获取关联规则挖掘的路由路径数据集.其次,利用国内互联网分层结构的路由特性和关联规则挖掘,在对路由路径数据集进行数据分析的基础上对数据集内的IP分类并定位,同时利用马尔可夫模型对数据集外的IP进行补充定位.最后,使用交叉对比验证和地标验证相结合的方法验证IPGEL方法的准确性.结果表明,IPGEL方法在提高IP定位准确率和覆盖率的同时有效提高了已有IP定位数据库的可信度.To update the IP geolocation information in time and improve the geolocation accuracy and coverage ratio for the IP geolocation database, a statistical IP geological location scheme （IP- GEL） is proposed based on association rule mining. First, the network measurement results are com- bined with IP geolocation databases＇ data to obtain a dataset for association rule mining. Secondly, the IPs in the dataset are classified and geolocated using the routing statistics of internet hierarchical structure in China and the association rule mining. Markov model is also used to supplement the geo- location for the IPs out of the dataset. Then both cross validation and landmarks validation are in- cluded in the evaluation to demonstrate the accuracy of the IPGEL method. The experimental results show that the IPGEL method increases the accuracy and the coverage ratio of the IP geolocation and improves the reliability of the existin~ IP ~eolocation databases at the same time.

关 键 词：网络测量 IP定位 IP定位数据库 关联规则挖掘 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]