基于信誉的P2P-VoIP垃圾语音过滤模型  

Reputation-based filtering model for P2P-VoIP spam

作　　者：王非[1] 莫益军[1] 黄本雄[1] 

机构地区：[1]华中科技大学电子与信息工程系,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2008年第8期62-65,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：湖北省重点科技攻关项目(2006AA102A04)

摘　　要：利用客观统计的用户话务模型以及用户的主观评价,提出了基于用户信誉的分布式VoIP垃圾语音过滤模型DAVS.该模型定义包括用户反馈和用户话务统计特性的信誉模型,并采用基于DHT网络的纯分布式结构保存、共享所有信誉数据.实验结果表明,DAVS模型可以准确、稳定地过滤P2P-VoIP网络中约98%的垃圾语音.By using objective user＇s statistical call model and subjective userts evaluation, a novel schema, DAVS （distributed anti voice spam） against spam calls on the P2P-VoIP networks, was proposed. The DAVS defines reputation model which based on the user＇s recommendation and statistic character of user＇s call model. Further, DAVS was constructed as the pure distributed architecture, DHT （distributed hash table）. All reputation data were stored in such DHT network, and were shared among all clients. The experiment result showed that DAVS could detect and filter about 98 % spam calls accurately and stably.

关 键 词：语音过滤 P2P-VoIP 垃圾语音 信誉 Dempster法则 分布式哈希表 

分 类 号：TN912.3[电子电信—通信与信息系统]