泊松过程在用户上下线模型中的应用  

An Application of the Poisson Process in a Model of on-off-Line Behavior of Network Users

作　　者：郭敏杰[1] 刘鹏[2] 刘芳[1] 雷振明[1] 

机构地区：[1]北京邮电大学网络体系构建与融合北京市重点实验室,北京100876 [2]中国标准化研究院食品与农业标准化研究所,北京100088

出　　处：《北京邮电大学学报》2013年第1期67-71,共5页Journal of Beijing University of Posts and Telecommunications

基　　金：国家自然科学基金项目(61072061);国家科技重大专项项目(2012ZX03002008);高等学校创新引智计划项目(B08004);中央高校基本科研业务费专项项目(2011RC0117)

摘　　要：提出了基于非齐次泊松过程的宽带拨号上网(ADSL)用户上下线模型.对用户上下线行为进行建模,并基于现网真实数据,使用模型计算用户24 h内的每时刻退出登录的平均概率,将模型的预测值与现网实测数据进行对比,分析了不同用户组退出登录概率一天内的变化和持续在线平均概率的变化.分析结果表明,提出的ADSL宽带用户上下线模型可以有效、量化地描述用户上下线行为,在ADSL宽带用户上下线行为的异常检测和舆情分析等领域具有较好的工程应用价值.A model of asymmetric digital subscriber line （ADSL） on-off-line behavior based on non-hom- ogeneous poisson process is presented. First, a model of ADSL on-off-line behavior is established by for- mula deduction. Then, the model is used to analyze and calculate the user average probability of logout based on real data from ADSL network. The value is predicted by using the model and analyzed by con- trasting with real data. Finally, the change of average probability of logout and continued online in differ- ent user group are given by the model. The analysis result shows that the model could give a quantitative description of On-off-line behavior of network users effectively, and will make great effort to anomaly de- tection for ADSL on-off-line behavior and public opinion survey.

关 键 词：上下线 用户行为 泊松过程 随机过程 

分 类 号：TP393.0[自动化与计算机技术—计算机应用技术]