本文選題：異常檢測　切入點：聚類　出處：《計算機科學》2014年08期

【摘要】：提出一種基于聚類的路徑偽造檢測方法。該方法將相鄰時刻路由路徑的變化集作為檢測對象,以前綴地址所屬國家為依據(jù),對路徑變化集進行聚類,引入各變化自治域的AS鏈接概率偏離度、中間國家出現(xiàn)概率和中間國家地理偏離度的定義,在此基礎(chǔ)上引入路徑級異常檢測指標,綜合利用這些指標檢測路由中的路徑偽造異常行為。選用真實的路徑偽造事件數(shù)據(jù)進行實驗,結(jié)果表明該檢測方法較以往的檢測方法更為有效、可行。
[Abstract]:A method of path forgery detection based on clustering is proposed.In this method, the change set of the routing path at adjacent times is taken as the detection object, and the path change set is clustered according to the country of prefix address, and the deviation degree of as link probability is introduced in each change autonomous region.Based on the definition of probability of occurrence in intermediate countries and geographical deviation degree of intermediate countries, the path-level anomaly detection indexes are introduced to detect path forgery anomaly behavior in the route synthetically.The experimental results show that this method is more effective and feasible than the previous methods.
【作者單位】： 電子工程學院;
【基金】：安徽省自然科學基金(1208085QF107)資助
【分類號】：TP393.04
，

本文編號：1718153