基于数据挖掘的网络业务流分析方法

Network traffic analysis method based on data mining

为了从业务角度对网络的性能进行评价和优化,提出了一种新的网络业务分析方法——具有时态路径约束的关联规则挖掘分析方法.该方法以网络业务为分析粒度,以与网络业务流相关的时态属性和路径属性为约束条件,对已经积累的反映网络状况的海量历史数据进行挖掘分析.在进行关联规则挖掘时,利用频繁数据项集的性质,通过引入事务标号,在求出候选频繁项集的同时也求出其支持度,避免了为求支持度而进行的扫描数据库运算,极大提高了挖掘的效率和速度.实验结果表明,进行挖掘分析的数据量越大,该方法的性能和效率就越好.

In order to evaluate and optimize the network performance from the view of the network traffic, a novel network traffic analysis method called time and path restrained association rules mining （TPRAR） is proposed. This method regards the network traffic as the analysis granularity and analyzes a mass of the historical data reflecting the network status by using data mining. During the course of mining, the time attribute and the path attribute related to the network traffic are regarded as the restraint conditions and the transaction ID is used to get the support of candidate frequent itemsets too while getting candidate frequent itemsets based on the character of frequent itemsets. This avoids scanning the database to get its support and the efficiency and the speed of mining are greatly improved. Experimental results indicate that the more the data are, the better the performance and the efficiency of TPRAR are.