适用于GIDS报文分类的P-HiCuts算法  

P-HiCuts algorithm for GIDS parket classification

作　　者：龚俭[1] 魏薇[1] 周鹏[1] 

机构地区：[1]东南大学计算机科学与工程学院,南京210096

出　　处：《哈尔滨工业大学学报》2008年第3期448-452,共5页Journal of Harbin Institute of Technology

基　　金：国家重点基础研究发展规划资助项目(2003CB314803);国家自然科学基金资助项目(90104031);江苏省网络与信息安全重点实验室资助项目(BM2003201)

摘　　要：针对HiCuts算法在NIDS应用上存在着空间异常膨胀和决策树不平衡性的问题,提出了一种P-Hi-Cuts算法.P-HiCuts(Pruned HiCuts)对原报文空间分组算法进行改进,采用覆盖规则上提和非均匀切分的技术解决原有问题,从理论上减小了决策树深度.实验结果显示,改进后决策树深度空间占用缩小到原来的10%,分类速度也提升了13.71%.To overcome the unusual expansion of the space consumption and unbalance of the decision tree in the application of HiCuts algorithm in NIDS, a P-HiCuts algorithm is presented. P-HiCuts improves the original space-cutting algorithm through node-overlapped rules upward moving and asymmetrical space cutting. The experimental result shows that the new decision tree is shorter, its memory consumption is ten percents of the original one, and the system performance is promoted 13.71%.

关 键 词：报文分类 分类算法 NIDS HiCuts P-HiCuts 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]