版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
1、analyzing the impact of granularity on ip-to-as mappingpresented by baobao zhangauthours:baobao zhang, jun bi, yangyang wang, jianping wu1 introductionndoing?nmap the ip address to the as that uses the ipnmeaningnhelp network managers diagnose network failurendiscover the as-level topology with trac
2、eroutensome other applications that need to map ip to asan example2 data collectionndata sourcentraceroute data (from caida)nbgp routing table (from routeviews)nprocessing into pairsnextract the prefixes and as paths from routing tablesnextract the destination ips and ip paths from traceroute datanf
3、ind the longest matching prefix for the destination ipnthe ip path associated with the destination ip and the as path associated with the longest prefix form one pairnorigin ip-to-as mappingnextract the prefixes and its origin ases from routing tablesnmap every prefix to its origin asdata collection
4、ndate: 04/22/2010nduring: one day3 methodologyndefinitionnexact matchnambiguous matchnmismatchnmethodsnprefix-granularity method (pgm)nip-granularity method (igm)nprefix-granularity limit method (pglm)nhierarchical mapping system (hms)nassumptionnthe traceroute path is consistent with the bgp as pat
5、h.methodsnprefix-granularity method (pgm)ni.e. maos methodnbind many ip addresses into one prefixnmap one prefix to many ases by setting thresholdntight couplingnprosncan modify the incorrect mappings for the ips that dont appear in the training dataset nconsnmistakenly modify the originally correct
6、 mappings for the ips that dont appear in the training dataset. (tight coupling)nthreshold. miss to modify the incorrect mappings for the ips that appear in the training datasetnthreshold. bring about ambiguous mappingsmethodsnip-granularity method (igm)nwe propose it for the first timenmap one ip t
7、o one only asnloose couplingnprosneliminate the ambiguous mappingsnconsnonly can modify the mappings for the ips that appear in the training dataset.methodsnprefix-granularity limit method (pglm)none fictitious methodnthe limit of pgm. set the threshold =0nit is only used to be comparedmethodsnhiera
8、rchical mapping system (hms)ncombine the igm with pgmnthree levels (/32 level, /24 level, origin level)nfirstly look up in the /32 level mapping, then /24 level mapping, finally the origin level mappingnprosncomplement the strength of tight coupling and loose coupling nconsn * inherit the characteri
9、stic of ambiguity from pgm4 evaluationndatasetevaluationntraining accuracyevaluationnvalidation accuracyevaluationncompare trained mapping with the origin mappingevaluation5 classification tree analysisnmotivationnquantify the pros and cons for the igm and pgmnanalyze the obstacles in the way of imp
10、roving the accuracy for the igm and pgm nother potential findingsnconstructing classification treetable 7 the improvement gained by correcting the mapping of the types for the pgm vds1gainvds2gainvds3gainvds4gaintype10.00%0.00%0.00%0.00%type20.71%0.02%0.27%0.05%type314.25%8.47%8.15%10.30%type40.00%0
11、.00%0.00%0.00%type52.37%1.55%0.35%2.47%type60.00%0.00%0.00%0.00%type70.80%1.57%1.47%1.05%type8(base)-0.29%(5.66%)-0.64%(7.34%)-0.15%(6.79%)-0.33%(6.20%)type1-2(base)0.00%(1.06%)0.00%(0.61%)0.00%(0.58%)0.00%(1.92%)type2-20.36%0.06%1.01%0.25%type3-20.42%1.12%22.29%15.08%type4-20.00%0.00%0.00%0.00%type
12、5-20.45%0.17%0.25%3.30%type8-2(base)0.00%(2.93%)0.00%(2.38%)-0.03%(2.22%)-0.01%(0.15%)type-all19.85%12.87%35.18%32.94%5.1 quantify the pros and cons for the igm and pgmnpros and consn(+) modify the incorrect mappings for the ips that dont appear in the training dataset (type 8-2, 1-2 for pgm, nothin
13、g for igm)n(-) mistakenly modifies the originally correct mappings for the ips that dont appear in the training dataset. (type 2-2 for pgm , nothing for igm)n(-) miss to modify the incorrect mappings for the ips that appear in the training dataset (type3 for pgm and igm)nquantifyingnfor pgm, base(ty
14、pe8-2)+base(type1-2)-gain(type2-2) is positive. 3.63%, 2.93%, 1.79% and 1.81% npgm(gain(type3)-igm(gain(type3) . 14.00%, 8.38%, 7.94% and 9.81% nconclusionnthe igm is superior to the pgm5.2 analyze the obstacles in the way of improving the accuracy for the igm and pgmnigmntype 7. (ips do not appear
15、in the training dataset) npgmntype 3. (ips appear in the training dataset, but miss to modify due to the tight coupling)ntype 3-2. (ips do not appear in the training dataset) 5.3 other findingsnthe limit of validation accuracy1-gain(type2) -gain(type3)-gain(type5)nfor igm98.87%,97.96%,98.43% ,98.96% nfor pg
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 二零二五年度二手房买卖垫资担保合同2篇
- 灭火器使用课程设计
- 2024模特经纪公司艺人经纪服务合同3篇
- 2025版绿色金融股权质押与环保风险控制担保服务协议2篇
- 焦虑状态护理常规
- 2024某公司电子商务事业部虚拟现实购物体验合作协议书3篇
- 2024年跨境电商服务合作协议
- 成都职业技术学院《创新综合实践》2023-2024学年第一学期期末试卷
- 2024年版建筑工程招投标与合同文本2篇
- 2024年项目中介服务协议细则版B版
- 2024-2025学年北师版八年级物理上册期末考试综合测试卷
- 2023-2024学年广东省广州市白云区八年级(上)期末数学试卷及答案解析
- 毕业设计工程造价预算书
- 幼儿园课件-神奇的中草药
- 起重机零配件(易损件)清单
- 锥坡工程量计算
- 植物园设计规范
- 深圳市建设工程施工围挡图集(试行版_下半部分).pdf
- 热水器3c安全试验报告及第三方检测报告dsf65mx ts tx ws wx ys yx ms
- 南洋电工GSB1A型16锭高速编织机使用说明书
- 大管轮见习记录簿范本汇总
评论
0/150
提交评论