一种新的网络故障诊断方法

ISSN100021239/CN1121777/TPJournalofComputerResearchandDevelopment42(11):19281933,2005:2004-05-25;:2004-11-01:(60273035);(200522006)FTFD1,212221(210003)2(210094)(liqianmu@1261com)ANewNetworkTroubleshootingMethodFTFDLiQianmu1,2,XuManwu1,YangYun2,ZhangHong2,andLiuFengyu21(StateKeyLaboratoryofComputerSoftwareandNewTechnology,NanjingUniversity,Nanjing210003)2(DepartmentofComputerScienceandTechnology,NanjingUniversityofScienceandTechnology,Nanjing210094)AbstractThroughformalizedanalysisofthesituationofnetworkandthetargetoffaultdiagnosis,anewmethodFTFDfornetworktroubleshootingbasedonfuzzyeventisproposed1Byintroducingsituation2de2tectionfunction,FTFDcancharacterizecomplicatedfuzzyfaultwithaccuratemathematicsconversion,andabnormaldegreecanbedefinedbythevectorofprobabilitywithbelieffunctions1Themethodcaneffec2tivelyreducefalsepositivesandnegativepositives1Itaimstobeappliedtoreal2timefaultdiagnosis1Theoperationalprototypicalsystemdemonstratesitsfeasibilityandgetstheeffectivenessofreal2timefaultdiag2nosis1Keywordsnetworktroubleshooting;fuzzyevent;realtimedetection,(FTFD),,,1FTFD,1,1;;TP3091,,,,1[1,2]:,,1[3],SPRTLR1WaldSPRT,1LR,,1Iserman[4],,1[5],,1Siraj[6]2001(fuzzycognitivemap)©1995-2007TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.,1FTFD,:;,,,;,12FTFD;3;4FTFD;5NS2BerkeleyETR;12FTFD:,,,FTFD,1FTFD()(),1:Fig11Flowchartsofthetroubleshootingsystem11,,,1,11,,FTFD,1FTFD:,,;,,,13S(E),E=(e1f1,e2f2,,enfn),,ei,fi1(A,E,G),G:AEE1G,A1,(),SS1E,P1,ei,i+1Ei(:i=1,2,,s-1),ei,i+1jfji,i+1,Pjfjp,31fjp=s-1i=1fji,i+1,Pj1,11fjp=s-1i=1fji,i+1,Pj1,1P1fjp=mini=1,2,,s-1{fji,i+1},Pj1fjp=maxi=1,2,,s-1{fji,i+1},Pj9291:FTFD©1995-2007TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.1;1P;,,1,,:S,f1(E),f2(E),,fp(E)1,:minpi=1w0ifi(x)s1t1fi(x)i,i=1,2,,p,xX,,Xx,i,w0i,:pi=1w0i=1pi=1w0i=114FTFD1ŽA[7,8]P(ŽA)þUŽA(x)dP=E[ŽA(x)],ŽA(x)f(x),P(ŽA)þUŽA(x)f(x)dx,f(x)1U={xi|i=1,2,,m},P(ŽA)=mi=1ŽA(x)p(xi),,p(xi)xi111ŽAŽA:P(ŽA)=10P(ŽA)d1(A)11,ŽA,,ŽAP^(ŽA)=1mmj=1P(ŽAj)1FTFD,,P^(ŽA)P(ŽA)1411[9]1DT(X)=ni=1mP2j=1xi,2j-1+1xi,2j+1,m,ni=1(m-1)P2j=1xi,2j-1+1xi,2j+1,m,,X=(X1,X2,,Xm)T,Xi1412FTFDFTFDSNMP,,,1:Step11(e1f1,e2f2,,enfn)1Step21DT(X),T1DT,0T1,,flag=0;,flag=11Step31fii,i1(xi)=fi-i,ŽAxi1:(ui)=1,ui0,0,ui01Step41(ui)fi,(e1(u1),e2(u2),,en(un))1Step51flag=0,(),,,,Step15;,Step61Step61,ŽA:P(ŽA)=ni=1[p(ei)(i)]1Step71FTFD,P(ŽA),1413,,t0,P|1mmj=1ni=1ŽA(xi)-P(ŽA)|tm22t0exp-u22du=1-,03912005,42(11)©1995-2007TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.,P(ŽAj)=ni=1p(xi)(ŽA(xi)-j),1-1,,,;t1,1-,m,P(ŽA)P^(ŽA)=tm1(P(ŽA))1mmj=1ni=1p(xi)(ŽA(xi)-j)-tm,1mmj=1ni=1p(xi)(ŽA(xi)-j)+tm1414(),,1M1,O(Mn)1,O(Mn),,15,,NS2BerkeleyETR1AB,10Mbps,2Mbps1AB,10300ms,1000B181duration:1bfro:1bto:1state:1010,SF,REJ,S0,S1,S2,S3,S4,SS,SH,SHR,OOS1,OOS21flag:,0(),1()1rin,rout,rin1routrin,rin{-3,-2,-1,0,1,2,3},NL,NM,NS,ZO,PS,PM,PL1104,,ŽA(xi)=1P1+019P2+017P3+015P41e:Width(p)=minep{B(e)}1J(p)=maxep{D(e)}-minep{D(e)}H(p)1InputError(e)=ifInErrorsTotallnputPackets1Discards(e)=ifOutDsTotal1O1Packs1Receive=ifInReceivesTotal1In1Packets12:Fig12Growthofdatabasesizeforbuildinganormaldatabase12,,,,,5000,1,(),,1,p,1,:015,ŽA(xi)=1P1+1P2+1P3+1P4,P(ŽA)=ni=1p(xi)(ŽA(xi)-)=p+(1-p)p+(1-p)2p+(1-p)3p;015017,ŽA(xi)=1P1+1P2+1P3+0P4,P(ŽA)=ni=1p(xi)(ŽA(xi)-)=p+(1-p)p+(1-p)2p;017019,ŽA(xi)=1P1+1P2+0P3+0P4,P(ŽA)=ni=1p(xi)(ŽA(xi)-)=p+(1-p)p;1391:FTFD©1995-2007TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.0191,ŽA(xi)=1P1+0P2+0P3+0P4,P(ŽA)=ni=1p(xi)(ŽA(xi)-)=p1P(ŽA)=10P(ŽA)d=0150ni=1[p(xi)(ŽA(xi)-)]d+017015ni=1[p(xi)(ŽA(xi)-)]d+019017ni=1[p(xi)(ŽA(xi)-)]d+1019ni=1[p(xi)(ŽA(xi)-)]d=0150P(ŽA)d+017015P(ŽA)d+019017P(ŽA)d+1019P(ŽA)d=p+019(1-p)p+017(1-p)2p+015(1-p)3p10101,100,ŽB,ŽC1004,106,P(‰B)=019168532,P(ŽC)=011765841,ETR(P(‰B)=019171246,P(ŽC)=011832458)1SogTip,341Fig13AnomalyprofileforarunoftheSog13Sog,Fig14AnomalyprofileforarunoftheTip14Tip,,,,,,,16,,,1FTFD,:,;,,;,;,,,11D1Gavalas,D1Greenwood,M1Ghanbari1AdvancednetworkmonitoringapplicationsbasedonmobilePintelligentagenttechnolo2gy1ComputerCommunications,2002,34(23):7207302R1Tagliaferri,A1Eleuteri,M1Meneganti,etal1Fuzzymin2maxneuralnetwork:Fromclassificationtoregression1SoftCom2puting,2001,43(5):69763M1Basseville,I1V1Nikiforov1DetectionofAbruptChangesTheoryandApplication1EnglewoodCliffs,NewYork:PrenticeHall,19934R1Iserman1Processfaultdetectionbasedonmodelingandestima2tionandknowledgeprocessingTutorialpaper1Automatica,1999,29(4):8158355P1Bullell,D1Inman1Anexpertsystemfortheanalysisoffaultsinanelectricitysupplynetwork:Problemsandachievements1ComputerinIndustry,1998,24(37):1131236A1Siraj,S1M1Bridges1Fuzzycongnitivemapsfordecisionsup2portinanintelligentintrusionsystem1ISFAWorldCongressandthe20thNAFIPSIntlConf1,TelAviv,20017J1E1Dickerson,J1A1Dickerson1Fuzzynetworkprofilingforintrusiondetection1In:Proc1IntlConf1NorthAmericanonAr2tificialIntelligence1LosAlamitos,CA:IEEEComputerSociety23912005,42(11)©1995-2007TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.Press,200213013068J1E1Dickerson,J1Juslin1Fuzzyintrusiondetection1ISFAWorldCongressandthe20thNAFIPSIntlConf1,TelAv