資源描述：
《基于密度演化與有限碼長(zhǎng)分析的低密度校驗(yàn)碼構(gòu)造》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在教育資源-天天文庫(kù)。
1、CONSTRUCTIONOFLOW-DENSITYPARITY-CHECKCODESBASEDONDENSITYEVOLUTIONANDFINITE-LENGTHANALYSISABSTRACTAsakindofchannelcodingtechniquewithwideapplication,LDPCcodeshavebeenadoptedinDVB-S2and802.16e.Sincetheperformancecanachieveorexceedturbocodeswhileencodinganddecodingcomplex
2、ityisquitelowduetothesparsenessofparitymatrix,LDPCcodeshavebecomethefocusofcodingtheory.IncomparisonwithregularLDPCcodes,irregularcodescanachievebetterperformanceintermsofnoisethreshold,whichislargelyaffectedbydegreedistributionpair.Usingdensityevolution,theperformance
3、ofcodeswithsamedegreedistributioncanbedetermined.Thus,degreedistributionpaircanbeoptimizedonthebasisofdensityevolution,whichmakesitpossibletofindgoodcodes.However,optimaldegreedistributionsalwaysresultinpoorerrorfloor,soitisnecessarytoaddconstraintondegreedistributionp
4、airsandtakefinite-lengtheffectsintoconsideration.Ontheotherhand,forparticulardegreedistributionpair,thoughcodesunderrandomconstructionmayhavesimilarperformance,theparitymatrixshouldhavewell-designedstructureforthesakeofimplementation.Withstudyofaboveissues,LDPCcodeswit
5、hgoodperformanceandlowimplementationcomplexitycanbeconstructedandappliedincommunicationsystemsforerror-correcting.Inthispaper,firstlydecodingalgorithmsincludingbeliefpropagation,min-sumalgorithmisanalyzedunderdensityevolution.Thendensityevolutionoflayereddecodingisdedu
6、ced,andperformancecomparisonofvariousdecodingalgorithmsismadeintermsofthresholdandconvergence.Thusoptimalscalefactorformodifiedmin-sumalgorithmisobtainedanditisshownthatwithlayereddecoding,iterationtimescanbereducedbyalmosthalf.Inaddition,withfinite-lengthanalysis,thei
7、mpactonerrorfloorbystructureofparitymatrixisstudied.Sinceerrorfloorismainlycausedbycycleswithlowextrinsicmessagedegree,someconstraintsonvariabledegreedistributionareputforward,whichcanbeusedtooptimizedegreedistributionpair.Finally,withconsiderationofencodinganddecoding
8、complexity,aconstructionmethodofirregularLDPCcodesispresented.Withconstraintonnumberofdegree-2variablenodesandapproxi
