資源描述：
《基于svm的智能天線算法分析》由會員上傳分享，免費在線閱讀，更多相關內(nèi)容在行業(yè)資料-天天文庫。
1、重慶郵電大學碩士論文AbstractSmartantennatechnologyisoneofkeytechnologiesinTD—SCDMAcommunicationsystemandtheresearchfocusincommunicationstechnologycurrently．SmartantennacansuppressinterferencesignalsbybeamformingandthustoimprovetheoutputSINRandcommunicationsystemcapacity．Therefore，Researchi
2、ngonthesmartantennaalgorithmissignificantandhasimportantpracticalvalue．SupportVectorMachine(SVM)isthelatestmachinelearningresearch，itonlytakesasmallamountofsampletobetestedonthesamedistributionofthesampleandhasgoodgeneralizationability,butalsodealwithhighdimensional，nonlineardata
3、andglobalconvergenceadvantages．Supportvectormachineshavebeenwidelyappliedtovariousareaofresearch，hasbecomeanewmethodinwirelesscommunicationsignalprocessing．Inthispaper,supportvectormachinesasasignalprocessingtoolapplytoresearchthesmartantennaalgorithm．Themainworkandinnovationofth
4、ispaperincludethat：Firstly,thepapergivesageneraloverviewofthebasicprinciplesofsmartantennaandclassicalalgorithmsofbeamformingandDOAestimation．VCdimensionandgeneralizationoftheVCboundary,lossfunctioninstructuralriskandsupportvectormachinesareincluded．Throughcomputersimulation，supe
5、riorcharacteristicsofclassificationinsupportvectormachineandfittingperformanceinsupportvectormachineregressionintwo-dimensionaldataarebeingshown．Secondly,itextendedsupportvectormachinestothecomplexplaneSOthatcanhandlethecomplexsignalfortheapplicationofsupportvectormachineinbeamfo
6、rmingandDOAestimationworkswellthecushion．Theoptimumbeamformerweightstranslateintosolvingapproximatelinearclassificationproblemsdealingwithsupportvectormachine．EstablishedbeamformerbytheLSVMandNSVMbasedonsupportvectormachines．Bysimulationanalysis，theresultsshowthat，Comparedwiththe
7、traditionalalgorithm，LSVMandNSVMusedinbeamformingalgorithmshasfasterconvergence，higherofoutputSINR，especiallyinthecaseofoverload．TheNSVMshowedbettershapingthantheLSVM，butslightlyhighercomplexity．Thirdly,solvingcoefficientsoftheARmodelbyusingtheSVMandgetthesignalspectrumofthedirec
8、tionofthemodel，adjustingtheparameterscan
