基于svm的智能天线算法分析

《基于svm的智能天线算法分析》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、重庆邮电大学硕士论文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