large dimensional random matrix theory for signal detection and estimation in array processing

《large dimensional random matrix theory for signal detection and estimation in array processing》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫(kù)。

1、LARGEDIMENSIONALRANDOMMATRIXTHEORYFORSIGNALDETECTIONANDESTIMATIONINARRAYPROCESSINGJ.W.SilversteinyandP.L.CombetteszyDepartmentofMathematics,NorthCarolinaStateUniversity,Raleigh,NC27695,USA.zDepartmentofElectricalEngineering,CityCollegeandGraduateSchool,CityUniversityofNewYork,Ne

2、wYork,NY10031,USA.ABSTRACTsensorsaremodeledasobservationsoftherandomvec-torX(t)=AS(t)+N(t),t2[0;+1[,whereAisaInthispaper,webringintoplayelementsofthespec-pqcomplexmatrixdependingonthegeometryoftraltheoryoflargedimensionalrandommatricesandthearrayandtheparametersofthesignals,andi

3、sdemonstratetheirrelevancetosourcedetectionandassumedtohaverankq.bearingestimationinproblemswithsizablearrays.Thedetectionproblemistoestimateqfromtheob-Theseresultsareappliedtothesamplespatialcovari-servationofnsnapshots(X(ti))1inofthedatapro-ancematrix,Rb,ofthesenseddata.Itiss

4、eenthatde-cess.Undertheaboveassumptions,therandomvec-tectioncanbeachievedwithasamplesizeconsiderablytors(X(t))t2[0;+1[arei.d.withspatialcovariancema-lessthanthatrequiredbyconventionalapproaches.AstrixR=EX(0)X(0)=ARA+2I,whereIde-Sppregardstodeterminingthedirectionsofarrivals,it

5、isnotestheppidentitymatrix.Moreover,thep?qarguedthatmoreaccurateestimatescanbeobtainedsmallesteigenvaluesofRareequalto2.Theseeigen-byconstrainingRbtobeconsistentwithvariousapri-valuesarereferredtoasthenoiseeigenvaluesandtheoriconstraints,includingthosearisingfromlargedi-remaind

6、erofthespectrumisreferredtoasthesignalmensionalrandommatrixtheory.Asettheoreticfor-eigenvalues.SinceRisnotknownitsspectrummustmalismisusedtoformulatethisfeasibilityproblem.beinferredfromobservingthatofthesamplecovari-PnUnsolvedissuesarediscussed.ancematrixRb=(1=n)i=1X(ti)X(ti).L

7、ooselyspeaking,onemustthendecidewheretheobservedPROBLEMSTATEMENTspectrumsplitsintonoiseandsignaleigenvalues.TheestimationproblemistodeterminethedirectionWeconsidertheproblemofdetectingthenumberqofarrivals(i)1ipofthesources.Understandardofsourcesimpingingonanarrayofp(q

8、potheses,thisproblemcanbesolvedviatheMUS