Ngày tải lên :
16/12/2013, 04:15
... Introduction
Thepast20yearswitnessedanexpansionofpowerspectrumestimationtechniques,whichhave
provedessentialinmanyapplications,suchascommunications,sonar,radar,speech/imageprocess-
ing,geophysics,andbiomedicalsignalprocessing[13,11,7].Inpowerspectrumestimationthe
processunderconsiderationistreatedasasuperpositionofstatisticallyuncorrelatedharmoniccom-
ponents.Thedistributionofpoweramongthesefrequencycomponentsisthepowerspectrum.As
such,phaserelationsbetweenfrequencycomponentsaresuppressed.Theinformationinthepower
spectrumisessentiallypresentintheautocorrelationsequence,whichwouldsufceforthecomplete
statisticaldescriptionofaGaussianprocessofknownmean.However,thereareapplicationswhere
onewouldneedtoobtaininformationregardingdeviationsfromtheGaussianityassumptionand
presenceofnonlinearities.Inthesecasespowerspectrumisoflittlehelp,andonewouldhaveto
lookbeyondthepowerspectrumorautocorrelationdomain .Higher- OrderSpectra(HOS)(oforder
greaterthan2),whicharedenedintermsofhigher-ordercumulantsofthedata,docontainsuch
information[16].Thethird-orderspectrumiscommonlyreferredtoasbispectrum,thefourth -order
oneastrispectrum,andinfact,thepowerspectrumisalsoamemberofthehigher-orderspectral
class;itisthesecond-orderspectrum.
HOSconsistofhigher-ordermomentspectra,whicharedenedfordeterministicsignals,and
cumulantspectra,whicharedenedforrandomprocesses.Ingeneral,therearethreemotivations
behindtheuseofHOSinsignalprocessing:(1)tosuppressGaussiannoiseofunknownmeanand
variance;(2)toreconstructthephaseaswellasthemagnituderesponseofsignalsorsystems;and
(3)todetectandcharacterizenonlinearitiesinthedata.
TherstmotivationstemsfromthepropertyofGaussianprocessestohavezerohigher -order
spectra.Duetothisproperty,HOSarehighsignal-to-noiseratiodomains,inwhichonecanperform
detection,parameterestimation,orevensignalreconstructionevenifthetimedomainnoiseis
spatiallycorrelated.Thesamepropertyofcumulantspectracanprovidemeansofdetectingand
characterizingdeviationsofthedatafromtheGaussianmodel.
c
1999byCRCPressLLC
... Introduction
Thepast20yearswitnessedanexpansionofpowerspectrumestimationtechniques,whichhave
provedessentialinmanyapplications,suchascommunications,sonar,radar,speech/imageprocess-
ing,geophysics,andbiomedicalsignalprocessing[13,11,7].Inpowerspectrumestimationthe
processunderconsiderationistreatedasasuperpositionofstatisticallyuncorrelatedharmoniccom-
ponents.Thedistributionofpoweramongthesefrequencycomponentsisthepowerspectrum.As
such,phaserelationsbetweenfrequencycomponentsaresuppressed.Theinformationinthepower
spectrumisessentiallypresentintheautocorrelationsequence,whichwouldsufceforthecomplete
statisticaldescriptionofaGaussianprocessofknownmean.However,thereareapplicationswhere
onewouldneedtoobtaininformationregardingdeviationsfromtheGaussianityassumptionand
presenceofnonlinearities.Inthesecasespowerspectrumisoflittlehelp,andonewouldhaveto
lookbeyondthepowerspectrumorautocorrelationdomain .Higher- OrderSpectra(HOS)(oforder
greaterthan2),whicharedenedintermsofhigher-ordercumulantsofthedata,docontainsuch
information[16].Thethird-orderspectrumiscommonlyreferredtoasbispectrum,thefourth -order
oneastrispectrum,andinfact,thepowerspectrumisalsoamemberofthehigher-orderspectral
class;itisthesecond-orderspectrum.
HOSconsistofhigher-ordermomentspectra,whicharedenedfordeterministicsignals,and
cumulantspectra,whicharedenedforrandomprocesses.Ingeneral,therearethreemotivations
behindtheuseofHOSinsignalprocessing:(1)tosuppressGaussiannoiseofunknownmeanand
variance;(2)toreconstructthephaseaswellasthemagnituderesponseofsignalsorsystems;and
(3)todetectandcharacterizenonlinearitiesinthedata.
TherstmotivationstemsfromthepropertyofGaussianprocessestohavezerohigher -order
spectra.Duetothisproperty,HOSarehighsignal-to-noiseratiodomains,inwhichonecanperform
detection,parameterestimation,orevensignalreconstructionevenifthetimedomainnoiseis
spatiallycorrelated.Thesamepropertyofcumulantspectracanprovidemeansofdetectingand
characterizingdeviationsofthedatafromtheGaussianmodel.
c
1999byCRCPressLLC
... modeling of the bispectrum have been shown to
yield superior resolution [30, 31] than the ones obtained with conventional methods.
76.6 Applications/Software Available
Applications of HOS span...