数字信号处理课件 第11章 Signal Averaging.ppt

ChapterEleven.SignalAveraging1

Theunwantedrandomdisturbancescontaminateourmeasurements.Howtogetavalidestimate,orthetruevalueofsomemeasuredparameter?Thesedisturbancesareduetoboththenatureofthevariablebeingmeasuredandthefallibilityofourmeasuringdevices.Thoseunwantedfluctuationsinameasuredvaluearecallednoise.Itistominimizenoisethroughtheprocessofaveraging.Thestatisticalparameterisknownasthemean,variance,andstandarddeviation.2

Indigitalsignalprocessing,averagingoftentakestheformofsummingaseriesoftime-domainsignalsamplesandthendividingthatsumbythenumberofindividualsamples.Mathematically,theaverageofNsamplesofsequencex(n),denotedxave,isexpressedasWhatwecalltheaverage,statisticianscallthemean.Thevarianceofthesequences2definedasThestandarddeviationdefinedasthepositivesquarerootofthevariance,3

11.1.COHERENTAVERAGINGThecoherentaveragingprocessisalsoknownaslinear,predetection,orvectoraveraging.Thatis,thetimephaseofthesignalineachsetisidentical.Withcoherentaveraging,thesinewavewillaveragetoitstruesinewaveamplitudevalue.thenoiseineachsamplesetwillaveragetowardzero.Coherentaveraging,wecanactuallyimprovethesignal-to-noiseratioofanoisysignal.4

Figure11-1.Signalpulseplusnoise:(a)onesampleset;(b)averageof32samplesets;(c)averageof256samplesets.5

Wedliketoknowisthesignal-to-noiseimprovementbycoherentaveragingasafunctionofN.WewanttomeasuresomeconstanttimesignalwithamplitudeA,moreconcisely,WeaverageNseparater(n)measurementsamplesandcalculateasingleaveragevalueraveToseehowaveragingreducesourmeasurementuncertainty,weneedtocomparethestandarddeviationofourrave(k)sequenceofaverageswiththestandarddeviationoftheoriginalr(n)sequence.6

Ifthestandarddeviationofouroriginalseriesofmeasurements