Thereismuchmoreinformationinastochasticnon-Gaussianordeterministicsignalthanisconveyedbyitsautocorrelationandpowerspectrum.Higher-orderspectrawhicharedefinedintermsofthehigher-ordermomentsorcumulantsofasignal,containthisadditionalinformation.TheHigher-OrderSpectralAnalysis(HOSA)Toolboxprovidescomprehensivehigher-orderspectralanalysiscapabilitiesforsignalprocessingapplications.Thetoolboxisanexcellentresourcefortheadvancedresearcherandthepracticingengineer,aswellasthenovicestudentwhowantstolearnaboutconceptsandalgorithmsinstatisticalsignalprocessing.TheHOSAToolboxisacollectionofM-filesthatimplementavarietyofadvancedsignalprocessingalgorithmsfortheestimationofcross-andauto-cumulants(includingcorrelations),spectraandolyspectra,bispectrum,andbicoherence,andomputationoftime-frequencydistributions.Basedonthese,algorithmsforparametricandnon-parametricblindsystemidentification,time-delayestimation,harmonicretrieval,phase-coupling,directionofarrivalestimation,parameterestimationofVolterra(non-linear)models,andadaptivelinearpredictionareimplemented.AlsoincludedarealgorithmsfortestingofGaussianityandLinearityofatimeseries.Afulltutorialanddemosetareincludedinthetoolbox.
1