• DocumentCode
    3096549
  • Title

    Decoupling higher order cumulant sequences resulting from three wave coupling processes

  • Author

    Sherman, David ; Zoltowski, Michael

  • Author_Institution
    Sch. of Electr. Eng., Purdue Univ., West Lafayette, IN, USA
  • fYear
    1990
  • fDate
    10-12 Oct. 1990
  • Firstpage
    227
  • Lastpage
    231
  • Abstract
    Using cumulant projections, the authors reduce the complexity of the frequency estimation problem for quadratically coupled sinusoids. The bispectrum is viewed along a single radial slice whose defining frequencies are related by a simple ratio. This procedure of taking projections along lines orthogonal to projection axes corresponding to those slices decouples the sinusoids. The projected cumulants contain one-dimensional sinusoids. Projections comprise frequency decoupling. Separating the projected cumulants into symmetric and skew-symmetric cumulants, phase decoupling is achieved. Sinusoidal frequency estimation can be derived from a purely autoregressive (AR) modeling of the skew-symmetric and symmetric cumulants. The predictor polynomials derived from this scheme combine both causal and non-causal components.<>
  • Keywords
    parameter estimation; polynomials; spectral analysis; statistical analysis; autoregressive modelling; complexity; cumulant projections; frequency decoupling; frequency estimation problem; higher order cumulant sequences; phase decoupling; predictor polynomials; quadratically coupled sinusoids; skew-symmetric cumulants; spectral analysis; symmetric cumulants; three wave coupling processes; Data analysis; Digital signal processing; Fourier transforms; Frequency estimation; Frequency synthesizers; Matrix decomposition; Polynomials; Spectral analysis; Surges; Symmetric matrices;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Spectrum Estimation and Modeling, 1990., Fifth ASSP Workshop on
  • Conference_Location
    Rochester, NY, USA
  • Type

    conf

  • DOI
    10.1109/SPECT.1990.205580
  • Filename
    205580