• DocumentCode
    825091
  • Title

    Circulant and skew-circulant matrices as new normal-form realization of IIR digital filters

  • Author

    Liu, Vincent C. ; Vaidyanathan, P.P.

  • Author_Institution
    Dept. of Electr. Eng., California Inst. of Technol., Pasadena, CA, USA
  • Volume
    35
  • Issue
    6
  • fYear
    1988
  • fDate
    6/1/1988 12:00:00 AM
  • Firstpage
    625
  • Lastpage
    635
  • Abstract
    Normal-form fixed-point state-space realization of IIR (infinite-impulse response) filters are known to be free from both overflow oscillations and roundoff limit cycles, provided magnitude truncation arithmetic is used together with two´s-complement overflow features. Two normal-form realizations are derived that utilize circulant and skew-circulant matrices as their state transition matrices. The advantage of these realizations is that the A-matrix has only N (rather than N2) distinct elements and is amenable to efficient memory-oriented implementation. The problem of scaling the internal signals in these structures is addressed, and it is shown that an approximate solution can be obtained through a numerical optimization method. Several numerical examples are included
  • Keywords
    digital arithmetic; digital filters; matrix algebra; sensitivity analysis; state-space methods; IIR digital filters; circulant matrices; eigenvalue sensitivity measure; fixed-point state-space realization; infinite-impulse response; internal signal scaling; magnitude truncation arithmetic; noise analysis; normal-form realization; numerical optimization method; skew-circulant matrices; state transition matrices; two´s-complement overflow features; Convolution; Digital filters; Eigenvalues and eigenfunctions; Finite wordlength effects; Hardware; IIR filters; Limit-cycles; Nonlinear filters; Quantization; Signal processing algorithms;
  • fLanguage
    English
  • Journal_Title
    Circuits and Systems, IEEE Transactions on
  • Publisher
    ieee
  • ISSN
    0098-4094
  • Type

    jour

  • DOI
    10.1109/31.1800
  • Filename
    1800