• DocumentCode
    1410905
  • Title

    A-Posteriori Symbol Probabilities and Log-Likelihood Ratios for Coherently Detected frac{π}{4}-DE-QPSK

  • Author

    Van Houtum, Wim J. ; Willems, Frans M J

  • Author_Institution
    Catena Radio Design CRD, Eindhoven, Netherlands
  • Volume
    15
  • Issue
    2
  • fYear
    2011
  • fDate
    2/1/2011 12:00:00 AM
  • Firstpage
    160
  • Lastpage
    162
  • Abstract
    In this letter, coherent detection of π/4-DE-QPSK is considered, but our analysis also holds for common DE-QPSK. It is shown that maximum a-posteriori (MAP) sequence detection can be regarded as an approximation, based on selecting dominant exponentials, of MAP symbol detection. A better approximation, relying on piecewise-linear fitting of the logarithm of the hyperbolic cosine, is proposed. This approximation results in a performance very close to optimal symbol detection. For the case where the symbols are produced by convolutional encoding and Gray mapping, the log-likelihood ratios are investigated. Again a simple approximation based on selecting dominant exponentials and an improved approximation relying on piecewise-linear fits, are discussed. As in the uncoded case the improved approximation gives a performance quite close to ideal. While the particular examples considered show modest gains in performance, this letter provides a way of improving performance when needed.
  • Keywords
    approximation theory; convolutional codes; maximum likelihood detection; maximum likelihood estimation; probability; quadrature phase shift keying; Gray mapping; MAP sequence detection; MAP symbol detection; a-posteriori symbol probability; coherent detected π/4-DE-QPSK; convolutional encoding; dominant exponential selection; hyperbolic cosine; log-likelihood ratios; maximum a-posteriori sequence detection; optimal symbol detection; piecewise-linear fitting; Differentially encoded phase-shift keying; MAP detection; coherent detection; log-likelihood ratios;
  • fLanguage
    English
  • Journal_Title
    Communications Letters, IEEE
  • Publisher
    ieee
  • ISSN
    1089-7798
  • Type

    jour

  • DOI
    10.1109/LCOMM.2011.122010.101842
  • Filename
    5673998