• DocumentCode
    422961
  • Title

    Constellation randomization (CoRa) for outage performance improvement on MIMO channels

  • Author

    Larsson, Erik G.

  • Author_Institution
    Dept. of Electr. & Comput. Eng., George Washington Univ., DC, USA
  • Volume
    1
  • fYear
    2004
  • fDate
    29 Nov.-3 Dec. 2004
  • Firstpage
    386
  • Abstract
    This work studies a new concept, "constellation randomization" (CoRa), for performance improvement of transmission over quasi-static (block fading) multidimensional ("MIMO") channels with small signal constellations. The motivation behind CoRa is to reduce some of the mutual information penalty associated with the use of a finite input alphabet for signalling on a static, possibly ill-conditioned, channel. We show, both via information-theoretic considerations, and via Monte-Carlo simulation of the frame-error-rate (FER) of a coded system, that CoRa can offer a lower outage probability than a conventional spatial multiplexing system that uses a fixed signal constellation. In our examples the improvement offered by CoRa is comparable to that offered by linear (complex-field) precoding. This is interesting, because the performance gain of CoRa comes at virtually no computational expense.
  • Keywords
    MIMO systems; Monte Carlo methods; channel coding; error statistics; fading channels; probability; CoRa; FER; MIMO channels; Monte-Carlo simulation; block fading channels; coded system; constellation randomization; frame-error rate; information-theoretic considerations; multidimensional channels; outage performance improvement; outage probability; quasi-static channels; signal constellations; Block codes; Constellation diagram; Covariance matrix; Crosstalk; Fading; MIMO; Receiving antennas; Samarium; Symmetric matrices; Vectors;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Global Telecommunications Conference, 2004. GLOBECOM '04. IEEE
  • Print_ISBN
    0-7803-8794-5
  • Type

    conf

  • DOI
    10.1109/GLOCOM.2004.1377975
  • Filename
    1377975