• DocumentCode
    3059150
  • Title

    Variable-length coding with feedback in the non-asymptotic regime

  • Author

    Polyanskiy, Yury ; Poor, H. Vincent ; Verdu, Sergio

  • Author_Institution
    Dept. of Electr. Eng., Princeton Univ., Princeton, NJ, USA
  • fYear
    2010
  • fDate
    13-18 June 2010
  • Firstpage
    231
  • Lastpage
    235
  • Abstract
    Without feedback, the backoff from capacity due to non-asymptotic block length can be quite substantial for block lengths and error probabilities of interest in many practical applications. In this paper, novel achievability bounds are used to demonstrate that in the non-asymptotic regime, the maximal achievable rate improves dramatically thanks to variable-length coding with feedback. For example, for the binary symmetric channel with capacity 1/2 the blocklength required to achieve 90% of the capacity is smaller than 200, compared to at least 3100 for the best fixed-blocklength, non-feedback code. Virtually all the advantages of noiseless feedback are shown to be achievable with decision-feedback only. It is demonstrated that the non-asymptotic behavior of the fundamental limit depends crucially on the particular model chosen for the “end-of-packet” control signal.
  • Keywords
    block codes; channel capacity; error statistics; feedback; variable length codes; binary symmetric channel capacity; end-of-packet control signal; error probability; fixed-blocklength coding; nonasymptotic block length coding; nonfeedback code; variable-length coding; AWGN; Additive white noise; Capacity planning; Channel capacity; Error probability; Feedback; Gaussian noise; Memoryless systems; Monte Carlo methods; Payloads; Shannon theory; achievability bounds; channel capacity; decision feedback; feedback; memoryless channels; non-asymptotic analysis;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Information Theory Proceedings (ISIT), 2010 IEEE International Symposium on
  • Conference_Location
    Austin, TX
  • Print_ISBN
    978-1-4244-7890-3
  • Electronic_ISBN
    978-1-4244-7891-0
  • Type

    conf

  • DOI
    10.1109/ISIT.2010.5513238
  • Filename
    5513238