• DocumentCode
    2931063
  • Title

    Joint source-channel coding of uniform memoryless sources over binary symmetric channels

  • Author

    Azami, Seyed Bahram Zahir ; Duhamel, Pierre ; Rioul, Olivier

  • Author_Institution
    Ecole Nat. Superieure des Telecommun., Paris, France
  • Volume
    6
  • fYear
    1998
  • fDate
    1998
  • Firstpage
    3614
  • Abstract
    A joint design of source and channel coding is considered for digital transmission over a binary symmetric channel (BSC). Three major techniques are used. For a binary source, channel optimized coder design (COCD) is proposed. This method has non-binary decoding values and minimizes the MSE distortion measure. It can be understood as a channel coder specifically designed to be robust under transmission errors. Then, this tool is applied to the transmission of a uniform source. This is done through a bitwise decomposition structure in which the samples of a uniform source are first expressed in binary representation and bits of the same weight are encoded together. Finally, a practical optimization algorithm is explained that delivers the best set of coders, for a given raw error probability (BER) of the channel
  • Keywords
    combined source-channel coding; digital communication; error statistics; mean square error methods; memoryless systems; optimisation; BER; BSC; COCD; MSE distortion measure; binary representation; binary symmetric channels; bitwise decomposition structure; channel coder; channel optimized coder design; digital transmission; error probability; joint source-channel coding; nonbinary decoding values; optimization algorithm; samples; transmission errors; uniform memoryless sources; Bit error rate; Bit rate; Channel coding; Image coding; Image processing; Information Theory Society; Lagrangian functions; Performance loss; Rate-distortion; Signal processing;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Global Telecommunications Conference, 1998. GLOBECOM 1998. The Bridge to Global Integration. IEEE
  • Conference_Location
    Sydney,NSW
  • Print_ISBN
    0-7803-4984-9
  • Type

    conf

  • DOI
    10.1109/GLOCOM.1998.775976
  • Filename
    775976