• DocumentCode
    252989
  • Title

    The Gaussian channel with noisy feedback: Near-capacity performance via simple interaction

  • Author

    Ben-Yishai, A. ; Shayevitz, O.

  • Author_Institution
    Dept. of EE-Syst., Tel Aviv Univ., Tel Aviv, Israel
  • fYear
    2014
  • fDate
    Sept. 30 2014-Oct. 3 2014
  • Firstpage
    152
  • Lastpage
    159
  • Abstract
    Consider a pair of terminals connected by two independent additive white Gaussian noise channels, and limited by individual power constraints. The first terminal would like to reliably send information to the second terminal, within a given error probability. We construct an explicit interactive scheme consisting of only (non-linear) scalar operations, by endowing the Schalkwijk-Kailath noiseless feedback scheme with modulo arithmetic. Our scheme achieves a communication rate close to the Shannon limit, in a small number of rounds. For example, for an error probability of 10-6, if the Signal to Noise Ratio (SNR) of the feedback channel exceeds the SNR of the forward channel by 20dB, our scheme operates 0.8dB from the Shannon limit with only 19 rounds of interaction. In comparison, attaining the same performance using state of the art Forward Error Correction (FEC) codes requires two orders of magnitude increase in delay and complexity. On the other extreme, a minimal delay uncoded system with the same error probability is bounded away by 9dB from the Shannon limit.
  • Keywords
    AWGN channels; Gaussian channels; error statistics; feedback; forward error correction; FEC codes; SNR; Schalkwijk-Kailath noiseless feedback scheme; Shannon limit; additive white Gaussian noise channels; error probability; feedback channel; forward error correction; modulo arithmetic; near-capacity performance; noisy feedback; power constraints; signal to noise ratio; Capacity planning; Error probability; Estimation error; Feedforward neural networks; Signal to noise ratio; Zinc;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Communication, Control, and Computing (Allerton), 2014 52nd Annual Allerton Conference on
  • Conference_Location
    Monticello, IL
  • Type

    conf

  • DOI
    10.1109/ALLERTON.2014.7028450
  • Filename
    7028450