• DocumentCode
    3508488
  • Title

    Sum-capacity of multiple-write noisy memory

  • Author

    Wang, Lele ; Kim, Young-Han

  • Author_Institution
    Dept. of Electr. & Comput. Eng., Univ. of California, San Diego, La Jolla, CA, USA
  • fYear
    2011
  • fDate
    July 31 2011-Aug. 5 2011
  • Firstpage
    2527
  • Lastpage
    2531
  • Abstract
    Motivated by the emerging interests in non-volatile solid-state computer memories such as flash memories, this paper studies the problem of repeatedly storing information on memory cells with noise and state. The goal is to reliably convey t messages by writing Xjj on an n-cell noisy memory p(yj |xj, yj-1), which stores Yjn at the j-th write. We model this problem as a channel with state and introduce the multiple-write noisy memory model, which includes the write-once memory and flash memory models as special cases. The t-write sum-capacity for the multiple-write noisy memory is established as Csum(t) = max[I(X1;Y1)+ Σj=2t(Uj;Yj)- I(Uj;Yj-1)], where the maximum is over all pmfs p(x1) Πj=2t p(uj |yj-1) and functions xj(uj, yj-1), j = 2, ..., t. We derive three outer bounds on the capacity region and discuss their extension to other classes of memory models. These results extend Wolf, Wyner, Ziv, and Körner´s work on the binary write-once memory and Fu and Vinck´s work on the generalized write-once memory to noisy memories.
  • Keywords
    flash memories; information storage; message passing; random-access storage; write-once storage; binary write-once memory; flash memory model; information storage; memory cells; message passing; multiple-write noisy memory; n-cell noisy memory; nonvolatile solid-state computer memory; t-write sum-capacity; write-once memory; Decoding; Encoding; Markov processes; Noise measurement; Phase change materials; Random variables;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Information Theory Proceedings (ISIT), 2011 IEEE International Symposium on
  • Conference_Location
    St. Petersburg
  • ISSN
    2157-8095
  • Print_ISBN
    978-1-4577-0596-0
  • Electronic_ISBN
    2157-8095
  • Type

    conf

  • DOI
    10.1109/ISIT.2011.6034023
  • Filename
    6034023