• DocumentCode
    2135629
  • Title

    Packet classification using the Hidden Markov Model

  • Author

    Khadimi, A.E. ; Lmater, M.A. ; Eddabbah, M. ; El Kayyali, M.S.

  • Author_Institution
    Dept. of Syst. & Commun., Inst. of Posts & Telecommun. (INPT) of Rabat, Rabat, Morocco
  • fYear
    2011
  • fDate
    7-9 April 2011
  • Firstpage
    1
  • Lastpage
    5
  • Abstract
    In the last decade, traffic classification has been the most interesting field for the majority of network searchers, and this behavior is caused by its importance, so many successful results have already improved reached. In this way, we present our method of packet classification which is based on the Hidden Markov Model (HMM); as an HMM states we are interested to the Flags bits of a TCP packet, and as an HMM observations we were convinced that we should focus on a packet property that didn´t depend on the payload information to avoid the encryption problem, so the right alternative was the distribution of the packet size for each state of our HMM. This process has been verified and validated by several applications like: Steaming video over HTTP (youtube.com, dailymotion.com, justin.tv), peer-to-peer video streaming, File transfer protocol (FTP). Based on the related woks in this field, our method has given us convenient results comparing to old works.
  • Keywords
    hidden Markov models; packet radio networks; telecommunication traffic; transport protocols; HMM observations; HMM states; TCP packet; encryption problem; file transfer protocol; hidden Markov model; packet classification; peer-to-peer video streaming; steaming video over HTTP; traffic classification; Databases; Hidden Markov models; IP networks; Payloads; Protocols; Streaming media; Hidden Markov Model (HMM); Packet classification; TCP packet; Traffic classification;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Multimedia Computing and Systems (ICMCS), 2011 International Conference on
  • Conference_Location
    Ouarzazate
  • ISSN
    Pending
  • Print_ISBN
    978-1-61284-730-6
  • Type

    conf

  • DOI
    10.1109/ICMCS.2011.5945705
  • Filename
    5945705