• DocumentCode
    498313
  • Title

    A Method Creating Different Packet Loss Membership Function in Heterogeneous Network

  • Author

    Yanxiang, Zhen ; Fang, Su ; Huikan, Xiang

  • Author_Institution
    Beijing Univ. of Posts & Telecommun., Beijing, China
  • Volume
    3
  • fYear
    2009
  • fDate
    19-21 May 2009
  • Firstpage
    315
  • Lastpage
    319
  • Abstract
    The packet loss differentiating mechanism of TCP for heterogeneous network has the important influence to the stability of wireless network, relative one way delay (ROD) of Loss Pair is adopted as the differentiating parameter to distinguish the packet loss pattern in reference [Su Fang, Fan Ying-lei, 2008]. The accuracy of this algorithm depends on ROD membership function. In order to differentiate packet loss pattern more accurately, the initialization method based on potential function is proposed by EM algorithm analysis, and the simulation result shows it is superior in convergence speed. Then the modified EM algorithm is applied in creating different packet loss membership function which congestion and wireless error coexist in the network. From the results of simulation, this algorithm has well distinguishing effects in the membership function of different packet loss pattern and practical application value.
  • Keywords
    computer networks; data communication; expectation-maximisation algorithm; packet radio networks; radio access networks; EM algorithm analysis; heterogeneous network; packet loss membership function; wireless network stability; Algorithm design and analysis; Analytical models; Clustering algorithms; Intelligent networks; Intelligent systems; Mathematical model; Pattern analysis; Stability; Wireless LAN; Wireless networks; Expectation maximization algorithm; Gaussian mixture model; Heterogeneous network; Membership function;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Intelligent Systems, 2009. GCIS '09. WRI Global Congress on
  • Conference_Location
    Xiamen
  • Print_ISBN
    978-0-7695-3571-5
  • Type

    conf

  • DOI
    10.1109/GCIS.2009.80
  • Filename
    5209151