• DocumentCode
    2081523
  • Title

    Head-to-Tail: Managing Network Load through Random Delay Increase

  • Author

    Dimitriou, Stylianos ; Tsaoussidis, Vassilis

  • Author_Institution
    Democritus Univ. of Greece, Xanthi
  • fYear
    2007
  • fDate
    1-4 July 2007
  • Firstpage
    257
  • Lastpage
    262
  • Abstract
    Window-based congestion control is typically based on exhausting bandwidth capacity, which occasionally leads to transient congestion. Moreover, flow synchronization may deteriorate conditions further, leading to persistent or more severe congestion, which is experienced by flows through increasing queuing delays and packet retransmission. Head-to-Tail is a new approach to queue scheduling that aspires to alleviate this problem. When conditions at the router´s buffer indicate high risk for congestion, Head-to-Tail delays packets intentionally to fabricate the senders´ impression about the network load. This implicit signal to reduce the transmission rate allows for a responsive behavior prior to congestion. In this paper, we evaluated Head-to-Tail with TCP Vegas and compared it with RED and other TCP variants. The initial results indicate that congestion events and retransmissions can be significantly eliminated.
  • Keywords
    delays; load management; synchronisation; telecommunication congestion control; transport protocols; Head-to-Tail; bandwidth capacity; congestion control; network load management; packet retransmission; random delay; transient congestion; Centralized control; Collaborative work; Computer network management; Computer networks; Delay; Engineering management; Load management; Potential well; Proposals; Transport protocols;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Computers and Communications, 2007. ISCC 2007. 12th IEEE Symposium on
  • Conference_Location
    Aveiro
  • ISSN
    1530-1346
  • Print_ISBN
    978-1-4244-1520-5
  • Electronic_ISBN
    1530-1346
  • Type

    conf

  • DOI
    10.1109/ISCC.2007.4381610
  • Filename
    4381610