review of paper 24

From: Shobhit Raj Mathur (shobhit@cs.washington.edu)
Date: Sun Nov 21 2004 - 20:55:56 PST

  • Next message: Alan L. Liu: "Review of MACAW:A Media Access Protocol for Wireless LANs"

    MACAW: A Media Access Proctocol for Wireless LAN's
    ==================================================

    This paper was motivated by the emergence of mobile deivices such as
    Palmtops, PDAs etc at that time. The MAC protocol proposed for wireless
    LANs prior to this paper (MACA) had many shortcomings which the proposed
    protocol MACAW tries to address.

    The congestion in a wireless LAN is at the receiver's end, hence carrier
    sense approaches like ethernet are not useful for wireless LANs. MACAW
    uses a RTS-CTS-DS-DATA-ACK protocol as opposed to RTS-CTS-DATA protcol
    used by MACA. MACAW gains a considerable performance improvement by using
    such a protocol. The ACK provided by the link layer allows for a much
    fater recovery from packet losses. Packets are lost more often in wireless
    networks due to collisions than in wired networks. If the link layer would
    depend on the application layer to detect the packet losses it would
    result in significant delays which can be avoided by using link level
    ACKs.

    The MACAW protocol propogates synchronization information about contention
    periods so that all devices contend effectively. Using the Data-Sending
    packet(DS) allows hosts which 'lose' the contention to compete effectively
    for access when the medium is available next time. Another synchronization
    information which MACAW uses is the RRTS. RRTS packet allows a receiving
    host to contend for bandwidth even when it is in the presence of
    congestion.

    Congestion is not homogenous, it varies with the location of the receiver.
    So instead of characterizing congestion by a single backoff parameter
    MACAW uses seperate backoff parameters for each stream and infact for each
    end of the stream. Backoff parameters are also copied from 'overheard'
    packets so that all the nodes have the same view of the congestion levels.

    These changes significantly improve the performance compared to MACA, but
    the paper still does not identify other concerns in wireless networks such
    as security. One important achievement of MACAW is the fair allocation and
    utilization of bandwidth. It should be noted that the experiments were run
    in the specific topography of Xerox Corporation's Palo Alto Reserach
    Center. The authors themselves agree that they are not sure how this
    protocol would perform in more general wireless LANs.

    It is interesting to note that, MACAW deviates from the end-to-end
    argument by using link level ACKs. This was necessary to improve
    performance, otherwise the applications would face severe delays due to
    packet losses. I think that the MACAW protocol is very close to the 802.11
    protocol used today. Hence MACAW scaled well to diverse topologies.


  • Next message: Alan L. Liu: "Review of MACAW:A Media Access Protocol for Wireless LANs"

    This archive was generated by hypermail 2.1.6 : Sun Nov 21 2004 - 20:55:56 PST