Monday, September 29, 2008

MACAW: A Media Access Protocol for Wireless LAN's

In light of the recent emergence of mobile computing devices that will necessitate controlled access to indoor wireless media, this paper proposes a new media access protocol for wireless LAN's: MACAW. Given that the relevant congestion in the described infrastructure is at the receiver, carrier-sense approaches are avoided and Karn's MACA (multiple access, collision avoidance) proposal is used as a starting point. The MACA protocol uses an RTS-CTS-DATA packet exchange and a binary exponential backoff (BEB) scheme. The authors argue that reliable transmission can be enforced more efficiently using link-layer recovery. Moreover, bi-directional transmission schemes require attention to congestion at both the receiver and the sender. Instead of adopting a standard CSMA/CA approach, the MACAW protocol sends a data-sending packet (DS) to block all other transmissions until the ACK packet has been transmitted - this prevents useless transmission requests in the exposed terminal case. Hence, MACAW uses an RTS-CTS-DS-DATA-ACK packet exchange. Moreover, a coordinated MILD (multiplicative increase, linear decrease) backoff scheme is used to ensure fairness of access.

The development of MACAW revolves around fair access to wireless media - the multiple stream model, RRTS mechanism and coordinated backoff scheme are all introduced with this goal in mind. That said, I wonder why the focus is completely on fairness and there isn't even the slightest trade-off with performance. Moreover, the discussion of the MILD (MIAD) backoff scheme made me wonder why the AIMD scheme is not used here, given that it helped achieve fairness in the congestion control scheme discussed earlier. These are different manifestations of fairness but I was surprised to see no mention of AIMD. Finally, the illustrations of the exposed terminal, hidden terminal and other situations were very effective in explaining the reasoning behind the features of the MACAW protocol. Overall, I really enjoyed this paper.

No comments: