Wednesday, September 30, 2009

MACAW: A Media Access Protocol for Wireless LAN’s

CSMA can't sense the hidden terminal and it provides only information about the collision at the receiver, so the authors tried to find another approach based on the previous one MACA which is based on RTS/CTS short and fixed size signaling packet. Both RTS and CTS packets contains the length of the proposed data transmission such that any node hears RTS and/or CTS defer its transmission. MACA does not provide adequate level of fairness within one cell because of the applied BEB algorithm, and the calculated congestion information is different at each pad.

The paper includes a design of multiple accesses called MACAW for indoor wireless LAN infrastructure with a goal to deliver fairness over optimal total throughput. The infrastructure uses single 256kHz channel between the base stations installed in the ceiling and the pads
· Pads and base stations transmit the same power
· Range of transmission is 3-4 meters
· The cell around each base station of very small size and sharply boundary
· Two stations are either in range or out of range of one another
· A station receives a packet correctly if and only if there is one active transmitter within its range
· Pads or base station know the presence of other devices only if they communicate

The authors applied the fairness's notion per stream rather than per station by allocating a queue for each stream and running a back off algorithm MILD independently for each queue. BO algorithm bases on multiplicative increase in back off time after occurrence a collision and linear decrease after successful transmission. Using UDP traffic data the fair allocation of the bandwidth is proved by the simulation.

MACAW also modifies basic link recovery RTS-CTS-DATA exchange of MACA, by sending DS packet by the sender to inform other stations about the existence and length of the following data transmission, and ACK packet by the receiver after successful reception of data. However, the additional ACK overhead was not helpful in the simulation at error rate=0.001.
The paper describes a scenario of two cells, each in coverage range of their respective base station and of other where one of the streams in a cell will be negatively affected (its access completely denied) by the transmission to the second pad (received all the requested throughput). The solution is to synchronize the information by introducing additional overhead packet called RRTS. Whenever a station receives an RTS to which it can't respond, it sends a RRTS packet to the sender during the next contention period. However all these exchanging messages don't help all scenarios for unicast transmission and muticast in general.
The author portrays clearly some challenges in media access of wireless LAN but I think the prposed solution here is not helpful well compared to the introduced overheads.

No comments:

Post a Comment