Professional Documents
Culture Documents
Macaw PDF
Macaw PDF
1
MACA continue … BEB (Binary Exponential Back-Off)
Stations that hear RTS but not CTS can Retransmission occur if and only if a
commence transmission. station does not receive a CTS in
Hidden Terminal: response.
C hears B CTS message. Back-off then retransmit.
Exposed Terminal Whenever a CTS is received
C hears B RTS message.
Back-off counter BO = Fdec (BO)
If a station did not get a CTS message for its
RTS, a collision is assumed after a timeout
Whenever a CTS is not received
period. Back-off counter BO = Finc (BO)
Another transmission is scheduled using BEB Fdec = BOmin
(Binary Exponential Back-off). Finc(x) = MIN [2x, BOmax]
P1 P2 P3
2
MACAW- MSM Results MACAW – Basic Message Exchange
MACA uses RTS-CTS-Data model
MACA recovers from errors by the
transport layer
Slow
Solution : use RTS-CTS-Data-ACK
ACK is returned to the sender after
immediately upon completion of data
reception.
P1 P2 P3
Offline
3
MACAW– Back-off Alg. Revisited MACAW – Performance Evaluation
Solution : separate BO for each stream Using MACAW over MACA yielded an
All stations attempting to communicate improvement of over 37% in throughput.
with the same receiving station should use MACAW has yielded a “fairer” division of
the same back-off value. throughput.
Back-off value MACAW is able to cope with highly non-
Copied between stations homogenous congestion, and can shield
Separate BO for each station un-congested neighbors fro losing too
BO of both ends in each packet header much throughput due to the presence of a
congested neighbor.
1994 2004
4
Critique MACAW- Back-off Algorithm
Authors should have compared their approach to Scenario where all the stations are in range of
other MACA-improved protocols each other and two stations are sending data to
Authors did only compare the network the base station, and each of them can produce
throughput without any regard to computational
overhead or power consumption added in their traffic to consume all the bandwidth.
adjustments
Authors did not specify the likelihood of some
scenarios, in other words, if a specific adjustment B
to the algorithm will benefit the overall
performance.
Due to the fact that some problems were left
unsolved or partially solved, what is the
likelihood of theses problems to occur. P1 P2