From: Masaharu Kobashi (mkbsh@cs.washington.edu)
Date: Mon Nov 22 2004 - 00:29:44 PST
1. Main result of the paper
The paper presents a new media access protocol for wireless LANs, MACAW,
which was derived from MACA and accomplishes remarkable improvements over
MACA in fairness of bandwidth allocation without degrading the overall
network utilization significantly.
2. Strengths in this paper
The fundamental insight of the new protocol is the recognition of the
fact
that the relevant congestion is at the receiver. Upon this basic
framework
the new algorithm with added ACK, DS and RRTS was built. As the
results of
the simulations show, it really works and improves the fairness
remarkably.
Another strength of the algorithm is that it uses separate backoff
parameters for each stream and for each end of the stream based on the
insight that the level of congestion varies according to the location
of receiver.
3. Limitations and suggested improvements
The setup for the argument misses an important realistic element,
which is the problems that occur in the state of transition of
stations (pads). The paper assumes everything is stationary. But at least
pads can move around freely in the real situations even while they are
transmitting data. This possibility makes the argument more complex than
the ones presented in the paper, since the state of congestion and the
area of exposure to each station can change at any point of transmission.
Once this factor is taken into account the proposed protocol can have
difficulties in achieving the reported results.
The performance evaluation is mostly done with UDP, except one which is
TCP (Table 11). The effect of transport protocol on the network
utilization
is not discussed at all in the paper. It deserves attention since UDP
and TCP can have different sensitivity to congestion and interruptions
of transmission.
4. Relevance today and future
Wireless is the trend of current and future. By now, however,
the proposed protocol may be already obsolete due to the limitations
stated by the authors (i.e lack of multicast support etc.) as well as
the limitations pointed in the above section. But the basic insights
such as "congestion is at the receiver" are still valid into the future,
hence the paper will contribute to the design of the future protocols.
This archive was generated by hypermail 2.1.6 : Mon Nov 22 2004 - 00:29:44 PST