Paper review:
A Comparison of Mechanism for Improving TCP Performance over Wireless Links
Reviewer:
Mike Liu
- State the problem the paper is trying to solve.
The main problem the paper is trying to solve is how to classify, compare, and
evaluate different mechanisms for improving TCP performance over wireless links.
- State the main contribution of the paper: solving a new problem, proposing a
new algorithm, or presenting a new evaluation (analysis). If a new problem, why
was the problem important? Is the problem still important today? Will the
problem be important tomorrow? If a new algorithm or new
evaluation (analysis), what are the improvements over previous algorithms or
evaluations? How do they come up with the new algorithm or evaluation?
The main contribution of the paper is that it provides a framework for evaluating new
mechanisms for improving TCP performance over wireless links and shows how
improvements to each of the improvements to each scheme can maximize performance. Prior
to this, different schemes were proposed but no one had determined a way to classify
them or measure their performance. In addition, improvement under each class of
mechanisms were made to seek out the most optimal performance in each class. This
framework is still rather relevant today because wireless technologies for the
internet are still developing and it is yet to be decided which will be dominant
scheme before it is widely implemented amongst the general publi.
- Summarize the (at most) 3 key main ideas (each in 1 sentence.)
The three 3 key main ideas are:
(1) The three basic groups of schemes for improve TCP performance over Wireless Link
are: end-to-end proposals, split-connection protocols, and link-layer proposals, with
the last of the three having the best performance overall.
(2) Other discoveries amongst each of the three groups are: it is important for
link-layer schemes to be aware of TCP algorithms in order to achieve high end-to-end
throughput; using a SMART-based selective acknowledgement scheme for the wireless hops
yields higher throughput in a split-connection approach, though the throughput is
still slightly less than a well-tuned link-layer scheme, indicating that splitting the
end-to-end connections is a requirement of good performance; end-to-end
schemes, while not as effective as local techniques in handling wireless
losses, can use explicit loss notification to acheive an improvement in throughput and
are promising since significant performance gains can be achieved without any extensive
support from intermediate nodes in the network.
(3) The SMART-based selective acknowledgement scheme is quite efective in dealing with
a high packet loss rate, when employed over the wireless hop or by a sender in a LAN
environment while a SACK scheme based on the IETF Draft resulted in significantly
improved end-to-end performance in a WAN environment.
- Critique the main contribution
- Rate the significance of the paper on a scale of 5
(breakthrough), 4 (significant contribution), 3 (modest contribution), 2
(incremental contribution), 1 (no contribution or negative contribution).
Explain your rating in a sentence or two.
I give this paper a rating of 4 because it sets up a framework for understanding
Wireless technologies involved with the Internet and shows how to evaluate them.
However, it was not very clear in its description of the different algortihms and
strategies and may have been better if it showed actual code or used clear diagrams or
graphs.
- Rate how convincing the methodology is: how do the authors justify the solution
approach or evaluation? Do the authors use arguments, analyses, experiments,
simulations, or a combination of them? Do the claims and conclusions follow from the
arguments, analyses or experiments? Are the assumptions realistic (at the time of the
research)? Are the assumptions still valid today? Are the experiments well designed?
Are there different experiments that would be more convincing? Are there other
alternatives the authors should have considered? (And, of course, is the paper free of
methodological errors.)
Their methodology was somewhat convincing. They used a number of different protocols
and tested them against one another in actual experiments. These experiments, however,
could have been more clearly explained along with the protocols themselves. Because the
experimental methods were a bit vague and only listed the results, there is the
possiblity that there may have been errors in their methods but which are undetectable
since they only posted the data rates acheived from each of their tests. It would have
been more convincing to have graphs of the data rates as they changed with time and
then demonstrated the effects of introducing several pathological conditions to show
which had the best data rate given a harsh environment as one encounters in the real
world.
- What is the most important limitation of the approach?
The most important limitation of their experimental approach is that their experiments
were so specialized that they do not generate the wireless conditions one would
encounter in using such devices on the road or in a normal everyday setting.
- What lessons should researchers and builders take away from this work. What
(if any) questions does this work leave open?
Researchers and builders should take away from this work the fact that well-tuned
link-layer schemes that are TCP aware and use a SMART-based selective acknowledgement
scheme are currently the most optimal scheme to be used for maximizing performance in
wireless environments. The questions that the work leaves open are that are there
perhaps even better or more implementations of the link-layer scheme or also, are there
other schemes in the other two classes, end-to-end and split-connection, that
may acheive better performance but were not considered in this paper. Finally, further
work can also be done to evaluate the performance of different protocols under the
influence of wireless losses in different network topologies.