(no subject)

Summary of the paper

This paper proposes a new mechanism of congestion control. It is
called XCP. The Internet evolves to incorporate very high-bandwidth
optical links and more large-delay satellite links. For this reason,
this novel approach to Internet congestion control outperforms TCP in
conventional environments, and remains efficient, fair, scalable, and
stable as the bandwidth-delay product increases.

XCP is a window-based protocol which decouples utilization control
from fairness control. After analysis and measurement, authors
conclude that XCP achieves fair bandwidth allocation, high
utilization, small standing queue size, and near-zero packet drops,
with both steady and highly varying traffic. It does not maintain any
per-flow state in routers and requires few CPU cycles per packet.

XCP provides a joint design of end-systems and routers. Senders
maintain their congestion window and round trip time and communicate
these to the routers via a congestion header in every packet. Routers
monitor the input traffic rate to each of their output queues. Based
on the difference between the link bandwidth and its input traffic
rate, the router tells the flows sharing that link to increase or
decrease their congestion windows. In this way, XCP could control the
throughput and avoid the congestion. The job of the routers is to
compute the feedback to cause the system to converge to optimal
efficiency and min-max fairness. It uses an efficiency controller(EC)
and a fairness controller(FC). The efficiency controller's purpose is
to maximize link utilization while minimizing drop rate and
persistent queues. EC with compute a desired increase or decrease in
the number of bytes that the aggregate traffic transmits in a control
interval, then control the bandwidth of flows by feedback. The job of
fairness of FC is to apportion the feedback to individual packets to
achieve fairness. Paper has described the algorithm in detail. The
decoupling design could be implemented simply, and significantly
improves previous approaches. This also provides stability in face of
high bandwidth or large delay, and ensures both efficiency and
utilization of the network resources.

Authors provide the detail of testing and compare XCP with TCP and
some other schemes. At the aspect of Impact of capacity, Impact of
feedback delay, Impact of number of flows, Fairness and Impact of
short web-like traffic, all the experiments show that XCP has better
performance.

At last, authors give the deployment description and some conclusion
and future work.

Points in favour or against

The paper is well written, with fine and clear presentation.
Obviously, it improves the previous work. This paper mainly focuses
on high-delay network environment. The main contribution of it is
that it solves the congestion problems at this environment. The novel
point is to propose the decoupling design of efficiency control and
fairness control which makes the system more simple, stable and
efficient. Authors also use the concept of control theory to solve
the problem. However, in this scheme, protocol should add the
congestion header which will add the cost of connection, although the
algorithm is not complicated. Another problem is that this scheme
should be implemented by both end uses and routers. Both of them
should cooperate together. In another word, we should deploy the
whole scheme to all the uses and routers in the Internet to complete
this congestion control scheme. It makes high cost.
Received on Sat Sep 30 2006 - 16:56:48 EDT