Evaluation of Queue Management Algorithms
Group Member : Jichuan Chang
(cjc@cs.cmu.edu)
Ningning Hu (
hnn@cs.cmu.edu)
Liu Ren (
liuren@cs.cmu.edu)
Project Home Page:http://www.cs.cmu.edu/~hnn/cs744_project/project.html
Project Description: Recently, some novel queue
management and congestion control algorithms have been proposed to enhance
the stability and QoS support of TCP. Although there is no evidence of
serious operational problem with the congestion control of current Internet,
it is substantially agreed that at least it need to protect the Internet
from "misbehaving" flows that do not use conformant end-to-end. For example,
TCP Friendly Rate Control provides smooth changes in transmission rate
for continuous streaming applications; BLUE uses packet loss and link idle
event as alternative indicator for queue management;CHOKe works on simple,
stateless algorithms that achieve fair bandwidth allocation, etc.
As new services emerging quickly with quite different
requirements, before we update the Internet infrastructure and end hosts
with new TCP/IP implementations, one question needs to be answered, which
congestion control algorithm will best suit the different traffic patterns
and requirements? Currently, the evaluations coming along with these proposals
are limited in several scenarios, thus a more comprehensive evaluation
is need to justify the effort to deploy them in the future.
Plan of Attack: We plan to implement two or threeof these new
mechanisms in NS-2, and evaluate the newer congestion control algorithms,
based on both generic and special traffic patterns. If possible, we?d like
to look at some improvement into current mechanisms.
Schedule:
Time
|
Task
|
Week 1
|
Read papers, install NS, get started
|
Week 2
|
Use NS-2 to simulate RED+ECN
|
Week 3-5
|
Implement BLUE in NS-2
|
Week 6-7
|
Compare BLUE with RED
|
Week 8-9
|
Implement CHOke and TRFC
|
Week 10-11
|
Evaluation of different mechanisms
|
Week 12
|
Write Report
|
*. All the tasks are supposed to be done by all the members in the group
Milestion: By Week 7, we should have finished the first congestion
control algorithm, both implementation and performance comparison, and
we should be on the way of working on some more algorithms.
Literature Search:
-
Floyd, S., TCP
and Explicit Congestion Notification. ACM Computer Communication Review,
V. 24 N. 5, October 1994, p. 10-23.
-
Floyd, S., and Jacobson, V., Random
Early Detection gateways for Congestion Avoidance V.1 N.4, August 1993,
p. 397-413.
-
A Comparison of Equation-Based
and AIMD Congestion Control. Sally Floyd, Mark Handley, and Jitendra
Padhye. May 2000.
-
Improving
Internet Congestion Control and Queue Management Algorithms. Wu-chang
Feng. (dissertation).
-
W. Feng, D. Kandlur, D. Saha, K. Shin, "Blue:
A New Class of Active Queue Management Algorithms" U. Michigan CSE-TR-387-99,
April 1999.
-
The ns Manual.
The VINT Project.
Resources Needed:
-
The Network Simulator - ns-2