EE384Y
Project Proposal

by:
George Mavromatis (gmavr@stanford.edu)
Ioannis Papadimitriou (jpg@stanford.edu)


During recent years, a lot of work has been done towards the theoretical
understanding of TCP and especially its congestion control
algorithm. Congestion control is viewed as a feedback system, descibed by
a deterministic flow model approximation which is usually based on differential
equations. Paper [1] is considered to have initiated all this
discussion. The next step was to take into account the communication
delays and view the network as a delayed feedback system. Paper [2]
tries to establish stability criteria when round trip times are equal and
papers [3], [4] when arbitrary. Following that, papers [5] and [6]
propose new congestion control systems. 

Another issue that is strongly connected with the 
discussion presented above is the use of control theory in order to prove that
the network system is stable. Actually, the network behavior is being
approximated by some mathematical equations (usually differential) and
then stability is proven using Nyquist and Bode diagrams. Papers [6], [7]
examine this method.


For the time being, interesting points that can be studied within this project
include:

(*) Understanding the methodologies used in order to prove the
stability of a congestion control algorithm
(*) Establishing the equations that desribe XCP when round trip times are not
equal as assumed in paper [6]
(*) Proving the stability or instability of the XCP algorithm when
round trip times are not equal.
(*) Using simulation in order to test whether the strong law of
large numbers holds true for the arrivals or not. If yes, then we will prove
analytically the stability of the system.
(*) Testing the behavior of XCP under specific traffic models (e.g. Web-based
type traffic) using simulation. Papers [8] and [9] that connect TCP
congestion control with TCP flow patterns can prove to be valuable to our
study. 


Schedule
~~~~~~~~

Week of:

April 15	Papers - Background

April 22	Papers - Background

April 29	Papers - Background - Simulation

May 6		Simulation

May 13		Analytical study - Simulation runs

May 20		Analytical study

		Sunday, May 26th: Draft Report

May 27		Analytical study

June 3		Final report


Final Presentation: May 22nd to June 5th
Final Report: June 5th



Related Papers:
[1] "Rate control for communication networks:shadow prices, proportional
fairness and stability", FP Kelly, AK Maulloo and DKH Tan 
[2] "End-to-end congestion control for the internet: delays and
stability", Ramesh Johari and David Tan
[3] "Stability of distributed congestion control with heterogeneous
feedback delays", L. Massoulie
[4] "On the stability of end-to-end congestion control for the Internet",
Glenn Vinnicombe
[5] "Scalable Laws for stable network congestion control", Fernando
Paganini, John Doyle, Steven Low
[6] "Internet congestion control for future high bandwidth-delay product",
Dina Katabi, Mark Handley, Charlie Rohrs.
[7] "A control theoretic analysis of RED", C.V. Hollot, Vishal Misra, Don
Towsley and Wei-Bo Gong
[8] "Statistical Bandwidth sharing: a study of congestion at flow level",
Ben Fredj, T. Bonald, A. Proutiere, G. Regnie, J. Roberts
[9] "Impact of fairness on Internet performance", T. Bonald and Laurent
Massoulie