 
Summary: EE 99, Third Term, 20012002
Random Marking Algorithms for Path Congestion Estimation
Jonathan Harel
California Institute of Technology
Pasadena, CA 91125, USA
email: harel@its.caltech.edu
Abstract  In this paper, I compare the perfor
mance of two probabilistic marking algorithms, REM
and SAM, which attempt to measure the sum "con
gestion" along a path in a network. I first briefly out
line the two marking schemes, then compare them us
ing the standard statistical measures of performance.
I then propose a new algorithm.
I. Introduction
Suppose we have a path of length l in a communication
network, where each link in the path has a congestion constant
si associated with it, i {1, 2, . . . , l}. We wish to estimate
S =
l
i=1
