Abstract

Session

TCP Analysis

Paper

7-2

Full Paper

ps.gz

Title

A Stochastic Model of TCP/IP with Stationary Random Losses

Author(s)

Eitan Altman (INRIA)
Kostia Avrachenkov (INRIA)
Chadi Barakat (INRIA)

Abstract:

We consider here a flow control mechanism in which the rate at which data is sent increases linearly in time until a loss occurs. At that point the transmission rate decreases by a multiplicative factor. This mechanism is a good approximation to TCP/IP, the congestion control in the Internet. Losses are generated by some exogenous random process, which is assumed to be stationary ergodic; no Markovian assumptions are made. We obtain an explicit expression for the average transmission rate and bounds in the case that there is a limit on the maximum rate. A set of experiments are conducted over the Internet to validate the analytical results as well as the motivations behind the work.