End-to-end congestion control and bandwidth estimation in high speed ATM networks

Abstract

We propose an end-to-end congestion control and bandwidth estimation algorithm for ABR traffic in high-speed ATM networks. The congestion control algorithm is derived using the Smith predictor and uses an estimation of the bottleneck queue level as feedback. The estimation allows the algorithm to be executed at the entry node of a high-speed ATM network without requiring any explicit feedback from the network switches.