Queue delay estimation and its application to TCP Vegas

Authors:
Kyungsup Kim;Chong-Ho Choi
Affiliations:
School of Electrical Engineering and Computer Science, Seoul National University, Shinlim-dong, Kwanak-ku, Seoul 151-742, South Korea;School of Electrical Engineering and Computer Science, Seoul National University, Shinlim-dong, Kwanak-ku, Seoul 151-742, South Korea
Venue:
Computer Networks: The International Journal of Computer and Telecommunications Networking
Year:
2003

Citing 7
Cited 0

Tracking and data association

Tracking and data association
Congestion avoidance and control

SIGCOMM '88 Symposium proceedings on Communications architectures and protocols
A control-theoretic approach to flow control

SIGCOMM '91 Proceedings of the conference on Communications architecture & protocols
Random early detection gateways for congestion avoidance

IEEE/ACM Transactions on Networking (TON)
Understanding TCP Vegas: a duality model

Journal of the ACM (JACM)
A study of a high speed packet scheduling method for achieving fairness among TCP connections

Journal of High Speed Networks
TCP Vegas: end to end congestion avoidance on a global Internet

IEEE Journal on Selected Areas in Communications

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes a modified Vegas algorithm, which can be adjusted to take a fair share of the bandwidth compared to other transmission control protocols (TCPs). In order to do this, we investigate the maximum round-trip time (RTT) achievable in steady-state flows fully utilizing available link capacities without marking packets. With this, the marking queue delay is defined. We also propose an estimation algorithm for the marking queue delay in TCP data traffic by using the Kalman filter. The modified Vegas algorithm supplements the shortcomings of Vegas such as unfairness due to over- or under-estimations of the round-trip propagation delay.This study analyzes the model of the marking probability in active queue management like random early detection in terms of RTT. This algorithm gives a robust estimate of the marking queue delay, and the Vegas algorithm in the congestion avoidance phase is modified using this estimation. The simulation results validate the advantages of the proposed algorithm in terms of performance and fairness.