Enhancement to RED Algorithm

Authors:
Ren Fengyuan;Ren Yong;Shan Xiuming
Affiliations:
Tsinghua University, China;Tsinghua University, China;Tsinghua University, China
Venue:
ICON '01 Proceedings of the 9th IEEE International Conference on Networks
Year:
2001

Citing 0
Cited 2

Active queue management algorithm considering queue and load states

Computer Communications
Random early detection for congestion avoidance in wired networks: a discretized pursuit learning-automata-like solution

IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics

Quantified Score

Hi-index	0.00

Visualization

Abstract

RED (Random Early Detection) has gained rathermore attention as AQM (Active Queue Management)recommendation algorithm. The instability and difficultyin tuning control parameter settings are two greatobstacles. In this study, we first identify several designproblems that lead the oscillation occurred in steadystate, such as, the control law is too simple to overcomethe effect caused by the queue system inertia, thenintroduce the varying speed of average queue size as anew state variable to predict the trend of the networkload condition, and modify the dropping function usingthe new variable as enhancement to the standard REDalgorithm. The simulation results verify the effectivenessof the enhanced algorithm. The enhancement makes theRED constrain the oscillation and become more robustand adaptable.