Hurst weighted random early detection algorithm based on self-similar traffic input

Abstract

RED as proposed by Floyd was based on the traditional Poisson input model,could not satisfy self-similar na-ture of network traffic.With this in view,an attempt has been made by proposing a modified RED algorithm,the HWRED algorithm with self-similar input.Modified RED algorithm can tune RED parameters according to the Hurst parameter of input traffic.Simulated results show modified algorithm increases evidently stability of queue size,and de-creases packet loss ratio,queue delay and queue delay variability,finally improves link use percentage of network.

Keywords

• congestion management;self-similar input;random early detection;IP networks