Objective: Perform per-flow traffic measurement (error tolerated) in a high speed with limited memory.
[more]
Lu et al (2007) ElephantTrap: A low cost device for identifying large flows (IEEE SHPI)
An algorithm is proposed to detect elephant flows. It assumes the flow rates are under Pareto distribution and the algorithm is as follows:
[more]
Mori et al (2004) Identifying Elephant Flows Through Periodically Sampled Packets (IMC'04)
Attacking the elephant and mice phenomenon (the vital few and trivial many rule) in a high speed Internet, try to determine the elephant flows. The elephant flows in this paper is defined as those who consumes more than 0.1% of link bandwidth.
[more]
Papagiannaki et al (2001) On the Feasibility of Identifying Elephants in Internet Backbone Traffic (Sprint TR)
Objective: Identify elephant flows from all the flows, given the elephants and
mice phenomenon (or mass-count disparity in statistics term) happens in the
Internet.
[more]
Griffiths, Livingood, Woundy (2008) Comcast's ISP Experiences In a Recent P4P Technical Trial (Internet Draft)
P4P study by Comcast and Yale.
[more]