[e2e] Traffic Burstiness Survey

[Richard G. Clegg]

On 10/09/12 09:22, Neil Davies wrote:
> I've always found the coefficient of variance of the offered load to be a very useful indicator.
>
I agree it is useful but one can construct hypothetical traces with the 
same CoV representing a constant linear increase (perfectly smooth) and 
sin wave of arbitrary frequency.  So perfectly smooth traffic and very 
ill-behaved traffic can have identical CoV.  (But one can construct 
similar counter examples with almost any measure picked).

Wischik and Ganesh's concept of Hurstiness is a mathematically rigorous 
quantity developed for analysing the burstiness of traffic queues which 
has some attractive known results for buffer asymptotics.

http://www0.cs.ucl.ac.uk/staff/ucacdjw/Research/hurstiness.html

-- 
Richard G. Clegg,
Dept of Elec. Eng.,
University College London
http://www.richardclegg.org/