A MAP Fitting Method to Approximate
Real Traffic
Behavior
András Horváth, Gergo István Rózsa,
Miklós Telek
Abstract:
This paper provides a heuristic fitting method to capture some
important features of real traffic sources by a Markovian arrival
process (MAP). The novelty of the proposed approach lies in the
separate treating of short and long range behavior of the considered
traffic sources. The proposed MAP is the superposition of two
elementary processes. A Phase type renewal process, whose interarrival
time exhibits heavy-tail behavior over some time scales, is used to
capture the long range dependent behavior, i.e., the empirical Hurst
parameter. While a two-state Markov modulated Poisson process (MMPP)
is applied to approximate the short range behavior. Different analysis
techniques are used to evaluate the goodness of the proposed fitting
method.
Postscript
András Horváth, 2008-06-25