Traffic modeling for non-bursty and bursty traffic

Abstract EN

This Paper discusses the mathematical modeling of the source traffic using Poisson, Pareto, and Weibull distributions along with performance comparison considering these three types of traffic generators: (1) Poisson distribution for modeling the BL ( BLPos ) and Exponential distribution for modeling the GT (GT Exp ), the corresponding traffic generator represented by / BL GT Pos Exp ; (2) Pareto distribution for modeling the BL ( BLPar ) and Pareto distribution for modeling the GT (GTPar ), the corresponding traffic generator represented by / BL GT Par Par ; and (3) Pareto distribution for modeling the BL ( BLPar ) and Weibull distribution for modeling the GT (GTW b ), the traffic generator represented by / BL GT Par Wb .

Nonbursty traffic was modeled vsing Poisson distribution for burst length and exponential distribution for gap time whereas bursty traffic modeling was achieved through heavy tailed Pareto and Weibull distributions.

The comparison between the three traffic generators has been verified through simulation for six sources Examining the simulation results for Allowed Cell Rate (ACR) and Memory Access which indicate the performance of the switch under / BL GT Pos Exp , / BL GT Par Par , and / BL GT Par Wb traffic generators, It is seen that the switch offers best performance under / BL GT Par Wb traffic generators.