Efficient modeling of merging and splitting processes in large networking structures

Simple models are developed for the description of a first-order Markovian (bursty) process modulated by merging and/or (independent or correlated) splitting operations. The models can be adopted for the packet traffic description in large networking structures, supporting multimedia packet traffic. This is due to the fact that the complexity of these models does not change as the number of points of transformation increases. The bursty traffic model is capable of describing a variety of packet sources. The induced packet delay at the merging points is evaluated. A queuing system is studied for this purpose under a general class of arrival processes, which include the adopted packet traffic models. Numerical results are obtained and are compared with simulations whenever approximations are involved.<
>