Comparison of hyperexponential and group Poisson traffic models of multiservice communication networks

The goal of the study was to compare hyperexponential and group Poisson flows of requests as models of burst traffic of multiservice communication networks. Various traffic models are considered for the problem of approximating the average queue created by video traffic of modern multiservice networks. Poisson and hyperexponential flows, both ordinary and group, are also considered. It is shown that group flows of both types are suitable for the task of approximating the video traffic queue. Group Poisson flows make it possible to obtain very simple analytical dependences of the average values of queues on the load factor of queuing systems and are therefore the most preferable. The inadequacy of using ordinary hyperexponential models for approximating burst flows of requests is also provided shown, since, at low values of the load factor, the dependences of the average values of queues for hyperexponential flows are practically close to zero, while the indicated dependences for burst flows have a very significant slope angle. The conclusions drawn are confirmed by the results of simulation modeling.