An efficiency study of different model-based and measurement-based connection admission control techniques using heterogeneous traffic sources

Numerous research has been done on modeling the traffic flows in a multiservice network. An alternative proposal was made by Addie et al. (1998). The authors argued that because the central limit theorem can be applied to heavy traffic processes, therefore as more traffic is aggregated together through the sharing of a transmission link, the aggregated traffic will weakly converge to a Gaussian process. In this paper, we apply this Gaussian model to the connection admission control (CAC) of a multiservice network. We consider a store-and-forward (packet or cell-switched) connection-oriented multiservice network that can be based on the concept of asynchronous transfer mode. We also make efficiency comparison with a CAC based on an effective bandwidth model. Two scenarios were considered: the first assumes that all a priori traffic specifications are known before a CAC acceptance/rejection decision is made, while the second assumes that only the peak rate of a new (requesting admission) connection is declared to the CAC mechanism. The latter scenario entails a measurement-based CAC approach.