Due to a large number of bursty traffic sources that an ATM network is expected to support, controlling network traffic becomes essential to provide a desirable level of network performance with its users. Admission control and traffic smoothing are among the most promising control techniques for an ATM network. To evaluate the performance of an ATM network when it is subject to admission control or traffic smoothing, we build a discrete-time single-server queueing model where a new call joins the existing calls.
In our model. it is assumed that the cell arrivals from a new call follow a general distribution. It is also assumed that the aggregated arrivals of cells from the existing calls form batch arrivals with a general distribution for the batch size and a geometric distribution for the interarrival times of batches. We consider both finite and infinite buffer cases, and analytically obtain the waiting time distribution and cell loss probability for a new call and for existing calls. Our analysis is an exact one. Through numerical examples, we investigate how the network performance depends on the statistics of a new call (burstiness, time that a call stays in active or inactive state, etc.). We also demonstrate the effectiveness of traffic smoothing to reduce network congestion.