Performance analysis of statistical multiplexer buffer in ATM networks with internal transfer mode

A statistical multiplexer with infinite buffer capacity, a finite number of inputs and underlying train arrivals is analyzed in this paper. The train arrivals model appears within the large self-routing ATM switching network, where an ATM cell is internally converted into a constant number of smaller data units called minicells, which are multiplexed and switched consecutively like a train. An alternative way to derive the mean buffer contents is described. In order to obtain the asymptotic buffer behavior, an analytical approach is developed. This approach is general, and can easily be extended to other types of correlated arrivals. The impact of train arrival processes on the buffer performance is illustrated via some numerical examples.