Loss correlation for queues with bursty input streams

The authors evaluate and characterize the correlation between packet losses for two queuing systems in discrete time that are motivated by broadband ISDN (BISDN) applications. The first, a two-class discrete-time queuing system, approximates the output queue of an asynchronous transfer mode (ATM) switch. The queue serves periodic foreground traffic and random background traffic. The background traffic is modeled as independently identically distributed batches of arbitrary distribution. It is shown that the conditional loss probability (CLP) is independent if the buffer size is at least as large as the period of the foreground traffic. Example calculations indicate that losses occur essentially randomly as long as the foreground traffic uses less than 10% of the channel capacity. The second analysis derives the CLP seen by a selected stream in a slotted finite-buffer system with a superposition of interrupted Poisson sources. Here, the total number of arrivals may be correlated from slot to slot. Traffic correlation is seen to have a strong influence on loss correlation, while buffer size is seen to have virtually none