Peakedness of stochastic models for high-speed network traffic

In networks based on the asynchronous transfer mode (ATM), information is transmitted asynchronously over high-speed links in the form of 53-byte units called cells. Accurate traffic characterization is a crucial step in performing network resource allocation and dimensioning. Peakedness was originally developed by teletraffic engineers as a tool for characterizing call arrival processes at a trunk group. We generalize the peakedness theory to include a class of stochastic models used in studies of high-speed networks and apply it to the approximate analysis of a statistical multiplexer.