Performance study on the "leaky bucket" usage parameter control mechanism with CLP tagging

An attempt is made to quantify the value of using the option of setting an asynchronous transfer mode (ATM) cell header (the cell loss priority or CLP indication field), i.e., CLP tagging, at the usage parameter control (UPC) mechanism. It is assumed that the UPC mechanism uses a leaky-bucket algorithm. The statistical gain realized in the ATM network with a UPC mechanism using CLP tagging is compared to that with a UPC mechanism not using CLP tagging. It is observed that in some cases CLP tagging does not provide any improvement on the statistical gain. In those cases where statistical gain is improved by CLP tagging a proper combination of the tagging probability and the cell discarding probability engineered for the tagged traffic is essential. A larger improvement on statistical gain is obtained when the sources have very small average bursts and the network elements have small buffer.<>