|
Abstract : |
This study deals with the monitoring of ATM cells from a CBR source which are subject to cell delay variation introduced by cells from other tra c sources. We present an exact solution for the dimensioning of the so-called peak cell rate monitor algorithm de ned by the CCITT Draft Recommendation I.371. The solution algorithm consists of a discretetime iteration and allows for the exact determination of probability that the monitor algorithm rejects a cell. Using the cell rejection probability, the parameters of the monitor algorithm can be chosen to meet the desired performance objectives like transparency and ability to detect misbehaving sources. We consider cells from a CBR source which are multiplexed together with background tra c. The background tra c can vary by its mean and coe cient of variation of the cell rate. The cell delay variation of the CBR tra c introduced by the background tra c is taken into account by the monitor algorithm. The numerical results show that the parameters of the monitor algorithm depend strongly on the coe cient of variation of the background tra c. 1, |
