Call admission control in ATM networks using neural networks trained with virtual cell loss probability method

In this paper, we propose a Neural Network (NN) approach to estimate Virtual cell loss probability (VCLP) of bursty sources for Call Admission Control (CAC) purpose in Asynchronous Transfer Mode (ATM) environment. Based on this approach, we have presented two schemes of estimating cell loss probability (CLP). For both the schemes training data set are obtained from VCLP methods advocated by, and this training is done off-line to estimate CLP in real time environment. While the first method performs consistently well to withstand changes in burst duration parameters, the second one is also suitable from the point of view of individual call quality. In order to discuss performance aspects the methods have been compared with other cell loss estimation methods. Our simulation shows that NN approach outperforms the conventional methods in terms of accuracy.

Actions (login required)

View Item