Call admission control in ATM using the diffusion model

Abstract
ATM technology will support a wide variety of traffic needs, and satisfy a range of quality needs and network performance objectives. Thus call admission control algorithms for ATM must be robust to accommodate various types of traffic and simple enough to be implemented in real time. A call admission control procedure is proposed using statistical bandwidth as the decision criterion. The statistical bandwidth takes into consideration the users' cell loss requirement, traffic characteristics and buffer size at statistical multiplexers, and is computed using diffusion approximations. Extensive simulation experiments show that our novel approach can significantly improve resource utilization compared to existing CAC (call admission control) criteria, while maintaining conservative bounds for the users' cell loss criterion.