Fuzzy connection admission control for self-managed networks

In this paper, we present an algorithm for connection admission control (CAC) in a high speed data network. Our algorithm uses on-line measurement and off-line simulation to estimate the amount of bandwidth required to meet the request of a set of active sources and the actual use of the bandwidth. We introduce the concepts of Customer Satisfaction Factor (CSF) and Proportion of the Used Bandwidth (PUB) and identify three domains for the CAC namely, strong acceptance (with guarantee), weak acceptance (without guarantee) and rejection. The algorithm presented here is computationally efficient and does not assume any detailed statistical characteristics of the traffic sources. Sources need to declare only their mean and peak rates.