A fuzzy approach to connection admission control in 802.11b wireless systems

Abstract

Connection admission control provides support for quality of service across heterogeneous networks. By restricting access through a traffic policing policy the network is prevented from becoming oversubscribed, in turn reducing latency for real-time variable traffic. Traditional methods to restrict admission involve the use of simple traffic descriptors (peak cell rate, cell loss ratio and cell to cell delay) that may not accurately characterise the network, and result in over cautious admission policies. Here we present a method using measurements taken from 802.11b WLAN devices to estimate the network state and then regulate admissions through a fuzzy logic controller (FLC). The FLC accurately determines network congestion by measuring network load and error rate. When the data transmissions are compared with a mechanism using a best effort service the proportion of successful admissions is significantly higher.