A fuzzy-based traffic controller for high-speed ATM networks using realistic traffic models

In modern ATM networks, usage parameter control (UPC) is necessary in order to ensure conformity of the incoming data streams to its negotiated parameters. This protects the quality of conforming applications from degradations that are due to additional network loading produced by nonconforming applications. The paper focuses on the congestion problem created in an ATM switch by the statistical multiplexing of various traffic sources. The objective is to achieve the maximum possible statistical multiplexing gain without violating the quality of service requirements of the individual streams. A fuzzy traffic controller is proposed, which performs policing and shaping of the input streams so that the above mentioned goals are satisfied. The controller is a fuzzy implementation of the two-threshold leaky bucket mechanism that is widely used in commercial ATM switches.