ATM congestion control using a fuzzy neural network

This paper presents a new mechanism to control the link-by-link traffic of an asynchronous transfer mode (ATM) switch. This method makes use of the linguistic ability of fuzzy set theory and logic to handle the complexity. A fuzzy neural network (FNN) will learn to control the injection rate of the previous ATM switch by issuing a signal. The FNN will learn to follow the inference method, and decide what kind of signal should be sent based on a set of rules as in the inference method.