ABSTRACT

This chapter introduces Asynchronous transfer mode (ATM) networks and their Connection Admission Control (CAC) and describes the problems in conventional methods for cell loss ratio (CLR) estimation. It discusses the feasibility of the fuzzy inference approach and the CLR estimation by using the fuzzy inference based on a weighted mean of fuzzy sets. In the methods, the CLRs are observed in ATM switches, and the learning systems extract the relation between the CLR and the number of connections in transmission rate classes. The relation between CLR and the number of connections is often nonlinear. The fuzzy sets in each fuzzy rule are automatically extracted and tuned by the learning algorithm explained later. The inference method based on a weighted mean of fuzzy sets can control the fuzziness and specificity of its final inference consequence by tuning the width parameters of membership functions in then-parts.