A neural-net based fuzzy admission controller for an ATM network

This paper proposes a neural fuzzy connection admission control (NFCAC) scheme, which combines benefits of fuzzy logic controller and learning abilities of the neural-net, to solve the connection admission control (CAC) problems in ATM networks. Fuzzy logic systems have been successfully applied to deal with the traffic control related problems and provided a robust mathematical framework for dealing with "real-world" imprecision; multilayer neural networks are capable of producing complex decisions with arbitrarily nonlinear boundaries and they have been used as a solution for CAC. However, the application of a neural network or a fuzzy logic system to CAC presents some difficulties in a real system operation. The proposed NFCAC solves the difficulties by combining the benefits of the existing traffic control mechanisms, linguistic control strategy of the fuzzy logic controller and the learning ability of the neural net. Simulation results show that the proposed NFCAC saves a large amount of training time and simplifies the design procedure of a CAC controller but provides a superior system utilization, while keeping the QoS contract, than either the neural network or fuzzy logic system does.