Adaptive per-application load balancing with neuron-fuzzy to support quality of service for voice over IP in the internet

Abstract

This paper presents a per-application load balancing with adaptive feedback control to optimize both QoS requirement for VoIP application (first target) and effective link usages (second target) over the Internet. In order to forward the Internet applications over the route coupling with respect to QoS for VoIP, the perceptron neural net is used to classify the applications with similar properties belong to one of two groups. However, the perceptron has some limitations such as non-optimization for several targets. Therefore, this paper proposes to develop the perceptron with fuzzy control approach to eliminate the limitation of perceptron and enhance the capability of perceptron.