Aggregate bandwidth allocation of heterogeneous sources in ATM networks with guaranteed quality of service using a well-trained neural network
Watit Benjapolakul, T. Rangsihiranrat · 2002
This paper proposes an application of neural network (NN) for aggregate bandwidth allocation of heterogeneous sources in ATM networks. The proposed method allocates bandwidth to guarantee the Quality of Service (QoS) for different service classes. The previous training algorithm, adaptive learning rate, was not efficient enough to recognize the relationship between traffic source parameters and their corresponding bandwidth. Thus, the Levenberg-Marquardt Algorithm is employed for training the neural network. The results show that neural network method trained by the Levenberg-Marquardt algorithm is a promising and effective method to accurately and immediately allocate the bandwidth requirement leading to higher resource utilization and fast response.