QoS routing based on artificial neural networks
基于人工神经网络实现的QoS路由决策

This paper discussed a solution for QoS routing based on neural networks and learning algorithms. In order to confirm the feasibility of QoS routing based on artificial neural networks, the experimental environment consisted of 17 interconnected servers (nodes),designed a network of several neurons on each node. Through learning algorithms, these neurons offered dynamic routing based on on-line network monitoring data. Experimental results indicate that routing decision provided by the neural network with minimum hops or minimum delay being the QoS goal can approach the best QoS. Moreover, when the neural network consideres minimum hops and minimum delay together, the overall delay of the network is smaller than the case where only one factor is considered.These results prove that the distributed routing decision on nodes provided by artificial neural network can optimize the QoS parameters.