Prediction of Network Traffic Using Dynamic Bilinear Recurrent Neural Network

Prediction of a network traffic using dynamic-bilinear recurrent neural network (D-BLRNN) is proposed and presented in this paper. D-BLRNN was developed to enhance the prediction capability of the BLRNN further by introducing dynamic learning control and optimization layer by layer procedure. Experiments are conducted on a real-world Ethernet network traffic data set. Results show that the dynamic BLRNN-based prediction scheme outperforms the conventional multi-layer perceptron type neural network (MLPNN) in terms of normalized mean square error (NMSE).