Intelligent Active Queue Management Predictive Controller using Neural Networks

Network congestion control, due to increasing number of Internet services with various quality of service (QoS) demands, has become the focus of current research. In this paper, a novel scheme of adaptive Smith Predictor (SP) controller using Neural Network for active queue management (NNAQM) is presented. Smith predictor is used to overcome the disadvantages such as influence of time delay, in particular, when it becomes significant in large TCP/IP networks. In this investigation, the well known Back-Propagation (BP) algorithm is used to train weights of the neural networks of the proposed design. Finally, a simulation platform is developed, tested and validated to demonstrate the merits of the scheme through a set of experiments.