End-to-End Delay Prediction by Neural Network Based on Chaos Theory

Internet End-to-end delay is an important QoS metric. It can be used in real-time application designing and optimizing - control system based on the Internet for example. In such applications, precisely predicted delay is needed. However, due to the complex of the Internet, end-to-end delay is dynamic and thus hard to predict. This paper first shows the jitter of end-to-end delay is chaotic based on a number of probing data sets, and use chaos neural network to predict jitter. Simulation results show chaos neural network is effective in jitter predict.