Jun Lv, Tsinghua University, China
Xing Li, Tsinghua University, China
Tong Li, Tsinghua University, China
Internet traffic analysis, models simulation and prediction play a very important part in the network management and design. This paper shows the new thought in network traffic modelling and prediction, and proposes a new algorithm to modify the model coefficient continuously. By far, most of models have not been considered to possess the adaptive ability. Our goal is to improve the model adaptive ability and make the model parameters update dynamically. We present the modified LSL algorithm, which can modify the parameters with each new input data update and has the property of fast convergence and high accuracy.
Citation:
Jun Lv, Xing Li, Tran Quang Anh, Tong Li, "A New Algorithm for Network Traffic Prediction," iscc, pp.1006-1012, 11th IEEE Symposium on Computers and Communications (ISCC'06), 2006