一种新的网络流量预测算法

A New Algorithm for Network Traffic Prediction

网络流量的预测对控制网络拥塞以及提高网络利用率有着重要的意义。为了更精确地预测网络流量,本文根据已有的流量矩阵,采用跟踪最好专家的思想,通过专家跟踪预测方法来获得新的流量矩阵。该算法不需要过度地依赖历史数据,只需要很少的先前流量数据就可以获得满意的预测结果。通过实验表明:此算法的预知结果和真实的网络流量极为接近,可以应用到实际中。

Prediction of network traffic is of a great significance to control web congestion and improve network utilization. In order to forecast network traffic more accurately, this paper follows the best expert minds on the established network traffic matrix, thus acquires new traffic matrices through experts tracking method. The algorithm does not need excessive reliance on historical data, and can obtain satisfying results with only a small amount of the previous flow data. Experiments show that the prediction is very close to the real net- work traffic and it is applicable.