【摘要】 拟通过单向时延的历史记录对其未来趋势进行预测。首先,通过粗粒度和细粒度两种方式采集世界各地若干目标节点的单向时延作为原始数据;然后,结合多种预测模型的特点和原始数据的内在属性,提出一种基于ARMA模型的双路径差异性分析方法;最后选取合理的预测窗口对方法的准确性进行验证。结果表明,该预测方法是正确、合理的,与灰色预测模型相比能更有效地预测单项时延差序列。更多还原

【Abstract】 In this paper we try to forecast the future trend of one-way delay through its historical record: First,we gather one-way delays of several target nodes around the world as the original data through both coarse-grained and fine-grained ways; After that,according to the characteristics of various prediction models and the intrinsic properties of original data,we propose an ARMA model-based dual-path difference analysis method; Lastly,we select the reasonable prediction window to verify the accuracy of the method. Results show that the prediction method is correct and reasonable,compared with the grey forecasting model it can more effectively predict the difference sequence of one-way delay.更多还原