【摘要】 多媒体网络流量具有重尾特性及自相似特性,已得到了广泛的认同。而Alpha-stable过程作为Gaussian过程的推广能很好地描述重尾特性及自相似特性。在简要介绍了Alpha-stable分布和Alpha-stable过程的基础上,本文研究分析了当前基于线形分形稳定噪声(LFSN)理论的网络业务流量模型,并提出了基于差分分析的新方法及基于差分分析的LFSN流量模型参数估计新方法。对Bellcore实验室采集数据的拟合验证结果表明,该估计方法能准确确定模型参数。更多还原

【Abstract】 The fact that multi-media network traffic is with heavy-tail character and self-similar character has been known to researchers. At the same time, Alpha-stable process, the extension of Gaussian process, has the advantage of modeling heavy-tail character and self-similar character. Alpha-stable distribution and Alpha-stable process are introduced at the beginning of this paper. Then after analyzing of current network traffic models based on Alpha-stable process and Linear Fractional Stable Noise (LFSN), we advance a differential analysis method and a new parameter estimation method based on difference analysis. Results of analysis of actual traffic data from Bellcore show the method can get parameters precisely.更多还原