网络流量的联合多重分形模型及特性分析

网络尺度行为的发现提供了用数学模型方法研究网络流量特性的可能性.本文基于连乘瀑布过程与K分布过程提出了联合多重分形(JMF)网络流量模型,该模型以尺度函数与矩因子的联合作为主要特征函数来研究网络流量的特性.理论分析及由实测网络流量数据的仿真结果表明,JMF模型可以较客观地同时描述网络流量短期分形行为与长期自相似行为,且实现复杂度小.其中尺度函数能够刻画时间尺度对流量特性的影响,矩因子描述了同一时间尺度上流量突发性的变化,二者的联合较好地描述了网络流量的短期行为,而模型的统计特性则刻画了流量的长期行为特征.

Joint Multifractal Model and Characteristics Analysis of Network Traffic

The discovery of scaling behavior in network has provided hope that mathematical models can be used to describe the nature of the traffic.A joint multifractal (JMF) model for network traffic is proposed based on the combination of multiplicative cascade process and independent K distributed process.The model focuses on both the scaling function and moment factor as the key for unraveling the multifractal characteristics of the network traffic.JMF model is objective and simple enough to describe the short-term fractal behavior as well as the long-term self-similar behavior of network traffic simultaneously,which is proved by theoretical analysis and simulated results from several real traffic data sets.The influence on the traffic characteristics due to timescale is expressed by scaling function,and traffic burstiness variation in a specified timescale is illuminated by moment factor.The combination of scaling function and moment factor describes the short-term behavior of network traffic,and the statistical properties of JMF model describe the traffic long-term behavior characteristics.