Abstract: Due to the nonlinear properties in the network traffic，the correlation dimension is introduced to study the chaotic characteristic of network traffic.At first，the phase space of LAN traffic is reconstructed by using C-C arithmetic and the optimal time delay is proposed.Then，the relationships between correlation dimension and the network traffic are analyzed in detail.Further，the effect of the phase space parameter on the correlation dimension by phase space parameter is studied.The result shows that correlation dimension can describe the complexity of network traffic，if the time delay is chosen at a certain range.The correlation dimension and saturated embedding dimension increase along with the using of network，but on the contrary，the relative correlation dimension will become less as the average traffic is increased.The result can be considered as a fundamental route for studying the correlations of the network traffic factors.

Key words: LAN traffic, phase space reconstruction, correlation dimension, correlation

摘要： 鉴于网络流量复杂的非线性特征，引入关联维数来研究网络流量的混沌特性。首先，采用C-C算法重构局域网流量的相空间，确定出最佳延迟时间。然后，分析了重构相空间的关联维数与网络流量的关联关系。进一步，研究了相空间重构参数对关联维数准确度的影响。研究结果显示，将重构相空间的延迟时间确定在一定范围内，关联维数可以较准确地反映网络流量的复杂程度。关联维数和饱和嵌入维数随着网络使用量的增加而增加，而相对关联维数随着网络平均流量的增加而降低。分析结果为进一步研究网络流量的各策动因素的关联特性奠定了理论基础。

关键词: 局域网流量, 相空间重构, 关联维数, 关联性