Many efficient approaches and analysis techniques are applied to analyze the macro and micro characterizes of network flow data.The attractors are reconstructed by making use of time-delay coordinates.Then the macro nonlinear characteristic quantities of the network flow time sequences such as the Fractal dimension、Kolmogorov entropy and the largest Lyapunov exponents are extracted in this multi-dimension phase-space.The study on the temporal characteristics of these three parameters discovered that the network flow is featured by some chaotic behaviors.The multifractal spectrums of the four difference network flow data are calculated in order to characterize and recognize the dynamic structure of network flow data more deeply

and thus can be effectively exploited for the controlling and modeling of the network behaviors.