Research On Spatiotemporal Characteristic Of Network Traffic Based On Flexible Multiscale Entropy

With the continuous development of the Internet,network has become an indispensable part of people's life.How to ensure network security,stability and efficient operation has become an urgent problem need to be solved in current network.As the carrier of data flow in network,the study of characteristics of network traffic can help us understand complex network structure and dynamic characteristics of network,which has import significance for the design,implementation and monitoring of network.Based on the research of multiscale entropy theory and the analysises of network traffic characteristics,this thesis proposes a flexible multiscale entropy theory based on the theory of multiscale entropy,and makes an innovative study on the spatial and temporal characteristics of network traffic.The main research work and contributions of this thesis are as follows:(1)In this thesis,we described the entropy theory and the characteristics of network traffic in detail.Including sample entropy,multiscale entropy theory,composite multiscale entropy theory.In the research of network traffic characteristics,the method of network traffic flow,the measurement of network traffic,and the characteristics of network traffic are described in detail.(2)The theory of Flexible Multiscale Entropy(FMSE)is proposed to improve the accuracy and stability of the time series complexity analysis.Firstly,based on the sample entropy,the method of flexible accumulation is introduced.On the basis of the method of combination calculation of composite multiscale entropy theory,the flexible multiscale entropy is put forward.Finally,based on the artificial white noise,1/f noise and real vibration data time series,the validity of flexible multiscale entropy theory is verified.The experimental results show that the flexible multiscale entropy is more accurate and stable than multiscale entropy and composite multiscale entropy.(3)An analysis method of spatiotemporal characteristics of network traffic based on the flexible multiscale entropy is proposed.Firstly,divide the open network traffic data into sub sequences of different levels and spatiotemporal dimensions respectively.Then,use flexible multiscale entropy proposed in this thesis as a theoretical tool to calculate the complexity of the divided traffic data.By studying the complexity of network traffic in different size of network,the complexity of the packet sequence is higher than flow sequence in whole scale.The complexity of bytes sequence is higher than the complexity of the flow sequence in most times,and the complexity of the flow sequence is higher than the complexity of the sequence of bytes when a lot of small flows are occurred.Small flow has important influence on the complexity between bytes sequence and flow sequence.(4)Summarizes the main work and contribution of the flexible multiscale entropy theory and the study of network traffic spatiotemporal characteristics based on the flexible multiscale entropy theory.