Research On Traffic Measurement Method Based On Traffic Characteristics In Software-defined Networks

With the development and popularization of network technology,Software-Defined Networks(SDN),as a new generation of network paradigm,has been put forward and received extensive attention.Unlike traditional network technology,SDN separates the control plane from the data plane,so that the control plane can be centralised logically,which brings convenience to the configuration and management of the networks.Traffic measurement is the basis of network configuration and management.When evaluating the performance of existing SDN traffic measurement methods,the characteristics of flow used for performance evaluation are usually simple,often ignoring the effect of flow characteristics on the performance of flow measurement methods.This paper mainly studies the traffic flow measurement method based on the traffic flow characteristics.The main contributions of the SDN traffic measurement method are as follows:1.In view of the demand for fine-grain flow measurement,the performance of the current SDN traffic measurement method is evaluated using the VBR traffic with long and intense throughput,and a traffic measurement method based on the adaptive throughput fluctuation(AdaRate)is proposed.This method calculates the fluctuation of throughput according to the recent historical statistical information of the traffic flow,and adjusts the polling frequency according to the fluctuation or not,so it almost does not need artificial parameters,and it also overcomes the problem that SWT can easily lose the peak information of throughput fluctuation when the traffic pattern is fixed.A random burst detection mechanism is also proposed to modify the measurement behavior of AdaRate.2.In order to meet the demand of more fine-grain flow measurement in VBR video flow scene,the effect of video scene switching on throughput in VBR video flow is analyzed and discussed in this paper,and the SW-ARIMA algorithm is proposed.The SWARIMA algorithm uses a long sliding window to find fragments suitable for modeling of the ARIMA model in the throughput sequence of the VBR video flow,and determines whether there is a video scene switching in the VBR video flow corresponding to the current sliding window based on the ARIMA model.When the controller determines that there is a video scene switch in the VBR video flow segment in the sliding window,the controller uses the maximum polling frequency.When there is no video scene switching in the VBR video flow corresponding to the sliding window,the controller adjusts the polling frequency to reduce the network bandwidth overhead and interpolates the unmeasured points based on the ARIMA model.Finally,the validity and practical value of the above method are verified by experiments,and the design and implementation of a SDN flow measurement framework integrating the above traffic measurement method is proposed.The user can configure the traffic measurement method by providing RESTful API with the framework.