An Adaptive Flow Measurement Method for VBR Video Flow in Software-Defined Networks

Abstract

The growth of mobile network scale has brought about the outbreak of mobile traffic, and the rapid growth of mobile traffic has brought challenges to network management and optimization. Effective network management and optimization require correct network parameters, so traffic measurement is very important for network management and optimization. In SDN networks, the adaptive flow measurement method is sometimes used according to the measurement requirements because the network overhead of traffic measurement is often proportional to the accuracy, and the bandwidth resources are tense. This paper mainly focus the research on adaptive flow measurement for VBR video flow. We analyze the influence of video scene switching on the throughput of VBR video flows, and propose an adaptive flow measurement method(SW-ARIMA) for VBR video flow. SW-ARIMA can be adaptive to adjust polling for the video scene switching, and the unmeasured point caused by adjusting polling could be interpolated based on ARIMA model in time. Finally, we evaluate the accuracy and overhead of SW-ARIMA with mininet, and the result present that SW-ARIMA is more accurate than the existing adaptive SDN traffic measurement methods, such as Payless, SWT and AdaRate, and has less overhead than polling on fixed period.