SDN Abnormal Traffic Detection Algorithm Based on Rescaled Range Analysis
Haiyan Lan, Yuchen Pan · 2019
When detecting the SDN (Software Defined Network) traffic state, some math errors occur owing to a standard deviation of zero from the constant value existing in the network traffic series of several nodes. An improved R/S (Rescaled Range Analysis) method is proposed to solve this problem. The algorithm inherits the mainstream R/S method and introduces new parameters to calculate the Hurst exponent of the series. A set of available parameters is established by applying the infinitesimal method and verifying experimental data. This algorithm is combined with Mininet environment to set up a virtual SDN simulation test. It is more significant to distinguish between normal and abnormal traffic with a lower delay of detecting the moment of abnormality. This algorithm provides a new perspective for studying SDN security issues. It has reference significance for researchers to detect SDN network traffic environment.