An Elephant Flow Detection Method Based on Machine Learning

Software-Defined Networking (SDN) is regarded as the next generation network. Current network is difficult to be configured and managed, and SDN is proposed to change this situation, which makes it attract a lot of attention of the academia and industry. The detection of Elephant Flow is an important service of SDN, based on which we can achieve the management of the network traffic and implement services such as the load balancing of traffic, congestion avoidance and so on. This paper focuses on the iterative method to detect Elephant Flow. We propose a method which uses the random forest to learn the arguments produced in the iterative detection and to improve the accuracy and speed of the detection. The experiments show that our method can efficiently improve the accuracy and speed of the detection compared to other methods.