Deep learning based detection method for SDN malicious applications

Abstract

SDN is a new type of network architecture. The core technology of the SDN is to separate the control plane of the network device from the data plane so as to achieve flexible control of network traffic. Such structure and characteristics have put forward higher requirements on the security protection capability of the SDN controller. However, there are still less researches on malicious applications for the SDN network architecture. This article aims at this problem, based on the analysis of the existing malicious application detection methods and on deep learning technology proposed by a detection method for SDN malicious applications. Finally, under the TensorFlow deep learning simulation environment Keras, 30 SDN malicious samples were studied and tested. The experimental data show that the detection rate of this method for malicious applications can reach 89%, which proves the feasibility and scientificity of the program.