基于切片的深度学习SDN恶意应用程序的检测方法

SDN是一种新型网络架构,其核心技术是通过将网络设备控制面与数据面分离。然而目前针对SDN网络架构的恶意应用程序研究还较少。针对这一问题,在总结分析现有恶意应用检测方法的基础上,采用代码切片技术并基于深度学习框架提出一种面向SDN恶意应用程序的检测方法。它旨在对样本进行模块化分割并提取特征后,将特征向量以矩阵形式重组。在TensorFlow深度学习环境Keras下对SDN恶意样本进行学习和检测,实验数据表明,该方法对恶意应用程序检测率可以达到93.75%,证明了方案的可行性和科学性。

SLICE-BASED DEEP LEARNING FOR DETECTING MALICIOUS SDN APPLICATIONS

SDN is new network architecture.Its core technology is to separate the control surface of network equipment from the data surface.However,there is little research on malicious applications for SDN network architecture at present.On the basis of summarizing and analyzing the existing malicious application detection methods,we propose a detection method for SDN malicious application based on code slicing and deep learning framework.It aimed at modularizing the samples and extracting the features,then reorganizing the feature vectors in matrix form.The malicious samples of SDN were studied and detected under the TensorFlow in-depth learning environment Keras.The experimental data show that the detection rate of malicious applications can reach 93.75%,which proves the feasibility and scientificity of the scheme.