SDN下基于人工智能算法的DDoS攻击检测研究

Research on AI-based DDoS Attack Detection in SDN Networks

软件定义网络技术(SDN)基于分层的思想将数据与控制分离,可以通过软件实现对网络控制的编程,实现灵活的网络管理和敏捷的应用开发,同时也引来了一些新的安全问题,如DDoS攻击。针对SDN网络中DDoS攻击的检测问题,构建基于BPNN的DDoS攻击检测系统,包括流表收集、特征提取、数据训练、攻击检测、攻击处理5个模块。其中攻击检测模块调用BPNN对特征值进行分类,并做出不同的处理。最后,经过实验测试分析,可知:使用BPNN可以更快地产生检测模式;当攻击规模较大时,检测可以达到较高的准确率和查全率。

Software defined network(SDN) divides data and control based on the thought of separation. It can realize programming of network control through software, flexible network management and agile application development, but also attracts some new safety problems, such as DDoS attacks. Against DDoS attack detection problem in the SDN network, we construct the DDoS attack detection system based on BPNN, including flow table collection, feature extraction, flow table training, attack detection, and attack call processing five modules.The attack detection module calls BPNN to classify the eigenvalues and do different processing. Finally, after experimental test analysis, it can be seen that: using BPNN can generate the detection mode faster. When the scale of attack is large, the detection can achieve high accuracy and recall.