Research On DDOS Attacks Detection And Defense Technology Based On Deep Learning Hybrid Model In SDN

Distributed Denial of Service(DDoS)has become the most serious aspect of network security.Existing DDoS detection and defense methods have problems with high detection delay,low detection accuracy and high false alarm rate.In addition,these methods can't detect new types of DDoS attacks.SDN(Software Defined Network)is a new type of network architecture and has great potential for development.The core idea of SDN is to separate the control plane and data plane,and concentrate intelligence on the controller for centralized control.Although it solves many hidden dangers of traditional network,SDN is a new technology with many problems.Among them,the security problem is one of the important problems.The internal security risk of the SDN is mainly the imperfection of the SDN protocol,and the external security risk is mainly the DDoS attacks.From a technical point of view,it is easy to launch DDoS attacks,but the method of defending against DDoS attacks is less effective.In this thesis,we research the security problems of SDN,and we propose a DDoS detection method based on the Deep Learning Hybrid Model(DCNN-DSAE)to improve classification accuracy of network traffic in SDN.The first level of DCNN-DSAE model is C3P2F2 model based on the Convolutional Neural Network,and the second level is SAE4 model based on the Stacked AutoEncoder.DCNN-DSAE model can not only improve the accuracy of classification,but also shorten the processing time of classification.Thus this method can effectively defense DDoS attacks and avoid resource exhausted in SDN.In this thesis,we will deploy DDoS detection module in control plane of SDN,and develope a DDoS detection and defense system based on hybrid model.This method can not only detect DDoS attacks in data plane but also control plane of SDN.Finally,we launch different types of DDoS experiments to verify the effectiveness of this architecture,and we achieve the effect of defense DDoS attacks.