Research On DDoS Attack Detection Under The Background Of Cloud Computing

In recent years,with the rapid development of cloud computing technology,people can more conveniently obtain storage and computing resources,and use cloud services to perform more complex computing tasks.However,a series of security problems such as Distributed Denial of Service(DDoS),system vulnerabilities,data leakage and malicious insiders have brought serious challenges to cloud computing.Among them,DDoS attack can take advantage of network vulnerabilities and consume the computing resources of the attack target,thereby causing the target host to crash.It is one of the most threatening attack methods in modern information networks.In addition,some new and unique forms of DDoS attacks have emerged in the cloud computing environment,such as insufficient bandwidth DDoS attacks,EDo S attacks,etc.In order to better defend against DDoS attacks against targets,this thesis proposes a DDoS attack detection model based on relative entropy and Convolutional Neural Network(CNN),and simulates related research using the Open Stack cloud computing platform.This thesis aims at the problem of DDoS attack detection in the context of cloud computing,the main contents are as follows.(1)A DDoS attack detection model based on relative entropy and CNN.First,in view of the problems of poor detection accuracy and slow speed of DDoS attack detection in existing research,this thesis proposes a detection model that combines statistical analysis methods and deep learning methods.Specifically,this study uses a statistical analysis method of relative entropy for the initial detection.This method monitors port traffic based on relative entropy to determine whether there is abnormal traffic,and extracts traffic characteristics after detecting suspected abnormal traffic.Subsequently,this study uses a deep learning method for secondary detection,which preprocesses the suspected abnormal traffic into a two-dimensional array,and uses the CNN re-inspection module to achieve accurate detection.(2)Application of DDoS attack detection model based on relative entropy and CNN in cloud computing environment.In order to verify the performance and advantages of the DDoS attack detection model proposed in this thesis in the cloud environment,this thesis studies and builds an experimental environment based on the Open Stack cloud computing platform.Use the same data sample set to conduct simulation experiments on the classic Long Short-Term Memory(LSTM),K-Nearest Neighbor(KNN)and Support Vector Machines(SVM)algorithm detection models,as well as the detection model in this thesis,the experimental results were compared and analyzed.The comparison results show that the DDoS attack detection model proposed in this thesis can operate effectively in the cloud computing environment,and it has higher advantages in detection accuracy,bit error rate,detection time and detection rate.In addition,an experimental analysis is carried out on the influence of the training sample data volume on the model accuracy.When the training sample data volume increases,the model performance is close to the performance under the initial sample data volume training.