Design And Implementation Of SDN Network Abnormal Traffic Detection System

With the continuous development of the Internet,traditional IP networks have become increasingly difficult to manage.For this reason,researchers put forward the concept of Software Defined Network(SDN),which solves the problems of configuration difficulties in traditional networks.However,as a new network architecture,the SDN network also faces serious network security problems.Therefore,how to prevent network attacks against the SDN network has become a current hot issue.Abnormal traffic detection technology is a commonly used method to prevent network attacks.In order to solve the network security problems existing in SDN networks,based on the research of abnormal traffic detection technologies at home and abroad,the K-nearest neighbor algorithm(KNN)and Supported Vector Machine(SVM)are closely related,and an abnormal traffic detection method based on incremental learning KNN-SVM algorithm(IL-KSVM)is proposed;at the same time,the method combines A new incremental learning method can learn a new sample set based on the original detection model.In order to solve the difficult problem of hyperparameter selection in the IL-KSVM algorithm,the particle swarm optimization algorithm(PSO)is used to optimize the IL-KSVM algorithm.The PSO algorithm can find the optimal parameter set of the algorithm in a short time.It has a great advantage compared with other search algorithms.In order to apply the IL-KSVM algorithm to the SDN network,a SDN network abnormal traffic detection system is designed,and the specific implementation of the system is given.In order to simulate the SDN network,the Mininet and Ryu controller are used to build the SDN network environment,and the IL-KSVM algorithm proposed in the Ryu controller is used to design and implement the SDN network abnormal traffic detection system.Through experiments,the effectiveness and reliability of the system are verified,and it is proved that the IL-KSVM algorithm has better detection performance and practicability for SDN network traffic.