Detecting DDoS Attacks with a Deep Mode
Dake He · 2008
For defending DDoS attacks effectively,the more information about the attacks is required.However,the present methods only stress the existence of attacks,so they are hard to indicate the attack protocols,attack density and attack patterns at one time.Advised to solve the problem with multi-class way.All attacks were classified into 24 classes according to attack protocols,attack density,attack patterns and HSMC-SVM,which is rapid and direct multi-class classifier,is employed to classify them.Shown as the experiments,the measure can implement training and testing processes rapidly and distinguish the attacks with a high true positive rate.The attack information about attack protocols,attack density,and attack patterns was provided by this way.At real network,the measure is feasible for its veracity and real time attribute.