Abstract: The growing use of computer networks, security has become a main challenging task for computer users. It is basically suffering from Denial of Service (DoS) attacks. DoS attack is an attack that sending large number of network traffic towards the Victim server of the organization that bring down the performance of networks. Existing techniques like firewall, access control and encryption mechanism is vulnerable to provide sufficient protection against Virus, warms and DoS attacks,etc.,. So we need new mechanism to monitoring computer systems and network traffic to identify any deviation of the original user behavior. This paper proposes a new hybrid based IDS model for DoS attacks and evaluate its performance based on Particle Swarm Optimization combine with Support Vector machine (PSO-SVM) to increase detection accuracy and reducing false alarms. Here, feature selection is one of the important processes to increase the classification accuracy of this model. So, Particle Swarm Optimization (PSO) is used for selecting necessary feature from the PMU 2015 dataset that will improve the performance of the proposed model. The proposed work was implemented in Mat lab7. 2. The result shows that the proposed hybrid IDS has high detection accuracy (99. 25%) and (0. 75%) of false alarms.
Keywords: Intrusion Detection System (IDS), Network Security, Denial of Service (DoS), PSO-SVM.