Anomaly-based intrusion detection system using Relational Detector Tree (RTD)

Network security is one of the most challenging issues in network communication world. As network communications grow, vulnerabilities and penetrating attacks are predicted to be as prominent factors. So in order to thwart these attacks an intelligent and powerful intrusion detection system in required. In this paper, a multilayer intrusion detection system is proposed. In first layer, four types of detector are created using genetic algorithm which are used in the second layer to detect some anomalies or abnormal traffic data using Negative Selection Algorithm (NSA) and finally in last layer, the detected anomaly data are classified into four types of attack: DoS, Probe, R2l and U2r. The results show better performance in detecting and classifying new or unseen abnormal data. All experiments are done using KDDCUP99 dataset

Keywords:

detector , genetic algorithm , Negative Selection Algorithm (NSA) , DoS , Probe , R2l , U2r , KDDCUP99