SECURING COMPUTER NETWORK by INCREASING the PERFORMANCE of INTRUSION DETECTION SYSTEM

Rapid development in the Internet and network technology has caused a considerable rise in the number of malicious intrusions. Hence, the security of the computer network is crucial for data confidentiality. Security measures are used as the first line of defense to protect network resources. However, these mechanisms are not enough because the growth of the information technology changes the ways of how the hacker conducts attacks. Therefore, an intrusion detection system is essential to detect the presence of attack among the network traffics. One of the principal problems is its high false alarm rate. Different feature selection has been used in developing a system. However, further research is needed to improve its performance. In this paper, we develop an anomaly-based intrusion detection system based on the ensemble voting technique and the principle component analysis to boost attack detection rate and at the same time minimize the high false alarm rate. Here, the performance of the proposed system is evaluated using KDD99 dataset, and the experimental results demonstrate a system accuracy detection of 99.960% while the false positive rate is only 0.039%, which is a better achievement. Therefore, an ensemble technique and a better feature selection yield to high-performance intrusion detection system model.