Network Anomalies Detection Using Statistical Technique : A Chi- Square approach

Intrusion Detection System is used to detect suspicious activities is one form of defense. However, the sheer size of the network logs makes human log analysis intractable. Furthermore, traditional intrusion detection methods based on pattern matching techniques cannot cope with the need for faster speed to manually update those patterns. Anomaly detection is used as a part of the intrusion detection system, which in turn use certain data mining techniques. Data mining techniques can be applied to the network data to detect possible intrusions. The foremost step in application of data mining techniques is the selection of appropriate features from the data. This paper aims to build an Intrusion Detection System that can detect known and unknown intrusion automatically. Under a data mining framework, the IDS are trained with statistical algorithm, named Chi-Square statistics. This study shows the plan, implementation and the analyze of these threats by using a Chi-Square statistic technique, in order to prevent these attacks and to make a Network Intrusion detection system (NIDS). This proposed model is used to detect anomaly-based network to see how effective this statistical technique in detecting intrusions.