NETWORK INTRUSION DETECTION SYSTEM USING DATA MINING TECHNIQUES

Abstract

In present scenario intrusion in the network is very critical concern for integrity, confidentiality, liability and many other security related term of the network. Due to this need of better network intrusion detection system is very important and critical need. Every Intrusion Detection System has some special features and these are totally depending on methodology used for development and in other word, training phase. There are two important techniques used in development of Intrusion Detection System, clustering and classification. Accuracy of the system is highly depended on classification although need of clustering cannot be change or replaceable. So for achieving more accuracy, high detection rate, high performance and low false alarm rate, this paper is using combination of clustering and classification technique. In this paper better approach of combination of techniques used that overcome short-fall of previous approach used for implementation for Intrusion detection system. This paper is using Fuzzy C-Mean algorithm as a clustering technique and Support vector machine as classification technique. KDD cup dataset is used for training and testing purpose in order to evaluate performance, accuracy, detection rate, false alarm rate and other important parameter. This paper presents a detail comparative analysis between combination of Fuzzy C-Mean and Support vector machine and combination of K Means and Support vector machine. Experiments and analyses show that the new approach is better in increasing the detection rate and accuracy as well as in decreasing the false positive rate. In previous paper’s related to intrusion detection system, all are based on detection of abnormal data’s. But these systems are highly depend on type of attacks, so it is better approach to track normal data’s because behaviour of normal data’s do not change in any time duration. That gives some advantage to the system on the basis of reliability, performance and accuracy.