Indian Journal of Science and Technology

Abstract

Today’s network world is facing lot of challenging task; Intrusion Detection is the method adopted to identify the unauthorised activities in the network. The type of malicious activities grows on increasing. There is very much essential to develop a new Intrusion Detection System, which can detect the malicious activities. The main purpose of our newly hybrid system is developed to identify both known and unknown attack. The proposed system is tested with the benchmark KDD ’99 intrusion data set. The proposed work also focuses on the detection rate and false alarm rate. The new system is developed with an optimized algorithm for feature to produce the reduced set of features. The attack detection rate is comparatively good; can be achieved by using layered approach with enhanced fuzzy multi-objective particle swarm optimization which does the feature selection effectively. The fuzzy based support vector machine algorithm is effectively applicable to detect anomaly attack. The newly described system is must more efficient in detection U2R attack, when compared with the existing methods takes long time for training to detect the unknown attack. Also our system is working with very less time for detection. The proposed system with the advantage of detection rate up to 99.1% and the false alarm rate is very much reduced.
Keywords: Attack, Computation, Fuzzy, Goodfit, Intrusion Detection, Objective Function, Optimization, Support Vector Machine