An Efficient Hybrid Multilevel Intrusion Detection System in Cloud Environment

Abstract

Cloud Computing offers latest computing paradigm where application, data and IT services are provided online over the Internet. One of the significant concerns in Cloud Computing is security. Since data is exposed to many users, security and privacy have become the key issues of Cloud Computing. Intrusion Detection System (IDS) plays an important role to identify intrusions by monitoring the activity of the system and alert the user about malicious behaviours and detect attacks. To detect those attacks, several classification methods have been used till now. This paper deals with Intrusion Detection System by the method of classification. In this paper, KNN is applied as binary classifier for anomaly detection. Neural Network is applied for detecting abnormal classes after KNN classification. Before classification, feature selection has been used to select relevant features. For our experimental analysis, we have used NSL-KDD dataset where all samples of "KDDTrain+" used as training dataset and "KDDTest+" samples are used as testing dataset. We use Rough Set Theory and Information Gain to select relevant features. Experimental results show that, we get better accuracy with our proposed hybrid KNN_NN classifier model for Intrusion Detection.