An IDS in Cloud Environment by using Feature Selection based Hybrid Multilevel Classifier

Cloud Computing has made resources and services accessible from any place of the world at any time. But with increasing popularity of Cloud Computing, it has also become vulnerable to intrusions. To defend the Cloud from attacks, Intrusion Detection Systems (IDSs) are developed. IDSs are deployed in the system or network to identify intrusions and report about them to the administrative system. Here the authors have proposed a novel IDS to provide security in the Cloud. For Feature Selection (FS), a Mutual Information-Information Gain based modified Ant Colony Optimization (MI-IG based modified ACO) method has been used which resorts to a feature scoring method. After FS, a hybrid multilevel Logistic Regression-Neural Network (LR-NN) classifier has been used for classification purpose where LR executes anomaly detection through binary classification and NN performs misuse detection. NSL_KDD dataset has been selected for conducting the experiments. From the experiments, it is observed that the proposed model has produced results with a high rate of accuracy. The results prove the excellence of the constructed IDS in Cloud Environment.