Ensemble of Ada Booster with SVM Classifier for Anomaly Intrusion Detection in Wireless Ad Hoc Network

Objectives: An ensemble of Ada Booster with SVM (EAB-SVM) classifier technique is proposed for detecting the network intrusions and monitoring the activities of the node as well as classifying it as either normal or anomalous. Methods/ Statistical Analysis: The optimal feature selection method is applied in EAB-SVM to select the relevant features to classify and detect the intrusion in wireless ad-hoc network. After that, an ensemble of Ada booster with SVM (EAB-SVM) classifier is used for classifying the intrusion through updating the weight of samples. Finally, the objective function of the EAB-SVM classifier is used to distinguish the anomalous and normal node behavior accurately. This in turn improves the anomaly intrusion detection accuracy. Findings: The EAB-SVM technique comprises of two model namely optimal feature selection model and intrusion detection model. The EAB-SVM technique employed Optimal Feature Selection to choose the features and to reduce the data space dimension. This in turn assists to increase the packet delivery ratio. After selecting the optimal features, the intrusion detection and classification is performed using AdaBoost with SVM classifier for intrusion detection. The AdaBoost with SVM classifier discover the node behaviors as normal or anomalous and thereby achieves reliable packet delivery in wireless ad-hoc network. The performance of EAB-SVM technique is evaluated in terms of packet delivery ratio, classification time, false positive rate and anomaly intrusion detection accuracy and compared with existing two methods. Application/Improvements: The simulation results shows that the EAB-SVM technique achieved the better performance in terms of packet delivery ratio, classification time, false positive rate and anomaly intrusion detection accuracy compared to state-of-the-art methods.