An enhanced whale optimizer based feature selection technique with effective ensemble classifier for network intrusion detection system

The increasing frequency of network attacks poses a significant challenge in maintaining network security. In response to security vulnerabilities, the employment of Intrusion Detection Systems (IDS) has become crucial. An IDS is a software application, which monitors network traffic to monitor and identify various malicious actions in the network. Network Intrusion Detection Systems (NIDS) specifically examine network traffic to identify and monitor the various malicious activities identified by the network nodes. However, many existing IDS solutions encounter difficulties in achieving optimal feature selection, aiming for high classification accuracy while minimizing false alarm rates. In the proposed system, the Linear Discriminant Analysis (LDA) technique is employed to reduce the dataset dimensionality. Further, an Enhanced Whale Optimization Algorithm (EWOA) is utilized for efficient feature selection for global optimization tasks. To perform classification, an ensemble classifier is proposed which employs Random Forest (RF) and XGBoost. The main objective of the proposed approach is to improve classification accuracy while minimizing the false alarm rate by combining the benefits of RF and XGBoost. The proposed work is implemented by using the python jupyter framework. The experimental results show that the proposed system has an overall classification accuracy of 96.08%, a false alarm rate of 0.01%, and an overall detection rate of 94.03%, by using the NSL-KDD dataset when it is compared with their counterparts.