Design of Intrusion Detection System for Wireless ADHOC Network in the Detection of DOS Attack using Oneclass SVM with Wrapper Approach Feature Selection Comparing with Information Gain Algorithm

• Published in: 2023 International Conference on Advances in Computing, Communication and Applied Informatics (ACCAI) pp. 1 - 6
• Main Authors: Kumar, M. Dinesh, Nagalakshmi, T J
• Format: Conference Proceeding
• Language: English
• Published: IEEE 25.05.2023
• Subjects: Denial of Service (DOS) Attack; Feature Selection; Innovative oneclass SVM (Support Vector Machine) with wrapper approach Feature Selection IDS; Intrusion Detection System (IDS); Machine Learning Models; Performance Metrics
• DOI: 10.1109/ACCAI58221.2023.10200296

The goal of this work is to design two intrusion detection systems to detect DOS attack in a wireless adhoc network by using Innovative oneclass SVM (Support Vector Machine) with wrapper approach Feature Selection IDS (Group 1) and Information Gain Algorithm (Group 2). Materials and Methods: To design an IDS model CICIDS 2017 dataset was taken. By using above mentioned algorithms the IDSs are modelled and are analysed with the SPSS tool with 19 samples in each group. Results: The IDS using Innovative oneclass SVM (Support Vector Machine) with wrapper approach Feature Selection IDS having 92% accuracy and 82% detection rate and the IDS with Information Gain Algorithm having 82% accuracy and 80% detection rate. The significance p<0.05 (Accuracy - 0.00; Detection rate - 0.00) indicates the good performance of IDS. Conclusion : This result clearly shows that the Intrusion Detection System designed using Innovative oneclass SVM with wrapper approach Feature Selection IDS significantly performs better than the IDS using Information Gain Algorithm.