Network Intrusion Detection Using Feature Selection Techniques: Bacterial Forage Optimization Algorithm

Network intrusion detection systems -NIDS are significant in potentially analyzing cyber-security and accurately categorizing attacks in current networks.Given the present circumstances, selecting an appropriate combination of anomaly detection features holds greater significance within NIDS.Diverse optimization algorithms offer a better solution: correctly choosing a near-optimal variety of features to achieve an improved NIDS.The Bacterial Forage Optimization (BFO) algorithm is an intelligent swarm algorithm commonly used for optimization problems.This paper proposes NIDS with optimal feature selection using Bacterial Forage Optimization techniques to detect the anomaly in the IDS (BFOFSIDS).Besides, it established Machine Learning-based Network Intrusion Detection and offered better performance for detecting User Remote (U2R Remote-to-Local (R2L), Probe and Denialof-service (DoS) attacks.The proposed model aims to identify anomaly detection features for an intrusion detection system, thus achieving a promising performance.The proposed method expands the BFO-based optimal features selection techniques, enhancing the accuracy among two high-dimensional intrusion detection datasets, such as NSL-KDD and UNSW-NB15.Experimental results demonstrate that the proposed model achieves an accuracy of 91.8% on the NSL-KDD dataset and 91% on the UNSW-NB15 dataset.Additionally, the BFOFSIDS model significantly reduces the average processing time, with values recorded at 13,756 ms for NSL-KDD and 13,748 ms for UNSW-NB15, outperforming state-of-the-art methods.The precision and F-score for BFOFSIDS are also notably higher, indicating its effectiveness in improving detection accuracy while minimizing computational complexity.