Hybrid intrusion detection system using particle swarm optimization and neural networks for external threat detection
Year: 2026
Type: article
Abstract: Intrusion is everywhere in the world, wherever there is networking. Intrusion Detection Systems (IDS) are of utmost importance for protecting data from breaches and intruders. Designing effective methods for detecting malicious intrusions as soon as they occur in any network becomes of utmost importance. The proposed hybrid methodology has been simulated with an enterprise secure network, trained with standard datasets like UNSW, KDD CUP, and NSDL, and evaluated using a variety of methods, including Man-in-the-Middle (MITM) and Denial of Service (DoS) assaults, sinkhole attacks, and probe attacks. The experimental setup has been framed with two distinct networks, one representing an enterprise virtual private network and another representing a secure blade-simulated virtual network. The experimental scenarios are framed with a hybrid approach that integrates PSO, Artificial Neural Networks (ANNs), K-Nearest Neighbor (K-NN), and decision trees (DT). Before that, the feature selection optimization method Particle Swarm Optimization (PSO) was employed. and then training with the dataset containing reduced features integrated into the various machine learning models, and their accuracy detection rates are compared. The proposed hybrid methodology of PSO with ANN outperformed all other existing IDS, generating the best outcome results in detecting and preventing cyber-attacks.
Institution: Priyadarshini College of Engineering
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7127054518