Improving Botnet Attack Identification in Iot Systems Through Gwo and Ofo Methods: An Innovative Strategy

Abstract

To protect computer networks from malicious threats it is necessary to have intelligent and adaptive Intrusion Detection Systems. To address this challenge, this article presents a new enhancement to the traditional Grey wolf optimization (GWO) algorithm for improving defensive capability.We propose Optical Flow Optimization (OFO) to intelligently initialize the population, which lays a solid foundation for global search and achieves faster and more effective threat detection. This optimized initialization is integrated with a Support Vector Machine (SVM) classifier in a supervised machine learning framework to form a powerful GWO-OFO-SVM hybrid. The proposed approach significantly improves the convergence speed, accuracy of the solution and resilience against local optima which are the main drawbacks of traditional GWO. Extensive performance validation evaluations are carried out on Mirai botnet attacks and compared with other state-of-the-art metaheuristic algorithms such as Salp Swarm-Ant Lion Hybrid Optimization (SSA-ALO), GWO and One-Class SVM (GWO-OCSVM), and Bi-directional LSTM with Recurrent Neural Networks for Botnet detection (BLSTM-RNN).Experimental results on the N-BaIoT dataset show the superiority of GWO-OFO-SVM with outstanding metrics, 97.30% accuracy, 97.09% recall, 97.08% precision, and F1-score of 97.45%. The results show the model’s ability to build reliable, high-performing intrusion detection systems in practical network settings.