OptiSecure: Hybrid SVM and ACO-based Intrusion Detection with Feature Optimization

Abstract

In the era of modern networking, detecting cyber threats with high accuracy and minimal computational overhead has become increasingly vital. This paper presents an enhanced anomaly-based intrusion detection system (IDS) using a hybrid model combining Ant Colony Optimization (ACO) and Support Vector Machine (SVM). The CICIDS2017 dataset is used to evaluate the proposed approach, representing real-world traffic with diverse attack patterns including DoS, DDoS, Brute Force, Botnet, and Infiltration attacks. Experimental results show that the ACO-SVM model achieves a detection rate of 90.56%, false alarm rate of 9.44%, and time complexity of 0.32 seconds, outperforming ACO-ANN, ACO-NB, and PSO-SVM models. The results confirm that ACO combined with SVM provides efficient, scalable, and accurate intrusion detection performance.