State-of-the-Art Research in Intrusion Detection Systems (IDS) for Network Attack Detection: A Review

Abstract

Intrusion Detection Systems (IDS) used in monitoring and analyzing network traffic have become essential measures in the detection of anomalies in network and in the mitigation of cyber threats. This review provides insights into the recent IDS developments for network attack detection, identifies research gaps, and discusses potential advancements to create adaptive and robust cybersecurity frameworks for combating evolving cyber threats effectively. It categorizes IDS into signature-based, anomaly-based, hybrid, network-based, and host-based systems. It explores the enhancing effects of the integration of deep learning (DL) and machine learning (ML) on IDS efficiency. It shows that Generative Adversarial Networks (GANs), Transformer-based models, and Federated Learning (FL) are demonstrating significant improvements. Nevertheless, adversarial attacks, computational overhead and high false positives persist as the major challenges in current IDS research. Therefore, we suggest that future studies should focus on improving real-time threat detection by maximizing adversarial resilience, Explainable AI (XAI), and quantum computing to create adaptive and robust cybersecurity frameworks.