AI-Driven Intrusion Detection and Prevention Systems for Next Generation Enterprise Networks: A Comprehensive Review and Algorithmic Framework

Abstract

As enterprise network infrastructures rapidly evolve toward multi-cloud architectures, edge computing environments, and Internet-of-Things (IoT) ecosystems, traditional signature-based Intrusion Detection and Prevention Systems (IDPS) are increasingly rendered obsolete against modern, sophisticated cyber threats. Advanced Persistent Threats (APTs), zero-day exploits, and encrypted malicious payloads present unprecedented challenges to contemporary network security defenses. This review paper provides a rigorous, state-of-the-art analysis of Artificial Intelligence (AI) and Machine Learning (ML) techniques applied to next-generation enterprise IDPS. We systematically examine shallow machine learning, deep learning architectures including Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, Transformers, and Graph Neural Networks (GNN), as well as Federated Learning (FL) frameworks designed for privacy-preserving threat intelligence sharing. Furthermore, we analyze key performance metrics, data preprocessing techniques, dynamic feature selection methods, and automated mitigation mechanisms. Based on qualitative and quantitative synthesis, we highlight major research gaps in high-throughput line-rate processing, adversarial robustness, model explainability (XAI), and false positive reduction. Finally, we present an algorithmic architectural framework and outline prospective future directions to guide the design of resilient enterprise security architectures.

KEYWORDS: Intrusion Detection and Prevention Systems (IDPS), Artificial Intelligence, Deep Learning, Transformer Models, Federated Learning, Zero Day Threat Mitigation, Explainable AI (XAI), Enterprise Network Security.