Intelligent Zero Trust Cybersecurity Framework for Enterprise Network Protection Using Autoencoder CNN-GRU Hybrid Learning

Abstract

As cyber threats grow more sophisticated and enterprise network architectures are becoming more complex, traditional perimeter-based cybersecurity mechanisms have proven inadequate. In recent years, Zero Trust Architecture (ZTA) has become a successful security paradigm that continuously checks the validity of all requests that are made. Recently, Zero Trust Architecture (ZTA) has become an effective paradigm that continuously checks the validity of all requests made, rather than taking trust for granted. Most current Zero Trust architectures, though, are based on complex architectures that are very compute-intensive or are mainly used in the authentication and access control domain without having the intelligent intrusion detection capabilities. To tackle these problems, this paper presents an Intelligent Zero Trust Cybersecurity Framework for Enterprise Network Protection Based on AutoEncoder CNN–GRU Hybrid Learning. The proposed system embeds feature compression using an AutoEncoder, learning of spatial features using a Convolutional Neural Network (CNN), and learning of temporal dependencies using a Gated Recurrent Unit (GRU) in a single deep learning system. In addition, a Zero Trust Decision Engine is added to transform the forecasted classification probabilities into dynamic risk scores for adaptive Allow, Monitor and Block access decisions. The framework was tested with the benchmark CSE-CIC-IDS2018 dataset in a standard deep learning environment with Python. Experimental results have shown that the proposed framework improved the accuracy to 99.12%, precision to 98.96%, recall to 98.81%, F1-score to 98.88%, and ROC-AUC to 99.47% with low inference latency and moderate computational complexity for real-world enterprise applications. The proposed framework is also found to give a good balance between predictive performance, computational efficiency, and adaptive Zero Trust decision-making through a comparative analysis with recent state-of-the-art approaches. The results highlight the promise of hybrid deep learning and Zero Trust approach in achieving scalable, intelligent, and reliable enterprise cybersecurity solutions.