Zero-Trust Security in Intrusion Detection Networks: An AI-Powered Threat Detection in Cloud Environment

Abstract

Cloud computing serves as a critical technology in modern digital systems because it provides organizations with benefits that equally come with corresponding problems. AI-powered cloud security now functions as a vital strategical tool which tackles rising complex and advanced cyber threats in the existing cloud computing era. This study investigates how to achieve zero trust security in intrusion detection networks using AI and how successful it is at resolving security issues in cloud networks. For the protection of IoT/IIoT networks, the zero-trust approach may function better. Access to network resources requires authorization and verification before any connection can be made because all users and devices are considered untrustworthy by default. This paper presents a zero-trust machine learning intrusion detection system (IDS) for protecting IIoT and IoT networks. The research proposes a Zero-Trust security model based on XGBoost for detecting attacks within the Edge-IIoTset dataset. Model performance enhancement required two steps: Min-Max scaling for normalization and SMOTE to balance classes during the data preprocessing process. XGBoost classifier operates on split training and testing data to detect threats that include Normal, DDoS, Enumeration, and Malware. To evaluate the performance of proposed XGBoost model according to accuracy, precision, recall, f1score, ROC, and confusion matrix. The proposed model surpasses traditional models by attaining 94.55% accuracy while K-Nearest Neighbors achieves 79.18% and AdaBoost reaches 86.29%, and Recurrent Neural Networks achieves 91% accuracy. The model shows reliable performance according to precision evaluation results of 95.46% combined with recall outcome of 98.38% and F1-score of 94.22%. The outcomes of comparative study and evaluation demonstrate the accuracy of risk detection abilities for Zero-Trust implementations which enhances security measures in contemporary digital systems.