Cloud-Based Digital Twin for Cybersecurity Threat Prediction
Venkata Ashok Kumar Boyina, Thiyagarajan Mani Chettier, Dhaval Patolia, Neha Gupta · 2025
Cyber threats are evolving into the ever-skyscraper-oriented monster with increasing scope and complexity and demand for contemporary security models that focus on live threat detection and response. In this study, a Cloud-Based Digital Twin (CBDT) approach is proposed for predicting and mitigating cybersecurity threats by employing machine learning in the digital twin domain to bolster cyber resilience. The framework proposed becomes a virtualized cloud environment replicating the IT infrastructure of the organization and allowing simulation, detection, and prediction of cyber threats in real time. Through the integration of machine learning models like Random Forest, Convolutional neural networks (CNN), and Long Short-Term Memory (LSTM), the CBDT framework enhances the processes of anomaly detection, threat classification, and incident response automation. LSTM has been shown to offer a best-in-class detection rate (96.1%) of a wide variety of cyber threats and consistently minimizes time to response vs. traditional intrusion detection systems. Furthermore, deployments in the cloud open new capabilities around scale, connecting vast threat intelligence data, and rapid analysis. Conclusively, the investigation highlights the promise of digital twins in cybersecurity, providing a proactive and adaptable defense strategy against the ever-changing landscape of cyber vulnerabilities.