Intelligent Anomaly Detection for Secure Data Transmission in Cloud Computing Systems over 6G Networks

The emergence of sixth-generation (6G) networks facilitates robust capabilities such as ultra-low latency and massive device connectivity, which simultaneously raise critical challenges in securing cloud-based data transmission. This study proposes a novel anomaly detection framework that integrates Autoencoders (AEs), Convolutional Neural Networks (CNNs), and Federated Learning (FL) to deliver real-time, privacy-preserving intrusion detection for 6G-enabled cloud computing environments. The framework is evaluated using four benchmark datasets, including NSL-KDD, UNSW-NB15, CIC-IDS2017, and CIC-DDoS2019. Across all datasets, the model achieved an average accuracy of 99.85%, precision of 99.76%, recall of 99.82%, and F1-score of 99.79%, while maintaining a False Alarm Rate (FAR) as low as 0.0011. The model also demonstrated high efficiency, operating with inference latency below 350 ms, making it highly suitable for the stringent requirements of 6G infrastructure. Enhanced with explainability tools, the system ensures transparent decision-making, offering an interpretable solution towards next-generation cybersecurity threats.