FedAnom: Strengthening Cloud Network Security with Federated Anomaly Detection
Manadadi Sriya Reddy, Kalyan Chatterjee, Muntha Raju, Samala Suraj Kumar, Tummala Abhinav Vardhan Reddy, Machakanti Navya Thara · 2024
Anomaly detection forms the cornerstone of ensuring the security of cloud networks. In this paper, we present an innovative approach aimed at enhancing anomaly detection in cloud networks by harnessing federated learning coupled with differential privacy. Building upon the foundation of autoencoder-based anomaly detection, our method integrates ensemble methods to amplify accuracy. Furthermore, our model incorporates adversarial robustness techniques to fortify resilience against potential attacks. Dynamic thresholding is employed to ensure adaptability to the evolving threat landscape. Through extensive evaluations conducted on diverse datasets, our approach showcases scalability, efficiency, and real-world effectiveness. Specifically, our model achieves a notable enhancement in accuracy and a substantial reduction in false positives compared to existing methods, signifying significant strides in bolstering cloud network security. By offering practical solutions for strengthening anomaly detection capabilities within cloud environments, our research contributes to the augmentation of cybersecurity infrastructure. This novel approach, empowered by federated learning with differential privacy, holds promise for effectively navigating the dynamic threat landscape and safeguarding cloud-based systems against malicious activities.