Emerging Federated Learning Frameworks Enhance Intrusion Detection Against DDoS Attacks

Recent advancements in intrusion detection systems leverage federated learning frameworks to combat distributed denial-of-service (DDoS) attacks. Two frameworks, SecureTrust-FL and FL-TWIN, were proposed to enhance privacy and security in distributed environments. SecureTrust-FL integrates trust management and differential privacy for effective intrusion detection across heterogeneous datasets, achieving an overall accuracy of 92.91%. Meanwhile, FL-TWIN pairs clients with Digital Twins and employs a four-stage poisoning defense pipeline, achieving peak test accuracies of 99.98% against various attack types. Both frameworks utilize blockchain technology for accountability and transparency in the learning process. These innovations aim to address the increasing complexity of networked environments and the challenges posed by DDoS attacks. The frameworks demonstrate significant improvements in detection performance while preserving data privacy. The research highlights the importance of adaptive learning and robust defense mechanisms in the face of evolving cyber threats.

Key Points: • SecureTrust-FL achieves 92.91% accuracy in privacy-preserving intrusion detection. • FL-TWIN pairs clients with Digital Twins, achieving 99.98% accuracy against DDoS attacks. • Both frameworks utilize blockchain for enhanced trust and accountability in collaborative learning.