Dynamic Trust-Weighted Federated Aggregation for Robust DDoS Detection in Cloud-Edge Networks

Cloud-edge networks are seriously threatened by Distributed Denial-of-Service (DDoS) attacks. Intrusion Detection Systems (IDS) based on Federated Learning (FL) allow privacy-preserving collaborative learning, although they are still susceptible to poisoning attempts and malicious model updates. A Dynamic Trust-Weighted Federated Aggregation (DTWFA) architecture for reliable DDoS detection is presented in this paper. Based on detection accuracy, update consistency, and past behavior, the framework gives participating edge nodes trust scores. These scores are then incorporated into the federated aggregation process. DTWFA increases resilience against adversarial attacks and improves detection performance by lessening the impact of faulty nodes. In dynamic cloud-edge scenarios, experimental results show better robustness, reduced false-positive rates, and superior accuracy when compared to traditional FedAvg-based FL techniques.