Hybrid Federated Ensemble Learning Approach for Re-al-Time Distributed DDoS Detection in IIoT Edge Compu-ting Environment

Development rapid from the Industrial Internet of Things ( IIoT ) and edge computing have revolutionize modern industry through distributed data processing with latency low . However , progress this also enlarges risk security cyber , in particular Distributed Denial of Service (DDoS) attacks can to disable operation industry that is critical . System Detection Conventional Intrusion (IDS) own limitations in matter scalability , data privacy , and capabilities generalization to environment Heterogeneous IIoT . For answer challenge said , research This propose A framework Hybrid Federated–Ensemble Learning (FL–EL) work to improve efficiency detection real -time DDoS attacks on networks IIoT edge -based . This model utilizing the Edge -IIoTset dataset which reflects pattern Then cross real in system industry . Federated learning is used For train the model collaborative across multiple edge nodes without need move data to center , so that guard data privacy . Each node performs training local using the basic model such as Random Forest (RF), XGBoost , and Support Vector Machine (SVM). Then , the central server do aggregation use ensemble techniques such as soft voting and stacking. The preprocessing process includes SMOTE technique and Z-score normalization for handle imbalance class and improve performance .Evaluation results show that This FL–EL hybrid approach capable reach performance high (F1-score > 99.5%) and significantly significant reduce level error positive as well as burden communication , compared with approach centralized . Framework this also shows ability detection fast with latency low , making it suitable For implementation in the system IIoT that requires resilience time real . Development advanced will covers Explainable AI integration for model interpretation and blockchain for secure and transparent logging .