Federated Hybrid Deep Learning Framework for IoT Intrusion Detection Using N-BaIoT, CICIoT2023, and BoT-IoT Datasets
Abstract
Abstract Internet of Things (IoT) environments expose many heterogeneous and resource-constrained devices to botnet, distributed denial-of-service, denial-of-service, reconnaissance, spoofing, malware, and service-scanning attacks. Centralized intrusion detection can achieve strong accuracy when all traffic is pooled, but it creates privacy, bandwidth, and deployment barriers for edge networks. This manuscript presents an end-to-end federated hybrid deep learning framework for IoT intrusion detection using N-BaIoT, CICIoT2023, and BoT-IoT data sources. The implemented workflow performs Phase-1 cleaning, encoding, train/validation/test construction, feature selection, class-imbalance handling, VAE-ready export, and IID/non-IID client preparation, followed by Phase-2 individual dataset training. The Phase-2 architecture uses CNN, LSTM, BiLSTM, attention, Tiny Transformer, Random Forest, XGBoost, and hybrid deep-feature XGBoost models. The completed evaluation shows strong detection performance on N-BaIoT and BoT-IoT and highlights CICIoT2023 as the most challenging multiclass setting because it contains 35 attack/traffic classes. These results provide a verified basis for the next federated aggregation and anomaly extension stages.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.