Авторы:
Издательство:
Дата издания
23.07.2025На каком языке издана
ISSN
22242708Область исследования
Тип исследования
Аннотация (краткое описание)
This paper presents a large-scale empirical study aimed at identifying the optimal local deep learning model and data volume for deploying intrusion detection systems (IDS) on resource-constrained IoT devices using federated learning (FL). While previous studies on FL-based IDS for IoT have primarily focused on maximizing accuracy, they often overlook the computational limitations of IoT hardware and the feasibility of local model deployment. In this work, three deep learning architectures—a deep neural network (DNN), a convolutional neural network (CNN), and a hybrid CNN+BiLSTM—are trained using the CICIoT2023 dataset within a federated learning environment simulating up to 150 IoT devices. The study evaluates how detection accuracy, convergence speed, and inference costs (latency and model size) vary across different local data scales and model complexities. Results demonstrate that CNN achieves the best trade-off between detection performance and computational efficiency, reaching ~98% accuracy with low latency and a compact model footprint. The more complex CNN+BiLSTM architecture yields slightly higher accuracy (~99%) at a significantly greater computational cost. Deployment tests on Raspberry Pi 5 devices confirm that all three models can be effectively implemented on real-world IoT edge hardware. These findings offer practical guidance for researchers and practitioners in selecting scalable and lightweight IDS models suitable for real-world federated IoT deployments, supporting secure and efficient anomaly detection in urban IoT networks. © 2025 by the authors.DOI
10.3390/jsan14040078Тип публикаций
Практическое значение
This study has important practical implications for the development of secure and efficient Internet of Things (IoT) networks. By systematically evaluating different deep learning architectures within a federated learning (FL) framework, the research provides valuable guidance for selecting intrusion detection system (IDS) models that balance detection accuracy with the computational constraints of IoT devices. The findings demonstrate that convolutional neural networks (CNNs) offer an optimal trade-off between security performance and resource efficiency, making them particularly suitable for deployment on edge devices with limited processing power, memory, and energy resources. The study also shows that high intrusion detection accuracy can be achieved without relying on excessively complex models, thereby reducing inference latency and storage requirements. The practical applications of this research include: Smart city infrastructures, where large numbers of interconnected IoT devices require scalable and secure anomaly detection mechanisms. Industrial IoT (IIoT) systems, for protecting manufacturing and automation networks from cyberattacks. Healthcare IoT environments, where timely and reliable detection of malicious activities is critical for patient safety and data protection. Smart homes and consumer IoT devices, where lightweight security solutions are necessary due to hardware limitations. Edge and distributed computing systems, where federated learning enables collaborative model training without exposing sensitive local data. Furthermore, the successful deployment and evaluation of the proposed models on Raspberry Pi 5 devices confirm their feasibility for real-world edge computing environments. The results help researchers, system designers, and cybersecurity practitioners make informed decisions regarding model selection, data allocation, and deployment strategies for federated IoT intrusion detection systems. Ultimately, the study contributes to the development of scalable, privacy-preserving, and resource-efficient cybersecurity solutions for next-generation IoT ecosystems.Квартиль
Источник квартиля
SCImago (Scopus)Impact Factor
11.3Количество цитирований
31Данный документ располагается в коллекциях
Международные публикации ученых SU
Scopus