×
Home Current Archive Editorial board
Instructions for papers
For Authors Aim & Scope Contact
Professional paper

FEDERATED LEARNING-BASED INTRUSION DETECTION FOR 6 G-ENABLED INTERNET OF THINGS IN SMART CITIES

By
Sanjay Kumar Orcid logo ,
Sanjay Kumar

Assistant Professor, Kalinga University , Naya Raipur, Chhattisgarh , India

Sapna Bawankar Orcid logo ,
Sapna Bawankar

Assistant Professor, Kalinga University , Naya Raipur, Chhattisgarh , India

Sindhusaranya Balraj Orcid logo
Sindhusaranya Balraj

Assistant Professor, Department of Computer Science and Engineering, Sona College of Technology , Salem, Tamil Nadu , India

Abstract

The high rate of Internet of Things (IoT) device proliferation in smart cities, along with the emergence of 6G technology, has tremendously augmented the network traffic and the issue of security. This paper proposes a Federated Learning-based Intrusion Detection System (FL-IDS) specifically designed for 6G-enabled IoT networks. The new system helps to solve the problem of scalability, privacy protection, and the possibility of detecting anomalies in time without any central storage of the data. FL allows training local models on edge devices and only the weights are transferred to a central model, which allows preserving sensitive information privacy. The methodology involves the use of intrusion detection using the local models of Random Forest, SVM, and KNN, and trained locally on the IoT devices. These models are subsequently federated by averaging to create a federated global model to be effective in detecting intrusion in large-scale IoT networks. The system identifies the anomalies of Denial of Service (DoS), spoofing, and data breach based on the network traffic patterns and device behavior variation. The system was evaluated using key performance metrics, namely, accuracy, precision, recall, and F1-score. These findings prove that the FL-IDS can attain an accuracy of 98, an increase of 12 % over the traditional intrusion detection systems. The system also decreases false positive rates by 20 %, as well as communication overhead by 35 %. The federated learning architecture enables scalable and efficient deployment, a large amount of data processing, and data privacy. Finally, the FL-based intrusion detection system provides a solution that is privacy-saving, scalable, and real-time to detect intrusion in 6G IoT networks in smart cities. The further study will concentrate on the optimization of the model updates and the system performance in dynamic urban settings, as well as real-time monitoring.

References

1.
Kalyanasundaram, K., Sivakolundu, V. P., & Mohan, J. (2025). Enhancing IoT Security in 6G Networks Using Federated Learning: A Decentralized Approach. In Secure Communication for the 6G and the Internet of Things Networks (pp. 107-134). CRC Press.
2.
Bamal S, Singh L. Detecting conjunctival hyperemia using an effective machine learning based method. Journal of Internet Services and Information Security. 2024;14(4):499–510.
3.
Pelekoudas‐Oikonomou F, Mirzaee PH, Hathal W, Mantas G, Rodriguez J, Cruickshank H, Sun Z. Federated Learning‐Based Intrusion Detection Systems for Massive IoT. Security and Privacy for 6G Massive IoT. 2025 May 19:101-28.
4.
Bahmaid S, Ghaleb SAM. Intrusion detection system using chaotic Walrus optimization-based convolutional echo state networks for IoT-assisted wireless sensor networks. Journal of Wireless Mobile Networks, Ubiquitous Computing, and Dependable Applications. 2024;15(3):236–252.
5.
Kim M, Oh I, Yim K, Sahlabadi M, Shukur Z. Security of 6G-enabled vehicle-to-everything communication in emerging federated learning and blockchain technologies. IEEE access. 2023 Dec 29;12:33972-4001.

Citation

This is an open access article distributed under the  Creative Commons Attribution Non-Commercial License (CC BY-NC) License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 

Article metrics

Google scholar: See link

The statements, opinions and data contained in the journal are solely those of the individual authors and contributors and not of the publisher and the editor(s). We stay neutral with regard to jurisdictional claims in published maps and institutional affiliations.