FedEdgeIDS: A Federated Intrusion Detection Framework for Edge-to-Cloud Networks

DOI

10.2991/978-94-6239-789-7_38How to use a DOI?

Keywords

Intrusion Detection Systems; Federated Learning; Edge Computing; Deep Learning; False Positive Rate; IoT Security; MITM Detection; Real-Time Anomaly Detection; Privacy-Preserving Machine Learning

Abstract

Increased proliferation of Internet of Things (IoT) devices and the emergence of distributed edge-to-cloud systems have exposed the critical weaknesses of conventional centralized Intrusion Detection Systems (IDSs), which are ill-adapted to the dynamic, heterogeneous, and privacy-sensitive needs of modern networks. FedEdgeIDS, as presented in this paper, represents a federated system of intrusion detectors designed to read and write at edge-to-cloud setups. The framework makes use of light-weighted CNN-LSTM models, which are capable of operating on the edge nodes to enable real-time, low-latency threat detection. But federated learning among devices located in different geographic locations does not involve transferring the data to the centralized server, hence ensuring the privacy of users. Another enhancement to the framework includes a Secure Communication Module (SCM), which protects against MITM attacks and packet injection. To address the long-standing issue of high false positives in existing IDS systems, we introduce a multi-stage False Positive Reduction Module (FPRM), which combines information-theoretic feature selection, adaptive threshold calibration, and ensemble verification to reduce false positive alerts without compromising the detection system’s performance. Three sets of benchmark data, including NSL-KDD, CICIDS2017, and Edge-IIoT, as well as non-IID federated data conditions, are used to test the framework. A large number of key performance indicators is also achieved with FedEdgeIDS, including a low false positive rate (FPR) of 0.42% and an average edge detection latency of 3.7 ms, compared to seven baseline approaches. These results indicate that FedEdgeIDS is a potential and privacy-sensitive intrusion detection system that can be scaled in the future for deployment in next-generation infrastructure networks.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Tanu Gupta
AU  - Rimmy Chhabra
AU  - Aditi Singh
AU  - Riyanshu Saini
AU  - Arpit Pant
AU  - Aastha Verma
PY  - 2026
DA  - 2026/10/08
TI  - FedEdgeIDS: A Federated Intrusion Detection Framework for Edge-to-Cloud Networks
BT  - Proceedings of theInternational Conference on Innovation and Practices in Artificial Intelligence (ICIPAI 2026)
PB  - Atlantis Press
SP  - 575
EP  - 596
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-789-7_38
DO  - 10.2991/978-94-6239-789-7_38
ID  - Gupta2026
ER  -