Distributed Intrusion Detection System For Edge of Things To Enhance Security

The increasing application of the Internet of Things (IoT) has raised significant security concerns, particularly at the Edge of Things (EoT). Traditional Intrusion Detection Systems (IDS) based on Centralized architectures have privacy issues and inefficiencies. This paper presents a Distributed Intrusion Detection System (DIDS) based on Federated Learning (FL) to enhance security in EoT environments. We use the UNSW-NB15 dataset for network intrusion detection to evaluate FL against Centralized Learning (CL) models. The experimental results show that FL outperforms CL with a test accuracy of 91.67% against 88.17%, confirming its potential towards secure, decentralized, and scalable intrusion detection in EoT networks.