Fed-CNN-IDS: A Privacy-Preserving Federated Learning-Based CNN Intrusion Detection System for IoMT

Abstract

The Internet of Things (IoT) has witnessed rapid growth in recent years, transforming numerous application domains. Among these, the Internet of Medical Things (IoMT) plays a vital role by enabling the collection and processing of vast amounts of sensitive personal and medical data. However, the security and privacy of such data remain critical concerns, especially in the face of increasing cyber threats. Traditional Intrusion Detection Systems (IDSs) based on centralized architectures are often vulnerable to various attacks and suffer from limitations in preserving data privacy. To address these challenges, this study proposes a Federated Convolutional Neural Network-based Intrusion Detection System (Fed-CNN-IDS) tailored for the IoMT environment. By leveraging Federated Learning (FL), the proposed model was trained locally on distributed devices without transferring sensitive data to a central server, thereby enhancing privacy. The model was trained and evaluated using the CICIoMT2024 dataset, and extensive experiments were conducted with varying numbers of clients to assess its performance in distributed settings.