Enhancing Cyber Security in IoT Systems using FL-based IDS with Differential Privacy

Abstract

Nowadays, IoT networks and devices exist in our everyday life, capturing and carrying unlimited data. However, increasing penetration of connected systems and devices implies rising threats for cybersecurity with IoT systems suffering from network attacks. Artificial Intelligence (AI) and Machine Learning take advantage of huge volumes of IoT network logs to enhance their cybersecurity in IoT. However, these data are often desired to remain private. Federated Learning (FL) provides a potential solution which enables collaborative training of attack detection model among a set of federated nodes, while preserving privacy as data remain local and are never disclosed or processed on central servers. While FL is resilient and resolves, up to a point, data governance and ownership issues, it does not guarantee security and privacy by design. Adversaries could interfere with the communication process, expose network vulnerabilities, and manipulate the training process, thus affecting the performance of the trained model. In this paper, we present a federated learning model which can successfully detect network attacks in IoT systems. Moreover, we evaluate its performance under various settings of differential privacy as a privacy preserving technique and configurations of the participating nodes. We prove that the proposed model protects the privacy without actually compromising performance. Our model realizes a limited performance impact of only ~ 7% less testing accuracy compared to the baseline while simultaneously guaranteeing security and applicability.

A Collaborative Software Defined Network-Based Smart Grid Intrusion Detection System

“…Moreover, Houda et al [34], proposed a federated learning approach combined with blockchain to establish a decentralised IDS for SDN, and Cui et al [35] presented a federated learning approach with CNN's for SDN-IDS creation. In addition in [36] authors introduce a privacypreserving IDS for the detection of DoS attacks in IoT networks, based on FL, utilising a FFNN with the FedAVG strategy. Furthermore, Mothukuri et al [37] proposed an IoT based FL-IDS using GRUs.…”

A Survey of Using Machine Learning in IoT Security and the Challenges Faced by Researchers

“…[176], [230] [77] UNSW-NB15 Intrusion Detection 175,341 training 82,332 testing 49 [125], [132], [143], [154], [157], [161], [162], [171], [172], [202], [205], [225], [234] [78] CICIDS2017 Intrusion Detection 30,540 80 [110], [115], [117], [122], [127], [153] [159], [166], [186] [79] N-BaIoT Botnet Detection 7062606 115 [126], [152], [177], [179], [183],…”

“…Anastasakis et al [230] take the huge IoT network log as an advantage to enhance the security in IoT systems using federal learning (FL). FL is a distributed machine learning approach that aims to build and train global models based on training datasets that are distributed across different remote devices while avoiding data leakage.…”

Fostering Trustworthiness of Federated Learning Ecosystem through Realistic Scenarios

“…A challenge that should be faced during the development of a secure FL system is understanding and balancing the trade-off between the privacy-preserving level and the achieved ML performance. The authors in [59] evaluates the performance of the proposed FL system under various settings of differential privacy as a privacy preserving technique and configurations of the FL nodes, investigating the trade-off between those components and achieved model performance. Specifically, they demonstrate how the model performance is affected at different level of DP, when the number of participants increases as well as in the case of imbalanced client data.…”