Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Internet of Things Security throughSelf-Supervised Graph Neural Networks

Safa Ben Atitallah, Maha Driss, Wadii Boulila, Anis Koubaa

TL;DR

The paper addresses IoT intrusion detection under severe data imbalance and the emergence of novel attack types. It introduces a self-supervised learning framework that combines MarkovGCN with edge-weight prediction to learn robust graph representations from IoT network data, pretraining on the network structure before fine-tuning for detection. On the EdgeIIoTSet dataset, the approach achieves high performance (e.g., 98.68% accuracy) and outperforms a supervised DNN baseline, demonstrating data-efficient and scalable intrusion detection. The work suggests significant practical impact for securing heterogeneous IoT environments and lays groundwork for extensions like federated learning and graph augmentations.

Abstract

With the rapid rise of the Internet of Things (IoT), ensuring the security of IoT devices has become essential. One of the primary challenges in this field is that new types of attacks often have significantly fewer samples than more common attacks, leading to unbalanced datasets. Existing research on detecting intrusions in these unbalanced labeled datasets primarily employs Convolutional Neural Networks (CNNs) or conventional Machine Learning (ML) models, which result in incomplete detection, especially for new attacks. To handle these challenges, we suggest a new approach to IoT intrusion detection using Self-Supervised Learning (SSL) with a Markov Graph Convolutional Network (MarkovGCN). Graph learning excels at modeling complex relationships within data, while SSL mitigates the issue of limited labeled data for emerging attacks. Our approach leverages the inherent structure of IoT networks to pre-train a GCN, which is then fine-tuned for the intrusion detection task. The integration of Markov chains in GCN uncovers network structures and enriches node and edge features with contextual information. Experimental results demonstrate that our approach significantly improves detection accuracy and robustness compared to conventional supervised learning methods. Using the EdgeIIoT-set dataset, we attained an accuracy of 98.68\%, a precision of 98.18%, a recall of 98.35%, and an F1-Score of 98.40%.

Enhancing Internet of Things Security throughSelf-Supervised Graph Neural Networks

TL;DR

The paper addresses IoT intrusion detection under severe data imbalance and the emergence of novel attack types. It introduces a self-supervised learning framework that combines MarkovGCN with edge-weight prediction to learn robust graph representations from IoT network data, pretraining on the network structure before fine-tuning for detection. On the EdgeIIoTSet dataset, the approach achieves high performance (e.g., 98.68% accuracy) and outperforms a supervised DNN baseline, demonstrating data-efficient and scalable intrusion detection. The work suggests significant practical impact for securing heterogeneous IoT environments and lays groundwork for extensions like federated learning and graph augmentations.

Abstract

With the rapid rise of the Internet of Things (IoT), ensuring the security of IoT devices has become essential. One of the primary challenges in this field is that new types of attacks often have significantly fewer samples than more common attacks, leading to unbalanced datasets. Existing research on detecting intrusions in these unbalanced labeled datasets primarily employs Convolutional Neural Networks (CNNs) or conventional Machine Learning (ML) models, which result in incomplete detection, especially for new attacks. To handle these challenges, we suggest a new approach to IoT intrusion detection using Self-Supervised Learning (SSL) with a Markov Graph Convolutional Network (MarkovGCN). Graph learning excels at modeling complex relationships within data, while SSL mitigates the issue of limited labeled data for emerging attacks. Our approach leverages the inherent structure of IoT networks to pre-train a GCN, which is then fine-tuned for the intrusion detection task. The integration of Markov chains in GCN uncovers network structures and enriches node and edge features with contextual information. Experimental results demonstrate that our approach significantly improves detection accuracy and robustness compared to conventional supervised learning methods. Using the EdgeIIoT-set dataset, we attained an accuracy of 98.68\%, a precision of 98.18%, a recall of 98.35%, and an F1-Score of 98.40%.

Paper Structure

This paper contains 17 sections, 3 equations, 4 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Proposed Approach
  4. Graph Constraction
  5. Self-Supervised Learning with Edge Weight Prediction
  6. MarkovGCN Architecture:
  7. Markov Process Aggregation:
  8. Pretext Task:
  9. Downstream Task
  10. Experiments
  11. Dataset
  12. Experimental Settings
  13. Experimental Results
  14. Comparison with Related Works
  15. Discussion
  16. ...and 2 more sections

Figures (4)

  • Figure 1: An Overview of the Suggested SSL-Based MarkovGCN for IoT Intrusion Detection
  • Figure 2: The distribution of Edge-IIoTSet dataset over the 15 classes
  • Figure 3: Performance Results: Precision, Recall, and F1-score by Attack Class
  • Figure 4: The ROC Curve for Multi-class Classification Using The Proposed Approach on The Edge-IIoTSet dataset.