Data generation and sharing for robust intrusion detection in IoT systems

Purpose and goal:

Intrusion detection systems (IDS) are a critical component for an efficient cybersecurity strategy in IoT systems. However, data driven IDS are only as good as their training and a fundamental challenge is to predict new attacks. We address this challenge by providing training data and ensuring confidential knowledge sharing between actors by leveraging and further developing advances in machine learning.

Expected results and effects:

The project goal is to further improve Sweden’s position at the forefront of cybersecurity for IoT networks with a wide range of applications. The project is targeting improvements of accuracy and robustness of IDS and will thus enable secure and robust digitalization involving IoT systems. The expected key results are novel technologies for privacy-preserving and robust knowledge sharing for IDS, specifically tailored to leverage the inherent properties of IoT systems, and datasets for robust training of data-driven IoT IDS.

Approach and implementation:

The project approach will be to work along two main research tracks in parallel, namely (a) data generation, and (b) private knowledge sharing. There are dependencies between the two tracks, for example private knowledge sharing requires data availability from the data generation track. We will utilize publicly available datasets and use testbed and simulation to generate realistic and synthetic data.

The main research will be done by a postdoc, to be hired, supported by two supervisors with expertise in IoT security and machine learning, respectively.