The University of New Brunswick

Published purpose

This project will develop new methods to ensure the security of IoT devices in large corporate networks. Data provenance and machine learning will be the main project tools. Provenance establishes the trustworthiness of the data generated by devices, which can in turn be used for training machine learning models that detect malicious behaviour. A major focus of the research will be to ensure that the new techniques do not introduce additional attack vectors. The project will investigate the vulnerability of ML-based systems to attacks that specifically target ML algorithm, and the project team will apply robust learning techniques where necessary to protect against such attacks. Distributed ledgers will be leveraged to provide tamper resistance to the data provenance information.