Research Topic: New Directions in Hardware Security through Cognitive Obfuscation [in Research Hub B, RC 4 (Platform Trojans), and Hub D, RC 10 (Engineers and Usability)]

Duration   01/2021-12/2023
Funded by   DFG [German Research Foundation]
Researchers   Prof. Dr. Nikol Rummel, Carina Wiesen
External website   
Partners   Prof. Dr.-Ing. Christof Paar (Embedded Security, Max Planck Institute for Security and Privacy in Bochum), Steffen Becker (Embedded Security, Ruhr University Bochum)

Project description

In the Cluster of Excellence “CASA – Cyber Security in the Age of Large-Scale Adversaries”, the interdisciplinary research team aims to analyze underlying human factors in Hardware Reverse Engineering (HRE) and to develop novel countermeasures (obfuscation) impeding HRE. HRE is a multilayered process, where human analysts employ sense-making processes and combine various semi-automated technical steps to extract high-level information from a low-level circuit. Those underlying cognitive processes that determine the success of an HRE attack remain poorly understood as only little prior research exists so far. Somewhat surprisingly, existing hardware obfuscation methods are largely based on ad-hoc methods, and do not take cognitive processes of reverse engineers into account, even though they are the main target of the obfuscation methods. In this research project, we strive towards a fundamental understanding of HRE, with the goal of designing cognitively difficult countermeasures (“cognitive obfuscation") that exploit limitations of human reverse engineers.
In this research project, we will built on our initial findings from previous exploratory studies (e.g., Becker et al., 2020) and pursue two strands of research. First, we will investigate the strength of existing hardware obfuscation techniques and will develop methods that exploit the cognitive limitations of reverse engineers. Second, we will develop a game-based methodology to obtain quantitative results regarding cognitive processes from experimental studies with a larger number (n > 200) of participants enabling the use of inferential statistics to gain more robust, reliable results for cognitive obfuscation. Finally, we will bring the work from both strands together to develop and evaluate cognitive obfuscation methods.

Back to current projects