Elena Dubrova has found the Achilles' heel of cryptosystems
Professor Elena Dubrova is one of the KTH's highest-ranked researchers in computer hardware and architecture and is among the two per cent highest-ranked researchers in the world. In the latest issue of Elektroniktidningen, she describes how she and her students have defeated one cryptosystem after another with the help of deep learning-based side-channel analysis.
Together with your students, you have found the Achilles heel of crypto systems; why do you believe this method has been so successful?
"Deep learning techniques are good at finding correlations in raw data. They enable the attacker to bypass many existing countermeasures against side-channel analysis, such as masking and shuffling, and break protected implementations, as we recently demonstrated on the example of Saber key encapsulation mechanism, a finalist of NIST post-quantum cryptography competition ."
Your research in recent years has focused on hardware security, and you have introduced this subject in undergraduate education; what has been the response from your students?
"The feedback on my Hardware Security course (IL1333) is very positive. Since hardware security is a young and growing area, the students feel at the forefront of research, where new things are discovered as they study the subject. They even have a chance to contribute to these discoveries by doing a research project. From 2021 the IL1333 course is also offered as a Lifelong Learning course."
You are currently working on defence methods together with Ericsson. What impact will that collaboration have?
"We aim to develop new hardware security assessment methods for Field Programmable Gate Array (FPGA) technology which is widely used to implement cryptographic algorithms and AI acceleration. Our expected results are to guide how to adapt existing FPGA design practices to ensure proactive vulnerability management."