New leakage models: How to deal with rejection
If I told you that a computer performed one of two computations—either factoring a 3-digit number or a 100-digit number—and then asked you to guess which one it was, you’d have a 50:50 chance of being correct.
What would your answer be if I gave you a hint—say, the computation took 3 hours, or the computer heated up by 5°C? Or what if the computer worked so hard that your crazy neighbor felt the WiFi waves through the wall?
We call such hints “leakage”, and very often, they are easy to come by. We can clearly measure computation time, temperature, power consumption, electromagnetic radiation, sound, etc.
This might seem like a tiny amount of information, but as in the case of the Titanic, what seems like a small leakage can quickly turn into a catastrophic event, leading to the complete collapse of a cryptosystem.
However, to prevent leakage, we first need to model it. In this talk, we will discuss various models of leakage and their equivalence. Our focus will be on a technique that allows us—in a black-box and lossless manner—to translate security against bounded leakage into security against many alternative (and more practical) models.
We will also show that every Lipschitz leakage can be efficiently simulated with a very small amount of bounded leakage. This means, in particular, that if our measuring equipment has noisy readings, it cannot leak large amounts of information.
This talk is based on the following papers:
1. The Mother of All Leakages: How to Simulate Noisy Leakages via Bounded Leakage (Almost) for Free
G. Brian, A. Faonio, M. Obremski, J. Ribeiro, M. Simkin, M. Skórski, D. Venturi
Eurocrypt 2021, IEEE Transactions on Information Theory 2022
2. Improved Reductions from Noisy to Bounded and Probing Leakages via Hockey-Stick Divergences
M. Obremski, J. Ribeiro, L. Roy, F.X. Standaert, D. Venturi
Crypto 2024
3. Simulating Noisy Leakage with Bounded Leakage: Simpler, Better, Faster
A. Mukherjee, M. Obremski, J. Ribeiro, L. Roy, F.X. Standaert, D. Venturi
Manuscript
Speaker Biography
Maciej is a Research Assistant Professor at the School of Computing, National University of Singapore. Before that, he spent five years in Divesh Aggarwal’s research group at the National University of Singapore—first as a Research Fellow, and since 2020, as a Senior Research Fellow. Prior to joining NUS, he spent three years as a Post-doc at Aarhus University (Denmark), working with Ivan Damgård and Jesper Buus Nielsen, and two years at the University of Warsaw (Poland). He earned his Ph.D. in 2013 from the University of Warsaw, under the supervision of Stefan Dziembowski. His research revolves around the theory of computer science, primarily focusing on problems inspired by modern cryptography. We live in a world where cryptographic devices are physical objects that can leak information and be tampered with. Maciej contributes to research that models and addresses the challenges of leakage- and tamper-resilient cryptography.
