Physical implementations of cryptographic schemes are vulnerable to side-channel attacks, which exploit leakages such as power consumption to recover secret information. A common countermeasure is masking, where sensitive data is split into multiple shares to limit information leakage.

To analyze the security of masked implementations, the random probing model provides a tractable and realistic framework, where each variable leaks independently with some probability. In this talk, I present recent advances on the design and analysis of masked implementations in this model.

In particular, I will present practical methods to evaluate the security of small masked circuits, as well as techniques to analyze larger systems using a modular, compositional approach. I will also introduce efficient masked gadgets satisfying these properties, and illustrate their benefits through a case study on AES.